A network coverage optimization method, system and storage medium for fire monitoring

Abstract

The present invention provides a network coverage optimization method, system, and storage medium for fire monitoring. The network coverage optimization method includes the following steps: Step A, using a nonlinear control strategy and a Lévy flight mechanism to improve the SOA algorithm to obtain an improved ISOA algorithm; Step B, combining the ISOA algorithm with the PSO algorithm to form an ISOAPSO algorithm. The ISOAPSO algorithm includes inner-layer optimization and outer-layer optimization. When solving the network coverage optimization problem for fire monitoring, the inner-layer ISOA algorithm is first used to optimize sensor nodes. The sensor nodes optimized by the ISOA algorithm are then transferred to the outer-layer PSO algorithm for secondary optimization, thereby achieving the purpose of improving network coverage. The beneficial effect of the present invention is that the network coverage optimization method of the present invention can better solve the network coverage optimization problem for fire monitoring and enhance fire monitoring capabilities.

Description

Technical Field

[0001] The present invention relates to the field of fire monitoring, and in particular to a network coverage optimization method, system and storage medium for fire monitoring. Background Art

[0002] Wireless sensor networks (WSNs) [1,2], as a key technology in the physical network, are responsible for data collection and transmission. Wireless sensors, composed of a large number of sensing nodes, are devices with data collection, computing, and communication capabilities. They are widely used in agriculture, oceanography, environmental monitoring, and safety monitoring [3-6]. To prevent fires, fire wireless sensor networks are deployed to monitor temperature, humidity, gas, and smoke, and transmit this data to users via the network. If anomalies are detected in the received data, users can take timely emergency measures to address the situation.

[0003] Coverage is an important indicator for measuring sensor networks, which determines the monitoring quality of fire wireless sensor networks. If sensor nodes are randomly deployed in space, it will cause problems such as reduced coverage and poor monitoring quality. Therefore, good node deployment can not only improve monitoring quality [7], but also reduce energy loss [8,9]. In recent years, research on sensor node coverage based on optimization has attracted great interest from scholars. Research on network coverage optimization can be divided into the following four categories:

[0004] Node optimization deployment based on virtual force algorithm. The basic idea of ​​this algorithm is to regard the sensor nodes in the deployment area as positively charged charges. Adjacent nodes can attract or repel each other, and the relationship between the two is used to move the nodes to the optimal position, thereby effectively improving the network coverage. Howard et al.

[10] first introduced the virtual force algorithm into the optimization deployment of wireless sensor networks. Zou et al.

[11] proposed a typical mobile deployment optimization method - Virtual Force Algorithm (VFA) based on the reference

[10] . Miao et al.

[12] proposed a wireless sensor network node deployment optimization algorithm based on a three-dimensional virtual force model, thereby realizing autonomous node optimization deployment in a three-dimensional space obstacle environment. Song et al.

[13] proposed a mobile sensor network coverage algorithm based on virtual molecular force and proved the effectiveness of this method through experiments. Luo et al.

[14] proposed a 3-D virtual force coverage algorithm, which can improve the network coverage. When deploying wireless sensor network nodes, the use of virtual force algorithm can better achieve the coverage of wireless sensor networks. However, traditional virtual force algorithms usually use a fixed step size to limit the movement distance of the sensor. Due to the lack of constraints, node oscillation will occur in the later stage of deployment, resulting in ineffective movement of the sensor.

[0005] Coverage optimization based on computational geometry. The basic idea of ​​this method is to divide the target monitoring area into several grids, and then approximate the coverage of the area by whether the grids are covered by nodes

[15] . Xu et al.

[16] proposed a wireless sensor network coverage calculation method based on geometric decomposition, which can increase the effective coverage of WSN nodes. Tan et al.

[17] proposed a 3D space self-deployment algorithm based on weighted Voronoi diagram to achieve full coverage of the monitoring area. In the network coverage optimization problem, the coverage optimization method based on computational geometry reduces the number of nodes through estimation operations. However, this method has the problem of grid size, which leads to low calculation accuracy and is not conducive to large-scale deployment of sensor networks.

[0006] Coverage optimization based on intelligent optimization algorithm. The basic idea of ​​swarm intelligence algorithm is to perform global optimization in the entire group, and guide the group to move towards the global optimal solution according to the fitness value through continuous iteration. Wang et al.

[18] introduced resampling technology to the PSO algorithm and used this method to solve the coverage problem of IoT sensor networks. Miao et al.

[19] proposed a gray wolf optimizer with an enhanced hierarchical structure, and experimentally proved that this method has advantages in wireless sensor network coverage. Deepa et al.

[20] introduced the Lévy flight mechanism in the enhanced whale optimization algorithm to improve the algorithm performance and used it for the deployment of sensor nodes. Wang et al.

[21] proposed a wireless sensor network coverage control optimization algorithm based on combinatorial mathematics. Altahir et al.

[22] used a dynamic programming method to optimize the coverage of visual monitoring sensors. Liao et al.

[23] used the ACO algorithm to optimize the deployment of sensor nodes. Song et al.

[24] proposed a variable step size fruit fly algorithm to optimize the coverage of wireless sensor networks and experimentally proved the effectiveness of this method. Reference

[25] used a bacterial foraging scheme to optimize wireless sensor networks. Reference

[26] used the AFSA algorithm to enhance the monitoring quality of WSNs. Swarm intelligence algorithms provide an effective method for deploying sensor nodes, thereby improving the service performance of the entire network. However, such algorithms are prone to falling into local optimality, so it is necessary to propose a reasonable solution to alleviate this shortcoming.

[0007] The Seagull Optimization Algorithm (SOA) is a new swarm optimization algorithm proposed by Dhiman et al.

[27] to simulate the migration and attack behavior of seagulls. Due to its advantages such as easy operation and high search efficiency, SOA has been applied in many fields. The research on SOA methods can be divided into two categories:

[0008] The first category is to introduce new search strategies. Since SOA is prone to falling into local optimal solutions, it is necessary to introduce search strategies to improve SOA so that the algorithm can effectively exert its optimization performance. In 2020, Dhiman et al. proposed the evolutionary multi-objective seagull optimization algorithm (EMoSOA)

[28] and a new multi-objective algorithm (MOSOA)

[29] for global optimization, and verified their feasibility on engineering optimization problems. Ewees et al.

[30] used the Lévy flight mechanism and mutation operator to improve SOA, so that the algorithm escaped from the local optimal solution. Li et al.

[31] introduced constrained non-dominant sorting and external archiving mechanisms to SOA, and used this method to evaluate the performance of the combined heat and power system. Reference

[32] introduced a collaborative optimization strategy to improve SOA and used it for testing IEEE8 and IEEE14 buses. Long et al.

[33] introduced a nonlinear escape energy factor and differential strategy based on the cosine function to SOA, and applied it to the estimation of volt model parameters. Che et al.

