Applied Research On The Artificial Neural Network In Power Load Identification Of Students' Dormitory

Along with the socialized reforms of the rear service, most universities and colleges have set no limit to the electricity consumed by students. As a result, the abuse of the large number of high-power electronic appliances brings about fires in dormitories frequently. The safe usage of electricity in university students'dormitories becomes an increasingly hot issue these days. Accordingly, how to identify these loads accurately and rapidly is the investigative content of dissertation.Based on the summarizing and analysis of the load varieties and characteristics in the students'dormitories, this article uses Fourier transformation to conduct a harmonic analysis to the load current combined the circuit model. There comes up the method of unsafe load identification in dormitory according as the changes of the power and the harmonic absolute area. This paper analyzed existed methods and techniques of harmonic detecting. According to the requirements of load identification for students'dormitories, the tenet of harmonic detecting in time domain was established. The convergence speed of harmonic current detecting algorithm basis of neuron is slow and it is unsuitable for the condition that load changes greatly. It is testified by this dissertation which the integral function of square difference between load current and detecting current can gain minimum value. Thus a harmonic detecting algorithm based on Hopfield neural network was given by establishing the relation between the integral function and the energy function of Hopfield neural network. This method which have many advantages such as with fast speed, high accuracy and good tracking performance can gain harmonic contents depending on adaptively detecting the fundamental component of load current as load current changes. Satisfactory results have been achieved in comparison with the performance index of harmonic current detecting algorithm basis of neuron by simulating under the MATLAB environment.The equivalent circuit model of typical nonlinear load was built. Unsafe load is identified by unsafe load identification method and harmonic current detecting algorithm based on Hopfield Neural Network. The validity and feasibility was showed by the simulation results.