SILIC: Intelligent On/Off Control for Networked Solar Insecticidal Lamps

I. Introduction

SINCE pests and diseases are among the main factors affecting the quality of crops, effective pest management is of paramount importance for enhancing crop yields [1], [2]. According to the Chinese Ministry of Agriculture of 2017, solar insecticidal lamps (SILs) are an innovative type of pest management device that can trap adult pests and reduce the pesticide use [3]. Compared with traditional pesticide pest management methods (e.g., chemical pesticide control and biological control), SIL plays an important role in organic farms [4]–[7]. The insecticidal lamp (IL) stands out as a prominent pest management device in agriculture. The continual advancement of technology across domains, such as optics, electric energy, and 4G networks has yielded a revolution in the design and performance of the IL [8]. This transformation has witnessed the evolution from a single insect-trapping light source to multiple light sources, resulting in a significant improvement in its insecticidal effectiveness. The energy supply for IL has progressed from relying solely on batteries to a combination of battery and solar energy. Simultaneously, the advancement of internet of things (IoT) technology has facilitated agricultural modernization and accelerated its integration into smart agriculture [9]–[12]. The combination of insecticidal lamp (IL) and IoT module has improved the pest management effect of ILs and formed a new type of agricultural IoT—solar insecticidal lamp IoT (SIL-IoT) [6]. By incorporating IoT module, SIL-IoT nodes can perform the following functions: 1) Monitoring weather conditions, pest number information, device fault status, and remaining energy status of the battery; 2) Predicting the number of pests trapped; 3) Transmitting the aforementioned data to servers.

The SIL predominantly employs the IoT module to relay crucial data such as pest information, weather conditions, energy status, and other device related data to servers, thus enabling real-time monitoring of pest density distribution. This information is primarily collected by the following sensors: solar panel voltage and current sensors, a humidity sensor (DHT11), a temperature sensor (DS18B20), a sound sensor, as well as voltage and current sensors [8]. Given the high concentration of pests near rivers, the number of SIL deployed in these areas and insecticidal working time have correspondingly increased. As depicted in Fig. 1, the SIL’s energy module comprises three core components: i) An energy harvesting module; ii) An energy storage module; iii) An energy consumption module. In the daytime, the solar panel harvests sunlight and converts it into energy, which is stored in the battery until it is fully charged. As night falls, the SIL lures phototropic pests primarily through its light source. Upon contact with the metal mesh, a high voltage pulse discharge, ranging from 2150 V to 6000 V, is released to trap these pests. Its design focuses on the effective harvest, storage, and consumption of energy to ensure the efficient utilization of SIL. Consequently, the energy consumed by SIL is predominantly attributed to: i) Turning on the lamp; ii) Executing pest extermination; iii) Operating IoT module [6].

Figure  1.  Modules of a SIL-IoT node.

DownLoad: Full-Size Img PowerPoint

The efficiency of SIL in pest control is closely related to the number of pests it can capture during specific time. The more pests captured, the better the pest control effect. Fig. 2 shows the phototactic behavior of Mythimna seperata during the night [13], indicating fewer pests in the late night compared to earlier periods. When the peak activity periods of pests do not align with the preset pest control times, SIL’s efficiency and effectiveness are reduced. To address this, the mathematical model of pest phototactic rhythm, is applied to adjust SIL’s insecticidal working time more precisely to align with the peaks of pest activity. Specifically, pests that exhibit significant phototactic behavior become more active during certain periods, potentially leading to outbreaks. Operating insecticidal work during periods of low pest number reduces energy efficiency and increases the risk of harming non-target insects. More importantly, excessive energy consumption during periods of low pest activity could lead to energy shortages when pests number peak, potentially missing the optimal insecticidal time. Therefore, there is an urgent need for an adaptive on/off scheme to enhance the pest control effectiveness and energy efficiency of SILs. This paper proposes a new approach combining the residual energy state of SIL-IoT with a mathematical model of pest phototactic rhythm, presenting an energy management scheme for SIL-IoT, as shown in Fig. 3(b). Initially, the number of active pests at night is calculated using the mathematical model based on pest phototaxis [8]. Subsequently, genetic algorithm (GA) and greedy algorithm are combined to optimize the model. Finally, the optimal insecticidal time for SIL is determined, achieving the best pest control effect with minimal energy consumption. Thus, intelligent energy management is a crucial component of SIL-IoT, optimizing insecticidal time and enhancing the energy use efficiency.

Figure  2.  Percentage of Mythimna seperata at night.

DownLoad: Full-Size Img PowerPoint

Figure  3.  Workflow of SIL on/off control method ((a) Workflow of traditional remote switching method (TRSM) for SIL; (b) Workflow of pest phototactic rhythm mathematical model for SIL).

DownLoad: Full-Size Img PowerPoint

SIL performs its insecticidal work during the non-rainy weather and charges during the clear weather conditions. Due to the high energy consumption of the SIL’s insect-trapping light source, the duration of the insecticidal work of this solar energy dependent device is limited by it energy. Currently, most SILs typically employ a “lights on at sunset, lights off at sunrise” method, defined as the traditional remote switching method (TRSM) [14]. As shown in Fig. 3(a), the insecticidal time for SIL is set by companies. However, the long-term fixed insecticidal time for pest control can lead to low energy utilization efficiency and may also result in poor pest control effectiveness. The inflexibility of TRSM makes it challenging for SIL’s insecticidal time to coincide with the actual activity periods of pests.

This work provides an intelligent on/off scheme for the SIL, aiming to trap more pests with less energy. This approach improves insecticidal effectiveness and energy efficient use of SIL-IoT. Specifically, the main contributions of this paper are summarized in the following:

1) This work is the first to integrate the pest phototactic rhythm mathematical model into the energy management of SIL [8]. A novel SIL intelligent energy management scheme (SIL-IEMS) is established based on the SIL’s energy consumption and pest phototactic rhythm model.

2) The on/off method in SIL-IEMS is proposed by combining genetic and greedy algorithms. The SIL-IEMS is framed as an optimization problem aimed at maximizing the insecticidal rate while ensuring insecticidal working sustainability under constraints.

3) The simulation result demonstrates that, in comparison to the TRSM, the SIL-IEMS achieves a significant 17.6% enhancement in insecticidal effectiveness. Furthermore, when initiated at a 15% energy level, the insecticidal rate under the SIL-IEMS surpasses that of the solar insecticidal lamp genetic algorithm (SILGA) by 6%.

The rest of this paper is organized as follows. Related work is presented in Section II. Section III lists the model assumptions and describes the system model. We formulate an optimization problem for SIL efficient insecticidal performance and propose a scheme in Section IV. The simulation results and performance analysis are shown in Section V. Finally, Section VI concludes this paper and discusses future work.