[34] proposed a mutual benefit mechanism and a symbiotic mechanism to improve the seagull optimization algorithm and used it to solve engineering optimization problems. The improved SOA has demonstrated good search efficiency and improved convergence accuracy on specific problems. However, the algorithm still has defects such as falling into local optimality and low solution accuracy on new problems.

[0009] The second category is the hybrid algorithm optimization strategy. The hybrid algorithm mainly integrates different algorithms to form a complementary relationship, thereby giving full play to the global search and local search capabilities of the algorithm. Wang et al.

[35] integrated the Yin-Yang pair idea into SOA, improving the performance of SOA. Mani et al.

[36] proposed a hybrid seagull optimization algorithm and used it for NLOS node detection. Almasri et al.

[37] mixed the sailfish optimization algorithm and the seagull optimization algorithm and used it for medical diagnosis. Jia et al.

[38] proposed a hybrid seagull optimization algorithm and used it for oil pollution image segmentation. Muthubalaji et al.

[39] combined the seagull optimization algorithm (SOA) and the owl search algorithm (OSA) and used it for energy management of the Internet of Things. The hybrid algorithm can enhance the diversity of the population, avoid the algorithm from falling into the local optimum, and thus improve the algorithm's optimization performance. The hybrid algorithm has a certain degree of singleness, and it can be considered to expand the algorithm to multiple application fields.

[0010] Although SOA has been successfully applied in many fields, research has found that this approach is prone to falling into local optimal solutions. In the implementation of network coverage optimization for fire monitoring, the following problems exist:

[0011] (1) When solving the network coverage optimization problem for fire monitoring, SOA often ignores the complexity of the problem, resulting in a decrease in network coverage.

[0012] (2) Due to the lack of population diversity in SOA, the algorithm is prone to fall into local optimality.

[0013] References

[0014] [1]Akyildiz,IF,Su,W.,et al.A Survey on Sensor Networks[J].IEEECommun.Mag.2002,40(8):102–114.

[0015] [2] F.Javed, MKAfzal, M.Sharif and B.-S.Kim. Internet of Things (IoT) Operating Systems Support, Networking Technologies, Applications, and Challenges: A Comparative Review [J]. IEEE Communications Surveys&Tutorials,2018,20(3):2062-2100.doi:10.1109 / COMST.2018.2817685.

[0016] [3]Jimmy LC, Choquehuanca-Zevallos JJ, ML Efraín. Sensor nodes fault detection for agricultural wireless sensor networks based on NMF[J]. Computers and Electronics in Agriculture, 2019,161:214-224.

[0017] [4]Li S,Qu W,Liu C,et al.Survey on high reliability wirelesscommunication for underwater sensor networks[J].Journal of Network andComputer Applications,2019,148(3):102446.doi:10.1016 / j.jnca.2019.102446.

[0018] [5]F.Xue,Y.Cai,Z.Cui.Bacterial foraging optimization algorithm forcoverage problem i-n wireless sensor network[J].Sensor Lett.2014,12(1):160-163.doi:http: / / dx.doi.org / 10.1166 / sl.2014.3234.

[0019] [6]C.H.Sun.A time variant log-linear learning approach to the SET k-COVER problem in wireless sensor networks[J].IEEE Transactions onCybernetics,2018,48(4):1316-1325.doi:10.1109 / TCYB.2017.2691772.

[0020] [7]Amitabha,Ghosh.Coverage and connectivity issues in wirelesssensornetworks:A survey[J].Pervasive and Mobile Computing,2008,4(3):303-334.doi:10.1016 / j.pmcj.2008.02.001.

[0021] [8]Q.Zhang,M.P.Fok.A two-phase coverage-enhancing algorithm forhybridwireless sensor networks[J].Sensors.2017,17(1):117.doi:10.3390 / s17010117. [9]Adulyasas A,Sun Z,Wang N.Connected Coverage Optimization forSensorScheduling i-n Wireless Sensor Networks[J].IEEE Sensors Journal,2015,15(7):3877-3892.doi:10.1109 / JSEN.2015.2395958.

[0022]

[10] Howard A.Mobile sensor network deployment using potential fields:A distributed,scalable solution to the area coverage problem[J].ProceedingsofDARS.2002:299-308.

[0023]

[11] Zou Y,Chakrabarty K.Sensor deployment and target localizationbasedon virtual forces[C].Joint Conference of the IEEE Computer&Communications IEEE Societies.IEEE,2003,2:1293-1303.

[0024]

[12] Miao C,Dai G,Zhao X M,et al.3D Self-Deployment Algorithm inMobileWireless Sensor Networks[C].China Conference on Wireless SensorNetworks.Springer,Berlin,Heidelberg,2015,11(4):721921.

[0025]

[13] Song L A,Rz B,Ys B.Design of coverage algorithm for mobile sensornetworks based on virtual molecular force[J].Computer Communications,2020,150:269-277.

[0026]

[14] C.Luo,Y.Cao,G.Xin,et al.Three-Dimensional Coverage OptimizationofUnderwater Nodes Under Multiconstraints Combined With Water Flow[J].IEEEInternet of Things Jour-nal,2022,9(3):2375-2389.doi:10.1109 / JIOT.2021.3094725.

[0027]

[15] Biagioni E S,Sasaki G H.Wireless Sensor Placement For Reliableand Efficient Data Collection[C].Proceedings of the 36th Annual HawaiiInternational Conference on System Sciences.2003.

[0028]

[16] Hui,X.,Bailing,W.,Jia,S.et al.An algorithm for calculatingcoverage rate of WSNs based on geometry decomposition approach[J].Peer-to-Peer Netw.Appl.12,568–576(2019).https: / / doi.org / 10.1007 / s12083-018-0653-1.

[0029]

[17] Tan L,Tang X,Hussain A,et al.A Weighted Voronoi Diagram-BasedSelf-Deployment Algorithm for Heterogeneous Directional Mobile SensorNetworksin Three-Dimensional Spa-ce[J].IEICE Transactions on Communications,2019:19-34.

[0030]

[18] Xiaohui,Wang,Hao,et al.Coverage Control of Sensor Networks in IoTBasedon RPSO[J].IEEE Internet of Things Journal,2018,5(5):3521-3532.doi:10.1109 / JIOT.2018.2829160.

[0031]

[19] Miao Z,Yuan X,Zhou F,et al.Grey wolf optimizer with an enhancedhierarchyand its application to the wireless sensor network coverageoptimizationproblem[J].Applied Soft Computing,2020,96:106602.doi:10.1016 / j.asoc.2020.106602.

[0032]

[20] Deepa R,Venkataraman R.Enhancing Whale Optimization Algorithmwith LevyFlight for coverage optimization in wireless sensor networks[J].Computers&Electrical Engineering,2021,94:107359.

