Research On Series Fault Arc Identification And Fire Alarm Based On Multi-information Fusion

Electrical fires account for the largest proportion of the total number of fires per year,with the largest number of electrical fires caused by fault arcs.The complex physical and electrical response of the AC series fault arc makes the fault identification difficult,and the fault arc will vary greatly with the load depending on the load.Moreover,the single identification feature can have a large limitation.Therefore,this paper aims to extract the identification feature values of fault current waveforms under different loads through simulation,to solve the limitations of single features and improve the accuracy of identification,and to identify the method in a new type of intelligent fire monitoring and early warning system.Applied.In this paper,the advantages and disadvantages of the classical arc mathematical model are compared.The Cassie and Mayr arc mathematical models are selected as the comparison model.In MATLAB,the simulation modules of two mathematical models were successfully built,and the simulation comparison circuit diagram was established.The AC series fault arc current waveform obtained by the research was found to be more suitable for the simulation research.The simulation model is used to build the simulation circuit of different value pure resistive load,RLC load and single-phase asynchronous motor load.The identification characteristics of each simulation waveform are analyzed and its identification characteristics are analyzed.The classical pattern recognition method is compared,the support vector machine for the two classification is selected,and the inference is made.The Gaussian kernel function which can map the finite dimension to the high-dimensional space is selected,and the penalty coefficient and Gaussian radius are obtained by the 3-fold cross-validation algorithm and the mesh search method.The model training and training model test were carried out by the method of missing value,and theidentification accuracy was 99.897%.Through the feature contribution analysis,the limitation of single feature to fault identification and the correctness of multi-information fusion identification are proved.Analyze the operation principle and shortcomings of the intelligent fire monitoring and early warning system in "Smart Fire",and improve it to obtain a powerful intelligent learning ability of integrated machine algorithm,powerful computing power of cloud computing platform and real-time transmission characteristics of mobile communication.The monitoring and early warning system was used,and the application of arc fire was carried out by taking the detection and identification of the core arc fire as a sample.