II. Related Work

A On/Off Scheduling Method for the Outdoor Devices

Currently, a multitude of investigations have been undertaken concerning the energy management of intelligent devices or sensor nodes across diverse monitoring frameworks. The fundamental goal of energy management resides in optimizing the utility of available energy, thereby increasing energy utilization efficiency. Given that energy management can be conceptualized as a scheduling conundrum, it has been systematically explored across various domains encompassing smart grids [15]–[18], intelligent industries [19], [20], and smart agricultures [21], [22], as shown in Table I.

Table  I.  Comparison of Different Intelligent On/Off Schemes

Paper	Method	Objective	Energy source	Research field
[15]	Evolutionary algorithm	Transform load energy management into a minimization problem	Power grid	Smart grid
[16]	Grey wolf optimizer (GWO)	Adjust on/off timing of household devices	Photovoltaic and power grid	Smart grid
[17]	Demand response management, peak shaving, and valley filling techniques	Make better use of renewable energy to achieve fairness in charging and discharging	Photovoltaic and power grid	Smart grid
[18]	Heuristic dynamic programming	Solve residential energy scheduling problems	Photovoltaic and power grid	Smart grid
[19]	Predictive control scheme based on deep reinforcement learning	Light switch control in wastewater treatment plants	N/A	Intelligent industry
[20]	Mixed integer optimization approach	Obtain optimal steady-state power allocation	Power grid	Intelligent industry
[21]	Predictive control, economic cost constraints	Improve traditional irrigation methods in strawberry greenhouses predictive control, economic cost	Power grid	Smart agriculture
[22]	Adaptive sleep scheduling algorithm	Coordinate node transitions between sleep and active states to minimize energy consumption	Battery	Smart agriculture
Ours	Intelligent on/off scheme	Optimize insecticidal working time of solar insecticidal lamp	Solar panel and battery	Smart agriculture

 | Show Table

DownLoad: CSV

In [15], the author proposed an evolutionary algorithm focusing on transforming the challenge of load energy management into a minimization problem. The method reduces energy consumption by controlling load on/off and makes full use of renewable energy sources by optimizing the smart grid configuration. Addressing the cost reduction concern from a time-based perspective, [16] introduced the grey wolf optimizer (GWO) to adjust the on/off timing of household devices. Due to the intermittent nature of renewable energy source, [17] implemented demand response management to better utilize these energy sources. The heuristic dynamic programming method was proposed in [18] for residential energy management based on varying energy demands. Additionally, they employed peak shaving and valley filling techniques in electric vehicles to reduce energy consumption. Reference [19] proposed a predictive control scheme based on deep reinforcement learning for light switch control in wastewater treatment plants. To obtain the optimal steady-state power allocation, [20] proposed a mixed integer optimization approach to find the optimal result for generation and on/off loads. In the case of energy and resource constraints, [21] improves the traditional irrigation method. Predictive control of periodic on/off irrigation in a strawberry greenhouse by adding economic cost constraints. Intelligent control of daytime running lights is achieved through a reward function that considers the amount of energy consumption and electricity price information to reduce energy consumption. An adaptive sleep scheduling algorithm is proposed in [22] to coordinate the transition of nodes between sleep and active states to minimize energy consumption. Unlike the aforementioned studies, SILs do not connect to a 220 V power grid and rely solely on solar panels to charge batteries for nighttime pest control. Therefore, improving energy utilization efficiency is crucial. The intelligent on/off scheme for SILs focuses on optimizing insecticidal working time to capture more pests with less energy, thereby enhancing trap pest efficiency and energy utilization. This method not only enables insecticidal work efficiency with limited energy but also maximizes the use of energy.

B Energy Management Method in SIL Commercial Products

SIL, as a green and pollution-free pest control method, has been widely adopted. Presently, major companies adopt TRSM for their ILs, as highlighted in Table II. Descriptions for each column are provided below.

Table  II.  Comparison of Different Companies’ Insecticidal Lamps

Company	Type of insecticidal lamp	Pest counting	Product life cycle	Intelligent on/off control method
Henan Yunfei Science and Technology Co., Ltd. [23]	YF-W-L40 smart IoT insecticidal lamp [24]	The count is achieved by counting the number of discharges from the high voltage metal mesh. This counting method is no longer reliable when multiple pests hit the metal mesh at the same time or when pests stick to the metal mesh causing continuous discharges.	1) Continuous cloudy working cycle: without charging, it can work continuously for three nights, 8 hours per night, until the energy is exhausted. 2) Design life: 5 to 15 years. (Except lamp and battery). 3) Lamp life: about 500 hours. 4) Battery life: 1000−1200 discharges of lead-acid battery (without overloading).	Traditional remote switching method (TRSM)
Zhejiang Top Cloud-agricultural Technology Co., Ltd. [25]	TPSC7-1-DC networked insecticidal lamp [26]
Zhejiang Longhao Agricultural Science & Technology Co., Ltd. [27]	IoT solar insecticidal lamp [28]
Chengdu Beang Technology Co., Ltd. [29]	BA-T/YCI remote intelligent cloud insecticidal lamp [30]
Henan Sailan Instrument and Equipment Manufacturing Co., Ltd. [31]	Smart IoT solar insecticidal lamp [32]
Changzhou Jinhe New Energy Technology Co., Ltd. [33]	Networked insecticidal lamp [34]

 | Show Table

DownLoad: CSV

These companies [23], [25], [27], [29], [31], [33] share common attributes concerning insecticidal number, product lifespan, and the incorporation of TRSM functionality within SILs. The prevailing approach in SIL products involves quantifying pest density primarily through the enumeration of voltage pulses emitted by the high-voltage mesh upon pest extermination. Currently, the companies produce SILs that are primarily controlled to turn on/off at the frequency of once per night.

Nonetheless, investigations in [23], [35] indicate that the lamp component can endure up to 500 hours of insecticidal work time without malfunction. The lead-acid battery, barring any leakage, can sustain between 1000 to 1200 discharge-recharge cycles. Although solar panels can power SILs, due to the cyclical nature of solar energy, they can only charge the lamps on sunny days. Without recharging, the battery can sustain insecticidal working for three consecutive nights, with each night providing 8 hours of insecticidal working, until the energy is exhausted. Particularly in remote areas, SILs often lack necessary maintenance and thus must rely on renewable energy sources such as solar panels [36]. However, most commercial SILs use the TRSM remote management system to control insecticidal times, leading to low energy utilization and suboptimal insecticidal effectiveness. Especially, when pest numbers are low, continuously turning on the SIL for insecticidal working not only fails to effectively attract target pests but may also inadvertently capture non-target insects. Consequently, intelligent energy management of SILs is required.