[0033]

[21] Wang Y,Li M.Coverage Control Optimization Algorithm for WirelessSensor Networks Based on Combinatorial Mathematics[J].Mathematical ProblemsinEngineering,2021,1-8.doi:10.1155 / 2021 / 6066379.

[0034]

[22] Altahir A A,Asirvadam V S,Hamid N,et al.Optimizing VisualSurveillance Sensor C-overage Using Dynamic Programming[J].IEEE SensorsJournal,2017,17(11):3398-3405.doi:10.1109 / JSEN.2017.2694385.

[0035]

[23] W.H.Liao,Y.Kao,R.T.Wu.Ant colony optimization based sensordeployment protocol for wireless sensor networks[J].Expert Syst,2011,38(6):6599–6605.doi:10.1016 / j.eswa.2010.11.079.

[0036]

[24] Song R,Xu Z,Liu Y.Wireless Sensor Network Coverage OptimizationBased on Fruit Fly Algorithm[J].International Journal of Online Engineering(iJOE),2018,14(6):58-70.doi:10.3991 / ijoe.v14i06.8698.

[0037]

[25] A.A.A.Ari,I.Damakoa,A.Gueroui.Bacterial Foraging OptimizationSchemefor Mobile Sensing in Wireless Sensor Networks[J].International journalofwireless information networks,2017,24(3):254-267.

[0038]

[26] D.W.Wang,C.L.Wang.Wireless sensor networks coverage optimizationbasedon improved AFSA algorithm[J].International Journal of FutureGenerationCommunication and Networking,2015,8(1):99–108.

[0039]

[27] Dhiman G,Kumar V.Seagull optimization algorithm:Theory and itsapplications for lar-ge scale industrial engineering problems[J].Knowledge-Based Systems,2019,165(FEB.1):169-196.doi:10.1016 / j.knosys.2018.11.024.

[0040]

[28] Dhiman G,Singh K K,Slowik A,et al.EMoSOA:A New EvolutionaryMulti-objective Seagull Optimization Algorithm for Global Optimization[J].International Journal of Machine Learning and Cybernetics,2021,12:571-596.doi:10.1007 / s13042-020-01189-1.

[0041]

[29] Dhiman G,Singh K K,Soni M,et al.MOSOA:A New Multi-objectiveSeagull Optimi-zation Algorithm[J].Expert Systems with Applications,2021,167:114150.doi:10.1016 / j.eswa.2020.114150.

[0042]

[30] Ewees,A.A.,Mostafa,R.R.,Ghoniem,R.M.et al.Improved seagulloptimization algorithm using Lévy flight and mutation operator for featureselection[J].Neural Comput&Applic,2022,34:7437–7472.doi:10.1007 / s00521-021-06751-8.

[0043]

[31] Ling-Ling Li,Sheng-Jie Zheng,Ming-Lang Tseng,et al.Performanceassessment of co-mbined cooling,heating and power system operationstrategybased on multi-objective seagul-l optimization algorithm[J].EnergyConversion and Management,2021,224:114443.

[0044]

[32] Mohamed A,Essam H.H,Mohamed A.M,et al.An improved seagulloptimization algorithm for optimal coordination of distance and directionalover-current relays[J].Expert Systems with Applications,2022,200:116931.doi:10.1016 / j.eswa.2022.116931.

[0045]

[33] Wen Long,Jianjun Jiao,Ximing Liang,et al.Parameters estimationofphotovoltaic mo-dels using a novel hybrid seagull optimization algorithm[J].Energy,2022,249:123760.doi:10.1016 / j.energy.2022.123760.

[0046]

[34] Che Y,He D.An enhanced seagull optimization algorithm for solvingengineering optimization problems[J].Applied Intelligence(2022).doi:10.1007 / s10489-021-03155-y.

[0047]

[35] Wang J,Li Y,Hu G.Hybrid seagull optimization algorithm and itsengineering applicat-ion integrating Yin–Yang Pair idea[J].EngineeringwithComputers,2022,38(3):2821-2857.doi:10.1007 / s00366-021-01508-2.

[0048]

[36] Mani,R,Jayaraman,S,Ellappan,M.Hybrid seagull and thermal exchangeoptimization algorithm-based NLOS nodes detection technique for enhancingreliability under data dissemination in VANETs[J].Int J Commun Syst.2020;33:e4519.https: / / doi.org / 10.1002 / dac.4519.

[0049]

[37] Almasri,M.M.;Alajlan,A.M.Artificial Intelligence-BasedMultimodalMedical Image Fusion Using Hybrid S 2 Optimal CNN.Electronics 2022,11,2124.https: / / doi.org / 10.3390 / electronics11142124.

[0050]

[38] Jia,H.;Xing,Z.;Song,W.Three Dimensional Pulse Coupled NeuralNetwork Based on Hybrid Optimization Algorithm for Oil Pollution ImageSegmentation.Remote Sens.2019,11,1046.https: / / doi.org / 10.3390 / rs11091046.

[0051]

[39] Muthubalaji,S,Srinivasan,S,Lakshmanan,M.IoT based energymanagement in smart energy system:A hybrid SO 2 SA technique[J].Int J NumerModel.2021;34:2893.https: / / doi.org / 10.1002 / jnm.2893.

[0052]

[40] Choi C,Lee J J.Chaotic local search algorithm[J].Artificial Life&Robotics,1998,2(1):41-47.doi:10.1007 / BF02471151.

[0053]

[41] Li X,Niu P,Liu J.Combustion Optimization of a Boiler Based on theChaos and Lév-y Flight Vortex Search Algorithm[J].Applied MathematicalModelling,2018,58:3-18.doi:10.1016 / j.apm.2018.01.043.

[0054]

[42] Kennedy J.Particle swarm optimization[J].Proc.of 1995 IEEEInt.Conf.Neural Networks,(Perth,Australia),Nov.27-Dec.2011,4(8):1942-1948.

[43] Eberhart R C,Shi Y.Particle swarm optimization:Developments,applications and resources[C]. / / 2001 Congress on EvolutionaryComputation.Piscataway,USA:IEEE,2001,1:81-86.doi:10.1109 / CEC.2001.934374.

[0055]

[44] S.Mirjalili.Moth-flame optimization algorithm:A novel nature-inspir ed heuristic paradigm[J]Knowl.-Based Syst,2015,89:228–249.doi:10.1016 / j.knosys.2015.07.006.

[0056]

[45] Arora S,Singh S.Butterfly optimization algorithm:a novel approachfor global optimization[J].Soft Computing,2019,23(3):715-734.doi:10.1007 / s00500-018-3102-4.

[0057]

[46] S.Mirjalili,S.M.Mirjalili,A.Lewis.Grey wolf optimizer[J].Adv.Eng.Softw,2014,69(3):46–61.doi:10.1016 / j.advengsoft.2013.12.007.

