Research On Harmonic Detection System Of Neural Network Optimized By Genetic Algorithm

Electric power is one of the most widely used energy sources in today’s society,good electric power quality is the premise for the rapid development of various industries and the guarantee for people’s stable life.But in recent years,the harmonic is caused in power system by a large number of nonlinear,impact power equipment putting into use,it leads to the decline of electric power,which directly affects the social security,stability and economic operation,so it is urgent to solve the problem of harmonic,the harmonic detection is the primary part of solving harmonic problem,thus it is necessary to design a harmonic detection system which performance is good.The research of this paper is the algorithm and system design of harmonic detection.In terms of detection algorithm,this paper first introduces several common methods for harmonic detection at the present stage,makes analysis and comparison between them,then this paper decides to adopt neural network as the core algorithm of detection system.After that,this paper describes the BP network which is one type of neural network,and uses various improved BP networks to detect harmonics,it is found that the BP network using LM algorithm has more advantages in harmonic detection,but this improved BP network also has a drawback,it is sensitive to the initial weights and thresholds of the network,which will affect the detection results of harmonics,therefore,the strategy of using GA to optimize is put forward,and the simulation research is carried out,the result shows that the BP network optimized by GA has faster training speed and more accurate harmonic detection than the unoptimized network.However,the GA also has defects,so the AGA is proposed to optimize the harmonic detection scheme of BP network,and a simulation experiment is conducted,the result shows that the BP network optimized by the AGA has a faster training speed,which is conducive to improve the real-time performance of harmonic detection,and the results of harmonic detection are more accurate.In the aspect of system design,this paper designs the hardware circuit and software program.The hardware circuit is mainly studied from the three aspects: data acquisition module,data processing module and human-computer interaction module.The data processing module adopts the current practical chip S5P4418 as the main controller,its powerful data operation ability can improve the detection speed of thesystem for harmonics.Software program design including two aspects: the driver design of system and the application design,the driver is mainly introduced the WIFI and SD card drive design process,while the application is introduced the design of the main program and subroutine basic process,basic subroutine includes sampling program,data processing and display program and data transmission program.Finally,a whole set of harmonic detection system is designed by combining hardware and software.Figure [44] Table [8] Reference [56].