Considering the sparse deployment characteristics of SIL-IoT, existing sleep scheduling management methods are not suitable for SIL energy management [37]. In this paper, the proposed SIL-IEMS becomes particularly crucial for enhancing the pest control effectiveness and energy efficiency of SILs. By optimizing the insecticidal time, SIL-IEMS ensures the sustainable insecticidal working of SIL in energy-limited remote areas and improves energy utilization.

III. System Model

Energy is fundamental for monitoring pest numbers, data transmission, and sensor device operations in the SIL-IoT. The core models in SIL-IoT include the mathematical model of pest phototactic rhythms, the energy consumption model, and the energy residual model. These models together constitute the SIL energy management models, aiming at optimizing the insecticidal working time of SIL, thereby enhancing pest control efficiency and energy utilization.

Table III lists the various notations used in this research. The proposed SIL-IoT model is detailed below.

Table  III.  Notations

Notation	Description
i	The time slot i where SIL turned on
j	The time solt j where SIL turned off
$t_{i1}$	The starting time point of the time slot $t_{i}$
$t_{j2}$	The ending time point of the time slot $t_{j}$
$t_m$	Time slots; per 60 minutes is a pest recording duration $0 <t\le11$
$f(t_m)$	The random distribution of pest occurrence at each $t_m$ time slot during the nighttime
$F(t_m)$	The fitness function
N	The total number of individuals in the population
n	The individual n in the population
$P(n)$	The probability of selecting individual n
k	The position or gene locus on a chromosome
$A_k$	The paternal individual 1
$B_k$	The paternal individual 2
$C_k$	The offspring individual
$\mu$	The cross-ratio parameters
$\delta$	The step size of the variation
$\gamma$	The maximum variation step size
$pick$	Represents a random number in the interval [0, 1]
$t_m'$	The current value to be mutated
$t_m''$	The new value after mutation
$E_{0}$	The initial remaining remaining energy of the SIL storage battery in the daytime
$E_{c}^{t_m}$	The energy consumed at night time
$E_{lc}^{t_m}$	The energy consumption of lamp
$E_{mc}^{t_m}$	The energy consumption of metal mesh
$E_{pc}^{t_m}$	The energy consumption of insecticidal
$E_{tc}^{t_m}$	The energy consumption of IoT communication transmission
$E_{a}$	The total capacity of the 48 V lead-acid battery
$E_{r}^{t_m}$	The residual energy of the SIL
$E_{c}^{t_m-1}$	The energy consumption of the SIL at the $t_m-1$ time slot
$SOC_{0}$	The energy state in the SIL’s 48 V lead-acid battery from the moment the lamp is turned on at night time
$SOC_{t_m}$	The energy state of the SIL at the $t_m$ time slot
$SOC_{t_m-1}$	The energy state of the SIL at the $t_m-1$ time slot
$\alpha_{t_m}$	The working state of SIL with lamps on to trap pests during the $t_m$ time slot

 | Show Table

DownLoad: CSV

A Assumptions and Definitions

Solar energy is harvested by the solar panel during the daytime, and subsequently stored in the battery to meet the insecticidal requirements of SILs during the nighttime.

Assumption 1: All the energy stored in the SIL battery is used for insecticidal work and no leakage occurs.

Assumption 2: SIL system will only perform once on/off scheme throughout the entire night.

B Pests Distribution Mathematical Model

The phototactic behavior that nocturnal pests exhibit towards lure light sources, also known as phototactic rhythm, has evolved over a long period in the ecological environment [1]. The combination of SIL and pest phototactic behavior will be increasingly applied in pest monitoring and green control. The specific outbreak period of pests may be influenced by a variety of factors, including, month, geographical location, crop type, temperature, humidity, precipitation, and wind direction. Generally, SILs are mainly used from late spring to autumn, especially during the months when pests are active. For East China, the high activity period of pests usually lasts from May to October.

Currently, the distribution types of the phototactic rhythm of pests mainly include single [38], double [38], and multiple peak types [39], as shown in Fig. 4. Among them, due to the different times when pests appear at dusk, midnight, and dawn, the single-peaked type mainly includes the dusk single-peaked type [40], midnight single-peaked type [13], and early dawn single-peaked type [39].

Figure  4.  The distribution types of the phototactic rhythm of pests.

DownLoad: Full-Size Img PowerPoint

In this paper, we leverage the mathematical model of pest phototactic rhythm that was previously investigated [8] as a foundation for our research. The model delineates a 24 hour cycle into periods of energy consumption time (e.g., from 18:00 to 5:00 the next day) and energy supply time (e.g., from 5:00 to 18:00). SIL intelligent energy management focuses on the rational allocation of energy consumption schedule. Consequently, the 11-hour insecticidal time from 18:00 to 5:00 (the next day) is considered to be optimal for effective pest management. The per hour time slot is a pest recording duration, denoted by $t_m$, where $0<t_m\le11$. Thus, the 11-hour insecticidal period is divided into 11 time slots (e.g., $t_m = 1$ represents the time slot from 18:00 to 19:00, $t_m = 11$ represents the time slot from 4:00 to 5:00).

The pests’ phototactic rhythm exhibits diverse activity peaks under different weather conditions. Considering that SIL insecticidal works only on rainless nights and automatically shuts down during rainfall. Based on the previously established mathematical model [8], this study focuses on analyzing the phototactic rhythm of pests on rainless nights. By optimizing the insecticidal time of SIL, the goal of killing more pests with less energy was achieved, thereby enhancing pest control efficiency.

As Mythimna seperata is the major pest of crops, such as wheat, rice, maize, cotton, legumes, and vegetables, its outbreak can cause significant damage to the yields of these crops. Therefore, it is important to maximize the number of traps for Mythimna seperata. The phototropic rhythms of Mythimna seperata is the early dawn single-peaked type. The primary objective of this work is to establish the mathematical model based on the phototactic rhythm of Mythimna seperata to achieve an intelligent on/off scheme for SIL. The equation of the phototactic rhythm curve of Mythimna seperata derived from [8] is shown in (1)

The insecticidal rate of SIL: The percentage of pest number eliminated $rate_-p (t_{i 1}, t_{j 2})$ from the $t_{i1}$ to $t_{j2}$ by SIL to the total number of pests $N_{p_{\text {all }}}$on that night, is shown in (2)

C Energy Consumption Model

The energy consumption of SIL consists of three main components: 1) The trap pest light source module; 2) The insecticidal module (including metal mesh and insecticidal pulse); 3) The IoT module. Specifically, the SIL’s IoT module mainly consists of a) an antenna to provide a signal, b) a CC2538 communication module for data transmission, c) 5 V to 3.3 V energy supply for wireless communication, d) 12 V to 5 V energy supply for chip and sensor, e) Raspberry Pi, and f) various sensing, which are shown in Table IV. Among them, the node signals are transmitted through the antenna to the CC2538 communication module, and the resulting data are stored in the Raspberry Pi. The same hardware platform is required for the subsequent experimental setup.