[0058]

[47] S.Mirjalili,A.Lewis.The whale optimization algorithm[J].Adv.Eng.Softw,2016,95:51-67.doi:10.1016 / j.advengsoft.2016.01.008.

[0059]

[48] ​​Chen

[0060]

[49] Cao Y, Li Y, Zhang G, et al. Experimental modeling of PEM fuel cells using a new improved seagull optimization algorithm [J]. Energy Reports, 2019, 5: 1616-1625. doi: 10.1016 / j.egyr.2019.11.013. Summary of the Invention

[0061] The present invention provides a network coverage optimization method for fire monitoring, comprising the following steps:

[0062] In step A, the SOA algorithm is improved by using nonlinear control strategy and Lévy flight mechanism to obtain the improved ISOA algorithm.

[0063] In step B, the ISOA algorithm and the PSO algorithm are mixed to form an ISOAPSO algorithm. The ISOAPSO algorithm includes inner layer optimization and outer layer optimization. In solving the network coverage optimization problem for fire monitoring, the inner layer ISOA algorithm is first used to optimize the sensor nodes. Then, the sensor nodes optimized by the ISOA algorithm are transferred to the outer layer PSO algorithm for secondary optimization, thereby achieving the purpose of improving network coverage.

[0064] As a further improvement of the present invention, in step A, a nonlinear control strategy is proposed for formula (2) of the SOA algorithm. The formula of the nonlinear control strategy is as follows:

[0065] A=f c / (1+e ((20×t) / Maxiter-10) ) (11)

[0066] Among them, A is the control factor, f c is the frequency, Maxiter is the maximum number of iterations, t is the number of iterations, and e is the base value of a natural logarithm.

[0067] As a further improvement of the present invention, in step A, the formula (10) of the SOA algorithm is s (t+1)=D s (t)×x×y×z+gbest(t) is improved as follows: First, the position of the seagull is guided by the current individual optimal position and the current population optimal position, thereby improving the diversity of the population. Then, the Lévy flight mechanism is introduced to improve the local optimization ability of the SOA algorithm. The formula is as follows:

[0068] P b (t)=(pbest(t)-Ps(t))×rand (15)

[0069] P g (t)=(gbest(t)-Ps(t))×rand (16)

[0070] P r (t+1)=P s (t+1)+P b (t)×Lévy(s)+P g (t)×Lévy(s)(17)

[0071] Among them, P b represents the new position of the seagull learned from the current individual optimal position, pbest represents the current individual optimal position, P g represents the new position of the seagull learned from the current population's optimal position, gbest represents the current population's optimal position, rand is a random number between [0,1], P r represents the new position of the seagull, Lévy(s) is the step size of the Lévy distribution, and P s Indicates the position where the seagull attacks its prey in a spiral.

[0072] As a further improvement of the present invention, in step B, the ISOAPSO algorithm includes the following steps:

[0073] Step 1, Start.

[0074] Step 2: Initialize parameters and generate initial positions. Initialization parameters include the maximum number of iterations Maxiter, the current number of iterations t, and the constant f c ,u,v,Maxiter,c1,c2,W max ,W min ,V max ,V min ,N and other parameters.

[0075] Step 3: Calculate the initial fitness value, individual historical optimal position and group optimal position.

[0076] Step 4: Sort the individuals by fitness value from large to small and select the top N individuals with good fitness.

[0077] Step 5: Determine whether the current number of iterations t is less than or equal to the maximum number of iterations Maxiter (ie, t≤Maxiter). If yes, proceed to the next step; if not, proceed to step 9.

[0078] Step 6: Update the position of the inner ISOA algorithm.

[0079] Step 7: Speed ​​and position update of the outer PSO algorithm.

[0080] Step 8: Update the fitness value, individual historical optimal position and group optimal position.

[0081] Step 9: Determine whether the current number of iterations reaches the maximum number of iterations, that is, t>Maxiter. If so, the iteration ends and executes step 10; otherwise, return to step 5.

[0082] Step 10: Output the optimal position and fitness value.

[0083] As a further improvement of the present invention, in step 6, the inner layer optimization adopts the ISOA algorithm. In the network coverage optimization problem, the ISOA algorithm uses formula (1), formula (11), formula (3-9) and formula (17) to update the position of the sensor node and output the result P r =(P r1 ,P r2 ,...,P rn ) is used as the initial control trajectory of the outer optimization PSO algorithm for secondary optimization.

[0084] C s (t) = A × P s (t) (1)

[0085] Among them, C s and P s Represent the new position and current position of the seagull group respectively, and t represents the current number of iterations.

[0086] M s (t) = B × (gbest(t) - P s (t)) (3)

[0087] Among them, M s is the new seagull position, gbest is the best seagull position in the current population, and the calculation formula of B is:

[0088] B=2×A×A×rd (4)

[0089] Among them, B is responsible for balancing local search and global search, and rd is a random value between [0,1];

[0090] D s (t)=|C s (t)+M s (t)| (5)

[0091] Among them, D s Refers to the new location the seagull has arrived at.

[0092] x=r×cos(α) (6)

[0093] y=r×sin(α)(7)

[0094] z=r×α(8)

[0095] r=u×e αv (9)

[0096] Where r is the radius of each seagull, α is a random number between [0, 2π], e is the base of natural logarithms, and u and v are constants.

[0097] As a further improvement of the present invention, in step 7, the outer layer optimization adopts the PSO algorithm, and the sensor nodes of the inner layer ISOA algorithm are used as the initial control trajectory. The nodes are further optimized and the speed and position are updated using formulas (18-20) to obtain the optimal coverage and the optimal sensor node position.

[0098]

[0099]

[0100] Among them, r1 and r2 are random numbers between [0,1], c1 and c2 represent the individual learning factor and social learning factor of the particle, t represents the current number of iterations, V id is the particle velocity, P id is the optimal value of the particle individual, P gd is the group optimal value.

[0101]

[0102] Among them, w max is the initial weight, w min is the final weight, and Maxiter is the maximum number of iterations.

[0103] The present invention also discloses a system for optimizing network coverage for fire monitoring, comprising: a memory, a processor, and a computer program stored in the memory, wherein the computer program is configured to implement the steps of the network coverage optimization method of the present invention when called by the processor.

[0104] The present invention also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the network coverage optimization method of the present invention when called by a processor.

