Research On Prediction Of Residual Current Parameter Causing Elecerical Fire

Electrical fire is closely related to people's live and life safety.Therefore,electrical safety protection is an important research topic.Based on this background,this paper analyzes the electrical parameters that caused the fire by studying various data mining algorithms,so as to realize the prediction of residual current and the diagnosis of electrical equipment operating faults.And this method can effectively prevent the occurrence of electrical fire.Firstly,the paper introduces the origin and research purpose of the subject,and explains the sources of experimental data.The experiment is based on the electric fire monitoring system of a company in Zhejiang province.This paper introduces the working principle of the system and the acquisition of experimental data in detail.The system is applied to a number of devices,each of which has about 40,000 sets of 11 parameters data,among which the electrical parameters include three-phase voltage,three-phase current,temperature and residual current.In this paper,through the analysis of electrical theoretical knowledge,the causes of electric fire and the causes of false residual current,the experimental regular and analysis methods are found.Secondly,the data is processed and the mathematical model is established.The first step is to explore and analyze the data of a single device,including data visualization analysis,correlation analysis and so on.The second step is to predict the residual current through the combined model of grey neural network.From the result,the predicted value of residual current is accurate.However,the working principle of the residual current transformer and the cause of the residual current are analyzed and found that the predicted results cannot be used as the judgment basis for the occurrence of electrical fire.Therefore,it is necessary to further analyze the residual current data.In the third step,it is found that the false residual current contained in the system will affect the prediction results of the residual current parameters that cause electrical fire.Therefore,the method of solving the real residual current is obtained by using the least square method and time series synthesis.In the fourth step,the real residual current and current data were clustered and analyzed reasonably to obtain the judgment basis of equipment operation failure.Finally,the data of residual current and other parameter are used to train the two models respectively.Firstly,the residual current is predicted by using the neural network algorithm optimized by grey prediction.Then,SOM algorithm is used to realize clustering for the real residual current data.Finally,by observing the experimental results and combining with the actual situation in the field,the running state of the equipment in each category is judged.Thus,reasonable fault diagnosis and prediction of future experimental data can be performed.In the verification experiment,the other two devices were analyzed and the experimental results were compared.The correctness and effectiveness of the model are proved by field investigation of the fault prediction results of electrical equipment.Therefore,this method can effectively prevent electric fire.