Table  IV.  The Parameters of Experimental Equipment

No.	Device	Parameter
1	Voltage and current sensors	INA219
2	Wireless communication modules	CC2538 + CC2592 ZigBee Module
3	Raspberry Pi 4 B	1 GB RAM + 64 GB storage
4	Batteries	12 V, 20 AH
5	Microprocessors	STC89C52RC
6	Triode	SO9012
7	Relays	DC-5 V
8	High voltage package	6 kV

 | Show Table

DownLoad: CSV

Hence the energy consumption of SIL $E_{c}^{t_m}$ in $t_m$ time slot is mainly 1) the energy consumption of trap pest light source $E_{lc}^{t_m}$, 2) the energy consumption of metal mesh $E_{mc}^{t_m}$, 3) the energy consumption of insecticidal pulse $E_{pc}^{t_m}$, and 4) the energy consumption of IoT module $E_{tc}^{t_m}$. Thus the energy consumption of SIL $E_{c}^{t_m}$ can be modeled as

In this paper, the lamps of SIL are mainly 15 W insect-trapping light source. When the high-voltage metal mesh is not discharged, the power of the SIL is 1.8 W. Considering the low energy consumption of the IoT module compared to the SIL lamps, the energy consumption per unit time of the IoT module can be considered as a constant.

Effective insecticidal energy use rate of SIL ${{rate}}_-{e_(t_{in},t_{jz}))}$: the percentage of energy consumed by SIL for killing pests $E_{pc}^{\left.(t_{i 1}, t j_2\right)}$ to the total energy consumed $E_{c}^{\left.(t_{i 1},t j_2\right)}$, can be modeled as

Ineffective insecticidal energy use rate of SIL $rate_-le_{(t_i,t_{iz})}$ can be modeled as

D Residual Energy Model

The SIL studied in this paper depends primarily on solar panels for energy during the daytime and then stored in 48 V lead-acid battery. At night time, it mainly relies on the battery for energy supply. Therefore, the energy remaining in the SIL at the time slot $t_m$ is mainly determined by the initial remaining energy of the SIL storage battery in the daytime, denoted as $E_{0}$, and the energy consumed at nighttime denoted as $E_{c}^{t_m}$. Consequently, the residual energy of the SIL is modeled as follows:

The SIL’s initial remaining state of charge $S OC_{0}$ is defined as the energy state in the SIL’s 48 V lead-acid battery from the moment the lamp is turned on during the nighttime. It can be expressed as the ratio of the energy capacity of the SIL at the moment the lamp is turned on to the total capacity of the 48 V lead-acid battery $E_{a}$. The SIL initial remaining state of charge can be calculated as

The energy state of charge of SIL at $t_m$ time slot can be calculated as

where $E_{c}^{t_m-1}$ is the energy consumption of the SIL at the $t_m-1$ time slot. $S OC_{t_m-1}$ is the energy state of the SIL at the $t_m-1$ time slot. The SIL energy scheme for the $t_m$ time slot is influenced by the energy state of charge of SIL at the $t_m-1$ time slot.

Specifically, the SIL system is designed to avoid deep discharge, which could significantly enhance the battery lifespan. When the remaining battery capacity of SIL in $t_m$ time slot is less than 10%, SIL’s battery will be deeply discharged. To improve the lifetime of SIL battery, the phenomenon of SOC_tm < 10% should be avoided. Therefore, for the intelligent energy management of SIL-IoT in this paper, the SIL will stop the insecticidal work when the remaining battery capacity falls below 10% (SOC_tm < 10%). Hence, the intelligent energy management scheme of SIL-IoT is mainly influenced by the initial remaining energy, energy consumption, and energy surplus.

IV. Proposed Intelligent Energy Management Scheme

A Parameter Transformation

Since SILs are sparsely deployed and have fewer redundant areas between them, their insecticidal ranges are relatively independent. To improve the effectiveness of pest management in the insecticidal area, the insecticidal working time of a single SIL can be adjusted before considering the optimization problem of multiple SILs. This paper aims to optimize the insecticidal working time of SIL to minimize total energy consumption within a specified period while ensuring optimal insecticidal effectiveness. Specifically, the on/off time of SIL ($t_{i1}$), ($t_{j2}$) is represented as two gene segments. Among them, i represents the time slot where SIL is turned on, j represents the time slot where SIL is turned off, 1 represents the starting time point of the time slot, and 2 represents the ending time point of the time slot. Since ($t_{i1}$), ($t_{j2}$) can form a pair of on/off time, these time form a single chromosome ($t_{i1}, t_{j2}$) as shown in Fig. 5. The on/off time of SIL is in the range of $t_{i1},t_j2 \in(0,11]$. Specifically, the time interval from 18:00 to 19:00 represents the first time slot, denoted as $t_{i1}$, where the start time of the first time slot 18:00 is represented as $t_{11}$ and the end time of the first time slot 19:00 is represented as $t_{12}$. The time interval from 19:00 to 20:00 represents the second time slot, denoted as $t_{j2}$, where the start time of the first time slot 19:00 is represented as $t_{21}$ and the end time of the second time slot 20:00 is represented as $t_{22}$. By employing selection, crossover, and mutation operations within a GA, this study seeks to identify the optimal combination of on/off time, thereby achieving the best balance between energy consumption and effectiveness.

Figure  5.  An illustration of parameter transformation. SIL’s on/off time is divided into two gene segments forming part of a chromosome.

DownLoad: Full-Size Img PowerPoint

B Problem Statement

Generally speaking, for energy efficiency, the residual energy in SIL should be reasonably utilized. In this paper, we mainly consider the initial remaining energy and energy consumption to optimize the on/off time of SIL to effectively trap pests within less than 50% of the energy.

Therefore, the objective function of SIL energy management is to maximize the number of pests trapped, which can be expressed as

where $F(t_m)$ denotes the fitness function and the optimisation objective. $f(t_m)$ denotes the pest distribution in $t_m$ time slot. $\alpha_{t_m}$ represents the insecticidal working state of SIL during the $t_m$ time slot. Here, 0 represents that the SIL is in the insecticidal working state of turning off the lamp, and 1 represents that the SIL is in the insecticidal working state of turning on the lamp and trapping pests.

C Energy Management Scheme Constraints

The intelligent energy management of SIL needs to satisfy at least the following constraints.

1) Load Energy Balance: To maintain the energy balance of the SIL, the energy consumption of the SIL during the nighttime should be less than the energy capacity of the lead-acid battery, which serves as the initial remaining energy source for the SIL to operate at night.

