Anti-stealing Detection System Based On Optimized Neural Network Algorithm

Electricity resources have always been the material basis for people ’s survival.For economic purposes or weak legal awareness,electricity theft has always existed.In this paper,data fusion technology,remote centralized meter reading,data wireless transmission technology and computer technology are used as the main technologies to study the problem of stealing electricity in the society,and an anti-stealing electricity detection system is built to establish a visual detection platform.Through the visual detection platform,the suspected users of stealing electricity are effectively detected.In order to realize the system design proposed in this paper,multi-faceted research has been carried out.The specific contents are as follows :(1)The electricity information acquisition system is constructed,and the acquisition terminal uses the TQ-LTE-DTU wireless data collector to collect the data of the meter.It includes the incoming line side,the load side,the three-phase line,the antenna,the card slot,the RS485 interface,the USB interface and the power supply port.The antenna needs to be placed separately when used.The data transmission terminal is the main control terminal,and the smart meter is the slave station.Each smart meter has its own address code.In the task scheduling module,the multi-threaded acquisition method is used to assign tasks to the types of collected data by issuing commands from the main control terminal.GPRS wireless network is selected as the transmission channel for remote data transmission.(2)The algorithm anti-stealing detection model is established for the power-cutting eigenvalues and key technologies in the process of stealing electricity,and the clustering algorithm,recurrent neural network algorithm,convolutional neural network algorithm and BP neural network algorithm are tested on the detection of stealing electricity.The experimental results show that the anti-stealing detection model based on BP neural network algorithm has higher accuracy,but the BP neural network algorithm detection model has other shortcomings such as too long training time,which affects the performance of anti-stealing model detection.(3)Aiming at the shortcomings of BP neural network algorithm in the detection of electricity theft,an anti-stealing detection method based on BP neural network algorithm optimized by genetic algorithm is proposed.The genetic algorithm is used to determine the population range and encode it.The adaptive function and the selection function are introduced to calculate the fitness value to screen out the individuals with the highest survival rate and decode them,so as to determine the initial value parameters in the forward propagation process.The gradient descent method and the back propagation method update the neural network forward propagation parameters to reduce the number of iterations in the model training process.Combined with the test set data of the acquisition system,the simulation experiment is carried out by MATLAB.The simulation results show that this method improves the detection accuracy of the anti-stealing detection model,shortens the training time of the model,and is reliable in the detection of stealing electricity.(4)The visual detection platform GUI is designed and built for visual analysis.The tool is MATLAB R2020 b,and the electricity information data is tested by the anti-stealing detection system.The test results show that the anti-stealing detection system can detect the suspected users of stealing electricity intuitively and quickly and display them through the visual detection platform.In terms of detection effect,compared with the detection results of BP neural network algorithm,the accuracy of electricity theft detection is increased by 8.37 %,and the training time of the model is shortened by 89.5 %.It effectively broadens the application prospect of the system in the detection of electricity theft.