[0105] The beneficial effects of the present invention are: 1. The network coverage optimization method of the present invention proposes a nonlinear decreasing control strategy, which can better balance the global search and local search of the SOA algorithm, better solve the network coverage optimization problem for fire monitoring, and improve the fire monitoring capability; 2. The network coverage optimization method of the present invention can improve the coverage rate of the algorithm in the fire wireless sensor network. BRIEF DESCRIPTION OF THE DRAWINGS

[0106] Figure 1 It is a comparison diagram of parameter A of the present invention;

[0107] Figure 2 It is an execution flow chart of the ISOAPSO algorithm of the present invention;

[0108] Figure 3 It is a comparison result diagram of functions F1-F12 of the present invention;

[0109] Figure 4 1 is a comparison result diagram of functions F13-F18 of the present invention;

[0110] Figure 5 is the average coverage graph of cases (1-3) of the present invention;

[0111] Figure 6 is the average coverage graph of Invention Case 4;

[0112] Figure 7 This is the best coverage curve of the first three cases of the present invention;

[0113] Figure 8 This is the optimal coverage curve of Case 4 of the present invention;

[0114] Figure 9 is a diagram of the node distribution optimization process of ISOAPSO in case (3-4) of the present invention; (a) is a diagram of the node distribution optimization process of case 3 of the present invention; (b) is the 35-node distribution optimization process of case 4; (c) is a diagram of the 45-node distribution optimization process of case 4. DETAILED DESCRIPTION

[0115] Fire monitoring is of great significance. The use of new technologies to rationally deploy fire wireless sensor networks and optimize network coverage can achieve good fire monitoring. Therefore, based on the analysis of the network coverage optimization problem for fire monitoring, the present invention proposes an improved seagull optimization algorithm-particle swarm optimization algorithm (ISOAPSO). In the ISOAPSO algorithm, nonlinear control strategies, Lévy flight mechanisms and hybrid optimization control strategies are introduced. This algorithm can improve the coverage of fire wireless sensor network deployment and thus enhance fire monitoring capabilities. The present invention compares and analyzes ISOAPSO and other 7 algorithms on 18 benchmark functions, and theoretically verifies the effectiveness of ISOAPSO. In the implementation of network coverage optimization for fire monitoring, ISOAPSO achieves the best coverage, verifying that this method has good fire monitoring capabilities.

[0116] Based on the characteristics of the network coverage optimization problem for fire monitoring, this paper proposes an improved Seagull Optimization Algorithm - Particle Swarm Optimization (ISOAPSO). ISOAPSO mainly includes three key points:

[0117] (1) When solving the network coverage optimization problem for fire monitoring, SOA often ignores the complexity of the problem, resulting in reduced network coverage. Therefore, this paper proposes a nonlinear decreasing control strategy to better balance the global and local search of SOA, which can better solve the network coverage optimization problem for fire monitoring and improve fire monitoring capabilities.

[0118] (2) Due to the lack of population diversity in SOA, the algorithm is prone to falling into local optimality. To address this problem, the present invention introduces two mechanisms. First, the current individual optimal position and the current population optimal position are used to jointly guide the position of the seagulls to enhance the diversity of the population. Second, the Lévy flight mechanism is introduced to prevent the algorithm from falling into local optimality, thereby improving the algorithm's coverage in the fire wireless sensor network.

[0119] (3) This paper improves the SOA by using a nonlinear control strategy and the Lévy flight mechanism, and proposes an improved seagull optimization algorithm (ISOA). Hybridizing ISOA with PSO can further improve network coverage. Specifically, the ISOA algorithm is used to solve the network coverage optimization problem, and the results of the ISOA algorithm are then used as the initial trajectory of the PSO algorithm for secondary optimization.

[0120] 1. Overview of Seagull Optimization Algorithm:

[0121] The SOA algorithm mainly consists of two steps, which are inspired by the migration and attack behaviors of seagull groups. During the migration process, the positions of individual seagulls are different to avoid collisions. During the attack process, each seagull attacks the prey in a spiral shape. Since SOA does not require crossover and mutation processes, the average running time of this algorithm is less than that of DE, GA, GSA and other algorithms

[19] . In the network coverage optimization problem for fire monitoring, the number of nodes in the optimal sensor deployment is large, and SOA can obtain a better sensor deployment plan in a shorter time. Algorithm 1 in Table 1 gives the pseudo code of SOA.

[0122] 1.1 Migration Behavior

[0123] Seagull migration behavior involves avoiding collisions, moving toward the optimal location within the population, and approaching the optimal location. The optimal location is the location of the individual seagull with the highest fitness value within the population. The algorithm for simulating seagull migration behavior consists of three steps. First, a control factor A is introduced to adjust the positions of the seagulls to avoid collisions. The formula is as follows:

[0124] C s (t) = A × P s (t) (1)

[0125] Among them, C s and P s They represent the new position and current position of the seagull group respectively, and t represents the number of iterations.

[0126] A=f c -(t×(f c / Maxiter)) (2)

[0127] Among them, formula (2) can avoid collisions between individual seagulls, f c =2 is the control parameter, and Maxiter is the maximum number of iterations. The value of A decreases linearly from 2 to 0 as the number of iterations increases. Secondly, the seagulls will move towards the location of the best seagull in the current population, that is, towards the direction of the seagull with the best fitness value. The formula is:

[0128] M s (t) = B × (gbest(t) - P s (t)) (3)

[0129] Among them, M s is the new seagull position, gbest is the best seagull position in the current population. The calculation formula of B is:

[0130] B=2×A×A×rd (4)

[0131] Among them, B is responsible for balancing local search and global search, and rd is a random value between [0,1].

[0132] Finally, the seagull group will move closer to the optimal seagull position, thus reaching the new position D s (t). The formula for this process is as follows:

[0133] D s (t)=|C s (t)+M s (t)| (5)

[0134] Among them, D s Refers to the new location the seagull has arrived at.

[0135] 1.2 Aggressive Behavior

[0136] When seagulls find prey, they spiral in the air, constantly changing their flight angle and speed, drawing on their migratory history and experience to attack their prey. This behavior can be described in the x, y, and z planes as follows:

[0137] x=r×cos(α) (6)

[0138] y=r×sin(α)(7)

[0139] z=r×α(8)

[0140] r=u×e αv (9)

[0141] Where r represents the radius of each seagull, α is a random number between [0, 2π], and e is the basis of natural logarithms. u and v are constants. According to formulas (1)-(9), the final position of the seagull is described as:

[0142] P s (t+1)=D s (t)×x×y×z+gbest(t) (10)

[0143] Among them, P s Indicates the position where the seagull attacks its prey in a spiral.

[0144]

[0145] 2. Improvement strategy of Seagull optimization algorithm

[0146] The key to optimizing network coverage for fire monitoring lies in improving network coverage. Therefore, this paper introduces a nonlinear control strategy and the Lévy flight mechanism to improve the SOA, proposing an improved Seagull Optimization Algorithm (ISOA). To improve ISOA's accuracy, ISOA and PSO are combined to form the ISOAPSO algorithm. This method is more suitable for solving network coverage optimization problems, improving network coverage and enhancing fire monitoring capabilities.