Lead-acid battery over-discharge tends to reduce the service life of the lead-acid battery and increase the cost of use. To reduce the usage cost of SIL, the energy capacity of the lead-acid battery should be maintained greater than 10%

Accordingly, the following residual energy constraints are defined and applied to the intelligent energy management optimization model of SIL-IoT:

2) Insecticidal Working Time Constraints: Since SIL is to turn on the lamp at night for insecticidal work, we only consider the insecticidal working time from 18:00 to 5:00 (the next day). The period from the $t_{i1}$ to $t_{j2}$ during which pests are killed is called the SIL insecticidal time, T. SIL continues to turn on the time to kill pests from the $t_{i1}$ to $t_{j2}$ is known as the SIL insecticidal time, T. For more accurate statistics of pest distribution, the insecticidal working time of 18:00−5:00 (the next day) is divided into 11 time slots. Therefore,

D Proposed Scheme

GA is the heuristic optimization algorithm that mimics natural selection and genetic mechanisms found in nature [41]. As depicted in Fig. 6, GA mimics the process of biological evolution by performing random searches, selection, crossover, mutation, and generation-by-generation evolution of candidate solutions to find the optimal solution to a problem quickly. Since the GA is an effective method for solving complex modeling optimization problems [42], it can produce globally optimal solutions.

Figure  6.  The flowchart of algorithm ((a) The flowchart of genetic algorithm (GA); (b) The flowchart of genetic algorithm (GA) + greedy).

DownLoad: Full-Size Img PowerPoint

In this paper, we combined the enhanced genetic algorithm with a greedy algorithm to determine the optimal on/off scheme for insecticidal working, which is a key component of our SIL-IEMS. The SIL-IEMS aims to maximize the number of pests trapped within a limited energy range while satisfying the energy balance and time demand constraints. The SIL-IEMS flowchart is illustrated in Fig. 7. According to the flowchart, the SIL-IEMS consists of three main parts: 1) Check the initial remaining energy; 2) Update the energy consumption; 3) SIL intelligent on/off scheme.

Figure  7.  Flowchart of the SIL-IEMS.

DownLoad: Full-Size Img PowerPoint

SIL intelligent energy management scheme: Due to the relatively deep search capability of GA, albeit with potentially slower convergence rates, and the rapid convergence of greedy algorithms during the search process, which can complement each other, combining the two forms a promising approach. By adopting these two methods, we aim to harness the comprehensive global search capability of GA while accelerating the convergence speed of the optimization process. To enhance the efficacy of SIL in pest control, this paper is based on the previously studied mathematical model of pest distribution [8]. Detailed statistical analysis of pest number dynamics is conducted for each hourly interval between 18:00 and 5:00 (the next day), elucidating the variations in their phototactic rhythm. Peak periods of pest abundance are identified through this analytical process. The optimal pest control time is determined by aligning the on/off insecticidal time of SIL with these peak periods. This methodology enables effective pest control during peak activity periods while minimizing energy consumption. Implementation of this strategy significantly enhances both pest management efficiency and energy utilization efficacy. In more detail, the SIL is turned on at the $t_{i1}$ time slot and turned off at the $t_{j2}$ time slot. Here, $0<t_{i1}\le11$, $0<t_{j2}\le11$, and $t_{i1}<t_{j2}$. Then, we calculate the energy consumption of the SIL for all time slots between $t_{i1}$ and $t_{j2}$. Check if the SIL’s remaining energy is sufficient to turn on the SIL for insecticidal work during the nighttime. If the remaining energy of the SIL SOC₀ > 10%, then the SIL could be turned on for insecticidal work.

1) Encoding: Encoding chromosomes is the first step in GA and is pivotal in enhancing solution efficiency. To achieve this, we begin by encoding the insecticidal working times of SIL (represented as $t_{i1}$ for the time of turn on the SIL and $t_{j2}$ for the time of turn off the SIL) into genetic fragments within the GA framework. Each set of on/off SIL time slots, denoted as ($t_{i1}$, $t_{j2}$), serves as a chromosome within the GA.

2) Initialization: Subsequently, employing encoding, a set of feasible solutions is randomly generated to establish an initial population with varying insecticidal working time.

3) Selection: This paper investigates the on and off times of SIL and aims to prioritize the reproduction of individuals with stronger adaptability in the next generation. The roulette wheel selection method in genetic algorithms offers advantages such as simulating natural selection, maintaining population diversity, and being easy to implement, making it an effective tool for solving optimization problems. Therefore, the selection process in this paper primarily employs the roulette wheel method for the natural selection of the next generation. This method mainly determines the probability of an individual being selected based on its fitness, thereby promoting the reproduction of individuals with higher adaptability in the population, which can be expressed as

where $P(n)$ represents the probability of selecting individual n; $F(n)$ represents the fitness value of individual n; N represents the total number of individuals in the population; n represents a position or gene locus on a chromosome.

4) Crossover: The crossover method used in this paper is arithmetic crossover, an innovative crossover technique that differs from traditional binary crossover, which produces new individuals by directly swapping gene positions. Arithmetic crossover is performed by executing a weighted average on the corresponding gene positions of two parent chromosomes, thus forming a new offspring individual. This method, by linearly combining the genes of the parents, effectively inherits the properties of the parents and introduces diversity into the population, enhancing the algorithm’s search capability and the global exploratory nature of the solutions, which can be expressed as

where $A_k$ represents the paternal individual 1; $B_k$ represents the paternal individual 2; $C_k$ represents the offspring individual; µ represents the cross-ratio parameters, the usual range of values is [0, 1]; k represents a position or gene locus on a chromosome.

5) Mutation: The improved GA in this study employs a non-uniform mutation method, which is a special type of mutation operation. As opposed to the GA’s random mutation, this method allows for the gradual reduction of mutation step size (i.e., the degree of change) as the number of iterations increases. By conducting an extensive exploration of the solution space in the early stages of the search process and fine-tuning the solutions in the later stages, this mutation strategy significantly enhances the convergence speed and quality of the solutions, which can be expressed as

where δ represents the step size of the variation; γ represents the maximum variation step size; pick represents a random number in the interval [0, 1]; n represents the individual n in the population; N represents the total number of individuals in the population; $t_m'$ represents the current value to be mutated; $t_m''$ represents the new value after mutation.

6) Greedy: The greedy algorithm is an efficient algorithm design strategy that can find near-optimal solutions within a reasonable time frame and is known for its fast computational speed. Therefore, combining the greedy algorithm with the GA to form an improved GA can fully leverage the GA’s powerful global search capabilities and utilize the greedy algorithm for detailed local optimization. This method searches for the maximum value of the fitness function by continuously comparing the insecticidal rate and energy utilization rate. Therefore, it is particularly suitable for optimizing the insecticidal working time of SIL and improving the insecticidal effect with the least amount of energy.

