The Research On Algorithm Of The NILM In Circuit Fault Detection And Household Appliances Power Consumption Monitoring

With the growth of the consumer’s demand，the rapid development of technologyand the growing consumption of energy, the people’s life-style is changing. Thefunction of the electric appliances and the electronic products is further improved. Theyprovide convenient to people’s life. On the other hand, the variety of theseproducts throws down the challenge to monitor the security and stability. The traditionalway is high cost and low efficient, and it’s not suitable for monitoring the new one. Thenon-intrusive load monitoring (NILM) reduces the cost of the monitoring greatly. Thispaper studied the algorithm of the NILM from two aspects, one was the circuit faultdetection and the other one was on-line monitoring to the power consumption of thehousehold appliances.In the circuit fault detection, this paper researched the quantitative relationshipbetween the effect of the blind separation and the correlation of the source signals firstly.And then contraposing the specific circuit puted forward the incomplete type NILM forcircuit fault detection, which based on the FastICA algorithm. Finally this paper verifiedthe correctness of this algorithm by simulation.①This paper analysed the effection of the sine signal’s correlation from it’samplitude and frequency, and simulated a series of sine signals whose correlation isdistributed evenly. This paper tested the blind separation by blending two sine signalsinto the mixed signal randomly and linearly, and by using FastICA algorithm separatethe mixed signals, calculating the F(WA) which is the parameter of the separation effect,and then fitting correlation coefficient and F (WA) into linear function.②This paper dedected the range of the source signals’s correlation coefficient bycalculating the correlation coefficient between the original signals and the separatedsignals, and by puting forward the conditions of the correlation between original signalsand the separated signals, which satisfy the separation effect, then by blending theresine signals into the mixed signal randomly and linearly and using the same method toanalysis this situation.③This paper puted forward the methods and steps to set up the test circuit, andthe rules to debug the test voltage, and the process of the incomplete NILM which basedon the FastICA algorithm. This paper used Matlab to simulate a circuit fault signal, andbrought it into fault diagnosis algorithm to validate. Monitoring the power consumption on-line of the household appliances isimportant to the demand response and the management of electric energy using. Toachieve the function above needs the complete data support of each householdappliances which obtained by adding the complete NILM module into the smart meter.This paper puted forward the complete NILM algorithm of the identification of thehousehold appliances’s switch state based on the characteristic parameters.①This paper researched the steady and transient characteristics of the homeappliances, and extracted the characteristic parameters by sampling the steady state andtransient operation informations of the20appliances, and calculating the haracteristicparameters and establishing the feature parameters database for identification.②This paper puted forward the algorithm to identification for householdappliances’s cast and cut state, and the algorithm to identification the type of thehousehold appliances. The experiment results confirmed the proposed identificationalgorithm can correctly identified multi-device hybrid startup.