[0147] 2.1 Nonlinear control strategy

[0148] In formula (2), the parameter A is divided by the frequency f c The linear decrease of parameter A causes parameter B to fail to effectively balance global and local search. However, the actual search process is an extremely complex nonlinear process, and this strategy will reduce the algorithm's optimization performance. In the network coverage optimization problem, the number of sensor nodes is large, and SOA easily ignores the complexity of the network coverage optimization problem, making it difficult to reasonably deploy sensor nodes, resulting in low coverage. Therefore, the present invention proposes a nonlinear control strategy for formula (2).

[0149] A=f c / (1+e ((20×t) / Maxiter-10) ) (11)

[0150] Here, e is the base value of a natural logarithm.

[0151] Figure 1 The curves representing the parameter A of formula (2) and formula (11) are shown respectively. Figure 1 It can be seen that the parameter A of the nonlinear control strategy proposed in the present invention decreases nonlinearly from 2 to 0. Therefore, the nonlinear control strategy can better balance the global search and local search capabilities of the algorithm, making the optimization process nonlinear, which can reduce the complexity of solving the network coverage problem.

[0152] 2.2 Lévy flight mechanism

[0153] The random step size of the Lévy flight mechanism [40,41] can expand the algorithm's search space and help the algorithm escape local optima. This mechanism can address the important shortcoming of SOA, which is that it is easy to fall into local optima. The Lévy flight mechanism is as follows:

[0154] Le(s)≈|s| -1-β (12)

[0155]

[0156]

[0157] Among them, μ and ν are random numbers from normal distribution, σ v =1, β=1.5, Γ(.) is a Gamma function, s represents the step size, and β represents the Lévy exponent.

[0158] During the migration process, the position of the seagulls in the SOA is only guided by the current optimal position of the population, resulting in a decrease in population diversity. By allowing the seagulls to learn from the current individual optimal position and the current optimal position of the population, the diversity of the population can be improved. The formula is as follows:

[0159] P b (t)=(pbest(t)-Ps(t))×rand (15)

[0160] P g (t)=(gbest(t)-Ps(t))×rand (16)

[0161] Among them, P b represents the new position of the seagull learned from the current individual optimal position, and pbest represents the current individual optimal position. g represents the new position of the seagull learned from the current optimal position of the population, gbest represents the current optimal position of the population, and rand is a random number between [0,1].

[0162] In order to improve the population diversity of SOA and prevent the algorithm from falling into local optimality, the present invention makes two improvements to formula (10): first, the position of the seagull is guided by the current individual optimal position and the current optimal position of the population, thereby improving the diversity of the population; second, the Lévy flight mechanism is introduced to improve the local optimization ability of the algorithm and prevent the algorithm from falling into local optimality. The formula is as follows:

[0163] P r (t+1)=P s (t+1)+P b (t)×Lévy(s)+P g (t)×Lévy(s) (17)

[0164] Among them, P r represents the new position of the seagull, and Lévy(s) is the step size of the Lévy distribution. Formula (17) enhances the diversity of the population and prevents the algorithm from falling into local optimality, which can improve the coverage rate of the network coverage problem.

[0165] 2.3 Hybrid Optimization Control Strategy

[0166] This paper improves on the SOA by using a nonlinear control strategy and the Lévy flight mechanism, proposing an improved Seagull Optimization Algorithm (ISOA). The ISOAPSO algorithm is then hybridized with the PSO to form the ISOAPSO algorithm. The ISOAPSO hybrid optimization control strategy consists of two parts: an inner optimization layer and an outer optimization layer. When solving the network coverage optimization problem, the inner ISOA algorithm is used to optimize sensor nodes. The optimized sensor nodes are then transferred to the outer PSO algorithm for secondary optimization. This hybrid strategy can improve network coverage.

[0167] 2.3.1 Inner Layer Optimization

[0168] The inner layer optimization adopts ISOA algorithm. The present invention uses nonlinear control strategy and Lévy flight mechanism to improve SOA. In the network coverage optimization problem, ISOA algorithm uses equation (1), equation (11), equation (3-9) and equation (17) to update the sensor nodes and output the result P r =(P r1 ,P r2 ,...,P rn ). The result is used as the initial control trajectory of the outer optimization PSO algorithm for secondary optimization.

[0169] 2.3.2 Outer Layer Optimization

[0170] Particle swarm optimization (PSO) is a new global search method proposed by Kennedy et al.

[42] , which is derived from the predation behavior of birds in nature. Assume that the swarm in the D-dimensional search space consists of N particles, and the current position of particle i is X i ={X i1 ,X i2 ,...,X iD}, the current speed is V i ={V i1 ,V i2 ,...,V iD}. The speed and position update formula is:

[0171]

[0172]

[0173] Among them, r1 and r2 are random numbers between [0,1], c1 and c2 represent the individual learning factor and social learning factor of the particle, and t represents the current iteration number. id is the particle velocity, P id is the optimal value of the particle individual, P gdis the optimal value of the group. In order to balance the global and local search capabilities of PSO, Shi and Eberhart proposed a linearly decreasing particle swarm algorithm

[43] , as shown in formula (20):

[0174]

[0175] Among them, w max is the initial weight, w min is the final weight, and Maxiter is the maximum number of iterations.

[0176] The outer layer optimization uses the PSO algorithm, and the sensor nodes of the inner layer ISOA algorithm are used as the initial trajectory to further optimize the sensor nodes. The speed and position are updated using equations (18-20) to obtain the optimal coverage and optimal sensor node position. Figure 2 The execution flow of the ISOAPSO algorithm is given. Algorithm 2 is the pseudo code of ISOAPSO.

[0177]

[0178] 3. Numerical optimization results and analysis

[0179] In order to test the performance of ISOAPSO, the present invention introduces 7 algorithms for comparative analysis. These algorithms include Moth Flame Optimization (MFO)

[44] , Butterfly Optimization Algorithm (BOA)

[45] , Grey Wolf Optimizer (GWO)

[46] , Whale Optimization Algorithm (WOA)

[47] , Seagull Optimization Algorithm (SOA)

[27] , Improved Seagull Optimization Algorithm (ISOA)

[48] , and Balanced Seagull Optimization Algorithm (BSOA)

[49] . At the same time, the present invention uses 18 benchmark functions to test the proposed method, see Appendix Tables A.1-A.3. Tables A.1-A.3 show the unimodal benchmark function, multimodal benchmark function, and fixed-dimensional multimodal benchmark function, respectively. All experiments were run on MATLAB R2017a.