V. Simulation

This section presents extensive simulation experiments to evaluate the performance of the proposed SIL-IEMS. We use effective energy utilization, insecticidal working time, and energy residual rate, which are compared with those of the TRSM scheme. The comparative analysis is conducted to comprehensively assess the efficacy of the SIL-IEMS and its potential to enhance pest management. The simulation results obtained in this study are presented and analyzed in the subsequent sections.

A Simulation Conditions

In the simulated experiments, our SIL used 365−400 nm wavelength of ultraviolet light, which has a significant insecticidal effect on rice, corn, wheat and fruit tree crops [1]. The analysis based on our previous studies on modeling of pest phototactic rhythms [8] culminated in the design of an environment that facilitates extensive evaluation of the proposed SIL-IEMS. This study, MATLAB R2019a was used to perform simulations on a laptop computer with 64-bit Windows 10 operating system, 16.0 GB RAM, and 2.6-GHz-Core i7-10750 CPU. To fairly compare our research results with traditional algorithms, all the simulation experiments in this paper are conducted in a uniform environment and parameter settings.

B Simulation Results

In this section, the SIL-IEMS is performed and compared with the TRSM scheme for the same SIL network configuration.

1) The TRSM Scheme: According to the pest’s phototactic rhythm [1], [8], most pests are active in the evening, midnight, and morning, so most product SIL companies [23], [25], [27], [29], [31], [33] generally set SILs to turn on pest management at 18:00 and end at 5:00 (the next day).

2) SIL-IEMS: SIL adjusts its on/off insecticidal working time based on pest distribution, initial remaining energy, consumed energy, and residual energy. In addition, this paper compares the simulation of the GA and the improved GA. The application of the standard GA in SIL is referred to as the SILGA, and the application of the improved GA is referred to as the SIL-IEMS.

Firstly, the simulation experiments are conducted to evaluate the performance of the SIL-IEMS concerning the SIL’s initial remaining energy. The initial remaining energy is divided into ten levels, ranging from 15% to 100%. As shown in Fig. 8, the SIL-IEMS exhibits rapid convergence after 20 iterations for all initial remaining energy levels, demonstrating the feasibility of the algorithm. We conducted six independent simulation experiments. The box plot shown in Fig. 9 demonstrates that the error decreases as the initial remaining energy level increases. The results indicate that the SIL-IEMS achieves high accuracy, regardless of the number of simulation experiments, thereby validating the realism of the simulation results. To reduce data error, the following simulation experiments present average values.

Figure  8.  SIL-IEMS iterations.

DownLoad: Full-Size Img PowerPoint

Figure  9.  Insecticidal rate box plots of SIL-IoT for varying battery capacity.

DownLoad: Full-Size Img PowerPoint

Algorithm 1 SIL Intelligent Energy Management Scheme (SIL-IEMS) Based on Improved GA

　　input: $t_{i1}, t_{j2},f(t_m),SOC_{0},i,N$

　　output: $T,Nq, SOC_{t_m},Best f(t_m)$

1 Create initial population;

2 for $i =1: N$ do

3 　for $\vee t_{i1}, t_{j2} \in\left(0,11\right]$ do

4 　　$T \gets t_{j2} - t_{i1}$;

5 　　if $T>0$ then

6 　　 　if $SOC_{0}>$ 10% then

7 　 　　　Calculate the number of SIL pest distribution $f(t_m)$ from $t_{i1}$ to $t_{j2}$, according to (1);

8 　 　　　$\alpha_{t_m}$ $\leftarrow$ $t_{i1}, t_{j2}$;

9 　 　　　if $\alpha_{t_m}$ = 1 then

10 　　　　　Update $\sum_{{t_m} \in T}\left[\alpha_t f(t_m)\right]$

11 　　　　end

12 　　　end

13 　　　Calculate the consumption of SIL $E_{c}^{t_m}$ according to (3)−(17);

14 　　　while $E_{0}>E_{c}^{t_m}$ do

15 　　　　$SOC_{t_m}$ $\leftarrow$ $SOC_{0} - SOC_{t_m-1}$;

16 　　　　if $SOC_{t_m}>$ 10% then

17 　　　　　$F(t_m) \gets \sum_{{t_m} \in T}\left[\alpha_t f(t_m)\right]$;

18 　　　　　$Best F(t_m) \gets F(t_m)$;

19 　　　　　Selection, crossover and mutation for $t_{i1}, t_{j2}$ according to (15)−(18);

20 　　　　　$t_{i1}', t_{j2}' \gets t_{i1}, t_{j2}$;

21 　　　　　if $SOC_{t_m}>$ 10% then

22 　　　　　　$F(t_m) '\gets t_{i1}', t_{j2}'$;

23 　　　　　　if $F(t_m) '>F(t_m)$ then

24 　　　　　　　$Best F(t_m) \gets F(t_m) '$;

25 　　　　　　　else;

26 　　　　　　　$Best F(t_m) \gets F(t_m)$;

27 　　　　　　　Calculate the number of pest is trapped by SIL;

28 　　　　　　　Calculate the proportion of effective insecticidal energy;

29 　　　　　　　$F(t_m) '' \gets t_{i1}'', t_{j2}''$;

30 　　　　　　　if $F(t_m) ''>F(t_m) '$ then

31 　　　　　　　　$Best F(t_m) \gets F(t_m)''$;

32 　　　　　　　end

33 　　　　　　end

34 　　　　　end

35 　　　　end

36 　　　end

37 　　end

38 　end

39 end

40 Update $E_{c}^{t_m}, t_{i1}, t_{j2}$;

41 Return $T,Nq, SOC_{t_m},Best f(t_m)$

To ensure comparability between the insecticidal effect and residual energy, the number of pests, the effective insecticidal energy, and the residual energy are normalized. The accuracy of the algorithm is verified by comparing the experimental results for different values of initial remaining energy. This paper uses four main metrics: 1) Insecticidal rate; 2) Effective insecticidal energy percentage; 3) Residual energy percentage; 4) The insecticidal period of SIL pest management. The insecticidal rate is the number of pests trapped by the SIL as a percentage of the total number. Effective insecticidal energy percentage is the proportion of energy used by the SIL for pest control out of the total energy consumed. The residual energy percentage is the percentage of energy remaining in the SIL to the total initial remaining energy.

As depicted in Fig. 10, the simulation results present the performance of three distinct methods applied to SIL at various initial remaining energy levels. It is evident that as the initial remaining energy increases, the insecticidal rate, remaining energy, and insecticidal time of SIL exhibit a progressive rise. Considering the heightened activity of targeted pests during the first half of the night, prolonged insecticidal time leads to a gradual decline in effective insecticidal energy. In scenarios where the energy of SIL proves insufficient, the SIL-IEMS emerges as distinctly superior to both the SILGA and TRSM methods. Specifically, within the initial remaining energy range of 15%−40%, both the SIL-IEMS and SILGA methods outperform the non-optimized TRSM method. These methodologies demonstrate the ability to exterminate more pests within a shorter timeframe, showcasing elevated insecticidal rates and enhanced energy utilization efficiency. Particularly noteworthy is the SIL-IEMS’s 17.6% increase in insecticidal efficiency compared to TRSM and a 6% improvement over SILGA when the SIL begins with an initial remaining energy level of 15%.

Figure  10.  Comparisons of the average simulation results for three methods: traditional remote switching method (TRSM); SIL genetic algorithm (SILGA); and SIL intelligent energy management scheme (SIL-IEMS).

DownLoad: Full-Size Img PowerPoint

Additionally, when there is a sufficient energy supply, SIL adopts SIL-IEMS can target pest capture during periods when pests are most attracted to the lure, thereby reducing the total insecticidal time required. It is important to note that the pests considered in this study exhibit phototactic behavior throughout the night. When the initial remaining energy levels range from 50% to 100%, even if all three methods provide equivalent effective pest control energy, SIL-IEMS significantly outperforms the other two methods in terms of pest capture efficiency.

In conclusion, the proposed SILGA and SIL-IEMS both outperform the TRSM method for SIL. Among the three methods, the SIL-IEMS stands out as the optimal choice. Notably, this methodology not only enhances the energy utilization efficiency of SIL, but also ensures effective pest control, thus exemplifying its significance and potential for application in pest management practices.

The average simulation results of the effective insecticidal energy consumption percentage and the total energy consumption as a percentage of the initial remaining energy are shown in Fig. 11. When the initial remaining energy is less than 50%, the SIL operates in an energy-deficient state. Conversely, when the initial remaining energy exceeds 50%, the SIL operates in an energy-sufficient state. In the SIL-IEMS and SILGA schemes, priority is given to turning on the lamp during periods with a higher density of pests, especially when the energy is insufficient. In such cases, both SIL-IEMS and SILGA schemes preferentially activate the lamp during periods when pests are more active. As the initial remaining energy level increases, the number of pests trapped during the period gradually decreases. Consequently, for both the SIL-IEMS and SILGA schemes, the percentage of ineffective insecticidal energy consumption decreases until it stabilizes at 50% of the initial remaining energy. In contrast, the TRSM scheme primarily employs remote control, unable to adaptively adjust the insecticidal working time. Therefore, the insecticidal quantity only increases when there is sufficient energy. As a result, the effective insecticidal energy consumption percentage for the TRSM scheme exhibits an increasing trend until it levels off at 50% of the initial remaining energy.

Figure  11.  Comparisons of the average simulation results of effective insecticidal energy consumption and total energy consumption as a percentage. The bar graph represents the percentage of effective insecticidal energy and the line graph represents the percentage of total energy consumption to total energy (The bar chart shows the effective insecticidal energy consumption as a percentage of the three algorithms, represented by TRSM-A, SILGA-A and SIL-IEMS-A respectively; and the line chart shows the total energy consumption as a percentage of the three algorithms, represented by TRSM-B, SILGA-B and SIL-IEMS-B respectively. Where A represents the proportion of effective insecticidal energy consumption for the three algorithms; B represents the proportion of total energy consumption of the three algorithms, used to distinguish between the bar chart and line chart in the figure)

DownLoad: Full-Size Img PowerPoint

Furthermore, with an increase in the initial remaining energy, the total energy consumption of SIL also rises accordingly. When the energy is sufficient, the number of pests trapped reaches its maximum value and stabilizes. Consequently, the total energy consumption as a percentage of the initial remaining energy for all three schemes (SIL-IEMS, SILGA, and TRSM) follows a rising and then declining trend. It is evident that when the energy is insufficient, both SIL-IEMS and SILGA schemes outperform the TRSM scheme in terms of effective insecticidal energy consumption.

To investigate the impact of SIL’s initial remaining energy level on its insecticidal time, this paper presents the proportion of ineffective insecticidal energy to total consumed energy and the percentage of total energy consumption to initial remaining energy for different time slots. The simulation results of this experiment are presented in Table V. SIL operates with insufficient energy for insecticidal work when the initial remaining energy levels are 15%, 20%, and 30%. Whereas it operates with sufficient energy for insecticidal work when the initial remaining energy levels are 50%, 80%, and 100%. Under insufficient energy conditions, both the SILGA and SIL-IEMS demonstrate high energy utilization efficiency (where ineffective energy represents energy other than insecticidal energy). When energy is sufficient, the SILGA, SIL-IEMS, and TRSM schemes exhibit the same percentage of ineffective insecticidal energy to total energy consumed and total energy consumption to initial remaining energy in all time slots, except for the time slots between 18:00−19:00 and 4:00−5:00 (the next day).

Table  V.  The Average Experimental Results for Different Initial Remaining Energy Levels (%)