[0180] 3.1 Comparison of ISOAPSO and other algorithms

[0181] In order to test the performance of ISOAPSO, the parameters of different algorithms are given in Table 1. The present invention performs a Wilcoxon rank sum test at a significance level of 0.05, where '+', '≈' and '-' respectively indicate that the performance of ISOAPSO is better than, similar to and lower than the test method. Table 2-3 records the average value (AV), standard deviation (SD), rank (R) and Wilcoxon rank sum test result (T) of each algorithm. Each algorithm is executed independently 20 times, and the 8 algorithms calculate the optimal fitness value of each benchmark function. The average value (AV) and standard deviation (SD) of each algorithm are calculated according to the optimal fitness value. All algorithms are sorted according to the average value (AV), and their rank (R) results are shown in Table 2-3.

[0182] Table 1 Algorithm parameter settings

[0183]

[0184] 3.2 Analysis of numerical results

[0185] Table 2 shows the test results for unimodal and multimodal benchmark functions. For unimodal benchmark functions (F1-F7), ISOAPSO outperformed the other seven methods on functions (F1-F4, F6, and F7). Furthermore, ISOAPSO found theoretical values ​​on four functions (F1-F4). For function F5, BOA showed the best average value, but ISOAPSO's average value was better than that of the other six methods. For multimodal functions (F8-F12), ISOAPSO located the theoretical optimum on five functions, demonstrating that ISOAPSO's Lévy flight mechanism can prevent the algorithm from falling into local optima.

[0186] Table 3 shows the results of the fixed-dimensional multimodal benchmark functions. Each method has different advantages. ISOAPSO achieved the best mean and standard deviation values ​​on three functions (F13, F16, and F18). MFO achieved the best mean and standard deviation on F14, and the average value of ISOAPSO was better than BOA, SOA, ISOA, and BSOA. GWO had advantages on functions F15 and F17, and the average value of ISOAPSO was better than SOA, ISOA, and BSOA. This shows that ISOAPSO has good search capabilities and that the hybrid control strategy of this method can improve the solution accuracy of the algorithm. As can be seen from Table 4, the ranking of the algorithms is IISOAPSO>GWO>BSOA>WOA>SOA>ISOA>BOA>MFO.

[0187] Table 2 Results of unimodal and multimodal benchmark functions

[0188]

[0189] Table 3 Results of fixed-dimensional multimodal benchmark functions

[0190]

[0191] Table 4 Overall Wilcoxon rank sum test results and ranking results

[0192]

[0193] 3.3 Convergence Analysis

[0194] Figure 3 is the convergence curve of the single-mode function (F1-F7) and the convergence curve of the multi-mode function (F8-F12), Figure 4 is the convergence curve of the fixed-dimensional multimodal benchmark function.

[0195] like Figure 3 As shown in the figure, ISOAPSO converges quickly on six functions (F1, F2, F3, F4, F8, and F10), and some WOA and GWO solutions approach theoretical values. ISOAPSO achieves the best convergence accuracy on five functions (F6, F7, F9, F11, and F12), while GWO and WOA also demonstrate good search capabilities. For F5, the BOA method demonstrates the best search capability, but ISOAPSO's convergence accuracy is significantly better than the other six methods.

[0196] like Figure 4 As shown in the figure, ISOAPSO has the best convergence accuracy on three functions (F13, F16, and F18). The convergence curves of functions F14 and F15 show that ISOAPSO's search accuracy is better than SOA, ISOA, and BSOA. Function F17 shows that ISOAPSO and GWO have similar convergence accuracy, but ISOAPSO converges faster.

[0197] 4. Network coverage optimization for fire monitoring

[0198] In terms of fire monitoring, the rational deployment of fire wireless sensor networks can facilitate overall fire prevention efforts and prevent losses. Fire wireless sensor networks primarily monitor and collect data such as temperature, humidity, and smoke. After processing, the collected data is transmitted to terminal devices for user service. Fire wireless sensor networks primarily consist of fire sensor nodes, sink nodes, and a network. Fire sensor nodes are primarily responsible for collecting information such as temperature, humidity, and images, while sink nodes process the data. The management center receives data from sink nodes via the network, allowing users to access data from the detection area. Therefore, the rational deployment of fire sensor nodes not only improves network coverage but also achieves the purpose of fire monitoring.

[0199] 4.1 Coverage Model of Fire Wireless Sensor Network

[0200] The coverage of a fire sensor node is a fixed circle. The distance between points is calculated to determine whether a location is within the coverage of the node.

[0201] Assume that there are N fire sensor nodes in the monitoring area, the sensing radius of the fire sensor node is R, and the node set can be described as S = {s1, s2, ..., s N}. Among them, node s i is (x i ,y i ), target node t j is (x j ,y j ), then node s i With t j The distance is:

[0202]

[0203] In the monitoring area, as long as the target node t j Fire sensor nodes i The target node is successfully sensed within the sensing range. The present invention adopts the Boolean perception model, and the formula is as follows:

[0204]

[0205] When p(s i ,t j ) is 1, indicating that the sensor node s i Covering the target node t j ; When p(s i ,t j ) is 0, indicating the target node t j Failed to be sensed successfully. Since multiple fire sensor nodes are required to conduct collaborative monitoring in the monitoring area, the fire sensor node set S senses the target node t j The probability is:

[0206]

[0207]

[0208] 4.2 Experimental simulation and analysis

[0209] The present invention tested 4 cases in the two-dimensional monitoring area

[0210] Case 1: The size of the monitoring area is 5×5m, the number of fire sensor nodes is 10, and the sensing radius R is 1m.

[0211] Case 2: The size of the monitoring area is 10×10m, the number of fire sensor nodes is 40, and the sensing radius R is 1m.

[0212] Case 3: The size of the monitoring area is 30×30m, the number of fire sensor nodes is 40, and the sensing radius R is 3m.

[0213] Case 4: The size of the monitoring area is 50×50m, the number of fire sensor nodes is 35 and 45 respectively, and the sensing radius R is 5m.

[0214] In the four case studies, the present invention used formula (24) as the algorithm's fitness function. A larger fitness value indicates better coverage and stronger fire monitoring capabilities. To verify the feasibility of ISOAPSO, the present invention introduced six algorithms (PSO, SOA, ISOA, BSOA, WOA, and BOA) for testing. Each algorithm had a population size of 80, a maximum number of iterations of 500, and each algorithm was independently run 20 times.

[0215] like Figure 5 As shown in the figure, the average coverage of ISOAPSO in cases (1-3) is 96.01%, 88.71%, and 94.48%, respectively. Compared with the PSO method, the average coverage of ISOAPSO is improved by 1.45%, 0.27%, and 3.51%, respectively. In addition, the average coverage of ISOAPSO is better than that of the other five methods (BOA, WOA, SOA, ISOA, and BSOA), showing the best fire detection capability.

[0216] Figure 6 Figure 2 shows the average coverage of different algorithms in Case 4 (35 and 45 nodes). ISOAPSO achieved average coverage of 91.38% and 98.30%, respectively, demonstrating the best average coverage and fire detection capabilities. When the number of nodes in Case 4 increased from 35 to 45, ISOAPSO's average coverage and fire detection capabilities improved by 6.92%. Because the increase in nodes will produce a large number of local optimal solutions, the Lévy flight mechanism proposed in this paper prevents ISOAPSO from falling into local optimal solutions.