Initial remaining energy	Parameter	Method	Time slot
			18:00− 19:00	19:00− 20:00	20:00− 21:00	21:00− 22:00	22:00− 23:00	23:00− 0:00	0:00− 1:00	1:00− 2:00	2:00− 3:00	3:00− 4:00	4:00− 5:00
15	Ite	TRSM	96.44	95.15	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
		SILGA	0.00	0.00	0.00	0.00	0.00	0.00	0.00	63.83	0.00	0.00	0.00
		SIL-IEMS	0.00	0.00	0.00	0.00	0.00	0.00	0.00	64.02	65.04	0.00	0.00
	Tte	TRSM	21.11	12.19	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
		SILGA	0.00	0.00	0.00	0.00	0.00	0.00	0.00	20.41	0.00	0.00	0.00
		SIL-IEMS	0.00	0.00	0.00	0.00	0.00	0.00	0.00	27.03	1.25	0.00	0.00
20	Ite	TRSM	96.44	94.75	92.78	91.61	0.00	0.00	0.00	0.00	0.00	0.00	0.00
		SILGA	0.00	0.00	0.00	0.00	0.00	0.00	0.00	65.81	63.94	0.00	0.00
		SIL-IEMS	0.00	0.00	0.00	0.00	0.00	0.00	0.00	65.31	64.03	66.48	0.00
	Tte	TRSM	31.66	32.23	32.91	3.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
		SILGA	0.00	0.00	0.00	0.00	0.00	0.00	0.00	34.91	42.12	0.00	0.00
		SIL-IEMS	0.00	0.00	0.00	0.00	0.00	0.00	0.00	26.06	47.69	26.26	0.00
30	Ite	TRSM	96.44	94.75	92.78	90.49	87.27	82.06	0.00	0.00	0.00	0.00	0.00
		SILGA	0.00	0.00	0.00	0.00	0.00	78.30	73.79	66.53	64.03	67.29	0.00
		SIL-IEMS	0.00	0.00	0.00	0.00	78.41	73.79	66.53	64.03	68.00	74.15	0.00
	Tte	TRSM	15.83	16.11	16.46	16.87	17.49	17.21	0.00	0.00	0.00	0.00	0.00
		SILGA	0.00	0.00	0.00	0.00	0.00	0.91	20.69	22.95	23.84	18.37	0.00
		SIL-IEMS	0.00	0.00	0.00	0.00	1.44	20.69	22.95	23.84	22.45	8.63	0.00
50	Ite	TRSM	96.44	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	74.11
		SILGA	96.33	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	73.25
		SIL-IEMS	95.91	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	77.05
	Tte	TRSM	7.92	8.06	8.23	8.44	8.75	9.34	10.34	11.47	11.92	11.23	4.22
		SILGA	6.73	8.06	8.23	8.44	8.75	9.34	10.34	11.47	11.92	11.23	2.50
		SIL-IEMS	2.42	8.06	8.23	8.44	8.75	9.34	10.34	11.47	11.92	11.23	9.81
80	Ite	TRSM	96.44	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	77.10
		SILGA	95.71	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	76.79
		SIL-IEMS	96.44	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	76.90
	Tte	TRSM	4.52	4.60	4.70	4.82	5.00	5.33	5.91	6.56	6.81	6.42	5.66
		SILGA	0.22	4.60	4.70	4.82	5.00	5.33	5.91	6.56	6.81	6.42	4.14
		SIL-IEMS	4.52	4.60	4.70	4.82	5.00	5.33	5.91	6.56	6.81	6.42	5.45
100	Ite	TRSM	96.44	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	77.10
		SILGA	95.71	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	76.79
		SIL-IEMS	96.44	94.75	92.78	90.49	87.27	81.77	73.79	66.53	64.03	68.00	76.75
	Tte	TRSM	3.52	3.58	3.66	3.75	3.89	4.15	4.60	5.10	5.30	4.99	4.40
		SILGA	0.22	3.58	3.66	3.75	3.89	4.15	4.60	5.10	5.30	4.99	4.14
		SIL-IEMS	3.52	3.58	3.66	3.75	3.89	4.15	4.60	5.10	5.30	4.99	4.11
1 Ite represents ineffective insecticidal energy as a proportion of total consumed energy. 2 Tte represents total energy consumption as a proportion of total initial remaining energy.

 | Show Table

DownLoad: CSV

Based on the results shown in the figure, it can be concluded that the SIL system performs significantly better with the SIL-IEMS and SILGA schemes than with the TRSM scheme. Particularly, when the initial remaining energy falls within the range of 15%−40%, the SIL system using the SIL-IEMS and SILGA schemes exhibits notably higher insecticidal rates compared to the TRSM scheme, and both of SIL-IEMS and SILGA schemes demonstrate shorter durations of illumination. To make a more detailed comparison of the energy proportions for each time slot among the three schemes, Fig. 12 presents a comparison of the percentage of ineffective insecticidal energy to total consumed energy and the percentage of total energy consumption to initial remaining energy for different time slots. The four sets of experimental line charts in Fig. 12 represent the comparisons with initial remaining energy levels of 15%, 30%, 50%, and 100%. As SIL primarily operates in two states, namely insufficient energy and sufficient energy, the initial remaining energy levels of 15% and 30% represent situations where the SIL system has insufficient energy, while the initial remaining energy levels of 50% and 100% represent cases of sufficient energy. To provide a detailed comparison of the ineffective insecticidal energy and total energy consumption relative to initial remaining energy during different time slots, representatives with insufficient energy (15%, 30%) and sufficient energy (50%, 100%) are selected for analysis.

Figure  12.  Comparison of ineffective insecticidal energy as a proportion of total energy consumed and total energy consumed as a proportion of initial remaining energy for different time slots. Percentage of different energy consumptions in the time slot at 15%, 30%, 50%, 100% initial remaining energy. (a) Ineffective insecticidal energy as a percentage of total consumed energy when the initial remaining energy is 15%; (b) Total energy consumption as a percentage of initial remaining energy when the initial remaining energy is 15%; (c) Ineffective insecticidal energy as a percentage of total consumed energy when the initial remaining energy is 30%; (d) Total energy consumption as a percentage of initial remaining energy when the initial remaining energy is 30%; (e) Proportion of ineffective insecticidal energy to total consumed energy when the initial remaining energy is 50%; (f) The percentage of total energy consumed to the initial remaining energy when the initial remaining energy is 50%; (g) The percentage of ineffective insecticidal energy to the total energy consumed when the initial remaining energy is 100%; (h) The percentage of total energy consumed to the initial remaining energy when the initial remaining energy is 100%.

DownLoad: Full-Size Img PowerPoint

VI. Conclusion and Future Work

In this paper, we have proposed a new scheme for energy management in SIL-IoT, the SIL-IEMS. This scheme takes into account several crucial factors, including initial remaining energy and energy consumption. Through a systematic exploration involving simulated experiments, we have substantiated the effectiveness of the proposed approach in intelligently adjusting the on/off SIL’s insecticidal working time, while working within the confines of limited energy resources. The simulation results indicate that both the SIL-IEMS and SILGA schemes surpass the conventional TRSM in insecticidal efficiency when the initial remaining energy is at or above a threshold of 40%, with SIL-IEMS exhibiting a remarkable 17.6% increase in insecticidal effectiveness. Concurrently, the insecticidal working time of the SIL is significantly reduced under the SIL-IEMS, thereby enhancing energy utilization efficiency markedly. Furthermore, at a starting energy level of 15%, the SIL-IEMS achieves a 6% improvement in insecticidal efficiency over the SILGA method. This optimized strategy not only effectively mitigates the issue of battery deep discharge but also offers a comprehensive solution to bolster pest management capabilities.

In our research, we considered the variation in pest conditions under sunny and rainless conditions. However, it is important to note that various environmental factors can influence the curve depicting pest distribution. Therefore, future research directions will delve into the distribution of pests and their patterns of change under the varying weather conditions. We plan to enhance the adaptability and universality of our approach by adopting more flexible predictive mathematical models and adjustment mechanisms to better accommodate a wider range of applications. Additionally, we will explore the potential for simplifying the model and implementation strategies to reduce the need for specialized knowledge among operators and streamline the method’s application process. These optimization strategies are expected to not only effectively alleviate the issue of deep battery discharge but also provide a more comprehensive enhancement scheme for pest management, while simultaneously improving the energy efficiency and insecticidal performance of SIL.