[0217] Figure 7 The best coverage curves of different algorithms on cases (1-3) are shown. ISOAPSO shows the best coverage and fire detection capabilities. Figure 8Figure 9 shows the optimal coverage curves for different algorithms in Case 4 (35 and 45 nodes). PSO and WOA methods demonstrate good coverage, but ISOAPSO achieves the best coverage. Figure 9 shows the node distribution of ISOAPSO in different monitoring areas. As the number of iterations increases, the number of overlapping nodes decreases, and coverage improves, indicating that ISOAPSO's fire monitoring capabilities are gradually improving.

[0218] Based on the characteristics of the network coverage optimization problem for fire monitoring, this paper uses a nonlinear decreasing strategy to balance the local and global search capabilities of the SOA. The current optimal position of the individual and the current optimal position of the population are used to jointly guide the positions of the seagulls to enhance population diversity. The Lévy flight mechanism is used to prevent the algorithm from falling into local optimal solutions. Finally, the ISOA is combined with the PSO to solve the network coverage optimization problem.

[0219] The present invention uses 18 benchmark functions to verify the theoretical effectiveness of ISOAPSO and experimentally compares ISOAPSO with seven other algorithms. The experiment confirms that ISOAPSO ranks first among all the tested methods. In the study of fire monitoring network coverage optimization, fire monitoring capability and coverage are proportional. The average coverage of ISOAPSO in cases (1-3) is better than that of the other six methods, and ISOAPSO shows good fire monitoring capabilities. In case 4, as the number of sensor nodes increases, the average coverage of all methods is improved, but ISOAPSO shows the best average coverage and fire monitoring capabilities.

[0220] The beneficial effects of the present invention are: 1. The network coverage optimization method of the present invention proposes a nonlinear decreasing control strategy, which can better balance the global search and local search capabilities of SOA, better solve the network coverage optimization problem for fire monitoring, and improve fire monitoring capabilities; 2. The network coverage optimization method of the present invention can improve the coverage rate of the algorithm in the fire wireless sensor network.

[0221] The above is a further detailed description of the present invention in conjunction with specific preferred embodiments, and the specific implementation of the present invention should not be considered to be limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be considered to fall within the scope of protection of the present invention.

Claims

1. A network coverage optimization method for fire monitoring, characterized in that: The following steps are involved: Step A, using nonlinear control strategy and Lévy flight mechanism to improve the SOA algorithm to obtain the improved ISOA algorithm; Step B: The ISOA algorithm and the PSO algorithm are mixed to form an ISOAPSO algorithm. The ISOAPSO algorithm includes inner optimization and outer optimization. When solving the network coverage optimization problem for fire monitoring, the inner ISOA algorithm is first used to optimize the sensor nodes. The sensor nodes optimized by the ISOA algorithm are then transferred to the outer PSO algorithm for secondary optimization, thereby achieving the purpose of improving network coverage. In step A, the formula (2) of the SOA algorithm is A nonlinear control strategy is proposed. The formula of the nonlinear control strategy is as follows: （11）， in, is the control factor, is the frequency, is the maximum number of iterations, Indicates the current iteration number, is the base value of a natural logarithm; In step A, the formula (10) of the SOA algorithm is The following improvements are made: First, the position of the seagulls is guided by the current individual optimal position and the current population optimal position, thereby improving the diversity of the population. Then, the Lévy flight mechanism is introduced to improve the local optimization ability of the SOA algorithm. The formula is as follows: （15）， （16）， （17）， in, Indicates the new location, represents the new position of the seagull learned from the current individual optimal position, represents the current individual optimal position, represents the new position of the seagull learned from the current optimal position of the population, represents the optimal position of the current population, is a random number between [0,1], represents the new position of the seagull, is the step size following the Lévy distribution, Indicates the position where the seagull attacks its prey in a spiral.

2. The network coverage optimization method according to claim 1, characterized in that: In step B, the ISOAPSO algorithm includes the following steps: Step 1, start; Step 2: Initialize parameters and generate initial positions. Initialization parameters include the maximum number of iterations. , current iteration number ,constant ,in, and is a normally distributed random number, and represents the individual learning factor and social learning factor of the particle, W max is the initial weight, W min is the final weight; Step 3, calculate the initial fitness value, individual historical optimal position and group optimal position; Step 4: Sort the fitness values ​​from large to small and select the top N individuals with the best fitness; Step 5: Determine the current number of iterations Is it less than or equal to the maximum number of iterations? ,Right now If yes, go to the next step, if not, go to step 9; Step 6: Position update of the inner ISOA algorithm; Step 7, speed and position update of the outer PSO algorithm; Step 8: Update the fitness value, individual historical optimal position and group optimal position; Step 9: Determine the current number of iterations Whether the maximum number of iterations has been reached ,Right now If yes, the iteration ends and executes step 10, otherwise returns to step 5; Step 10: Output the optimal position and fitness value.

3. The network coverage optimization method according to claim 2, characterized in that: In step 6, the inner layer optimization adopts the ISOA algorithm. In the network coverage optimization problem, the ISOA algorithm uses formula (1), formula (11), formula (3-9) and formula (17) to update the position of the sensor node and output the result As the initial control trajectory of the outer optimization PSO algorithm, secondary optimization is performed; （1）， in, and Represent the new position and current position of the seagull group respectively, Indicates the current iteration number; （3）， in, It's the new Seagull location, is the best seagull position in the current population, The calculation formula is: （4）， in, Responsible for balancing local search and global search, is a A random value between （5）， in, Refers to the new location the seagull has arrived at; （6）， （7）， （8）， （9）， in, represents the radius of each seagull, yes A random number between is the base of natural logarithms, and is a constant.

4. The network coverage optimization method according to claim 3, characterized in that: In step 7, the outer layer optimization adopts the PSO algorithm, takes the sensor nodes of the inner layer ISOA algorithm as the initial control trajectory, further optimizes the sensor nodes, and uses formulas (18-20) to update the speed and position, thereby obtaining the optimal coverage and optimal sensor node position; （18）， （19）， in, and yes A random number between and They represent the individual learning factor and social learning factor of particles, Indicates the current iteration number, is the velocity of the particle, is the optimal value of the particle individual, is the group optimal value; （20）， in, is the initial weight, is the final weight, is the maximum number of iterations.

5. A system for optimizing network coverage for fire monitoring, characterized in that: include: A memory, a processor, and a computer program stored in the memory, wherein the computer program is configured to implement the steps of the network coverage optimization method according to any one of claims 1 to 4 when called by the processor.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the network coverage optimization method according to any one of claims 1 to 4 when called by a processor.

Images