Research On Electrical Fire Detection System Based On Multi-sensor Information Fusion

With the daily popularization of electrical equipment and the rapid growth of power load,electrical fires occur from time to time.Especially for the power distribution room,where the electrical equipment is densely distributed and running for a long time,there are many potential electrical fire hazards.Once an electrical fire occurs,it will cause immeasurable losses.Therefore,how to carry out early prevention and accurate detection of electrical fire has become an important problem to be solved urgently.However,the traditional electrical fire detection has the problems of complicated wired transmission wiring and serious interference of various environmental information,which makes it difficult to adapt to the current electrical fire detection in complex environment.Based on this,this paper constructs a multi-sensor and multi-node electrical fire detection system,and applies the multi-sensor information fusion technology to the electrical fire detection system to further improve the timeliness and accuracy of electrical fire detection and the rationality of alarm decision-making.The main research work of this paper is as follows:(1)Aiming at the problems of dense distribution of electrical equipment and complex wired transmission wiring,Bluetooth Mesh networking technology is used to realize wireless transmission of multi-node detection information.Due to the limitation of communication distance of electrical fire nodes and the influence of metal objects in the detection area on detection information transmission,a spiral antenna with high gain and long-distance transmission is designed to extend the communication distance of networking nodes without adding electrical fire nodes,so as to ensure that the electrical fire nodes transmit detection information stably and reliably within the effective communication distance.(2)Aiming at the problem that the existing electrical fire detection methods can’t effectively use the time series characteristics of detection information,resulting in low detection accuracy,the time series characteristics of detection information are extracted by using the memory ability of the recurrent neural network.On this basis,an electrical fire feature recognition method based on LSTM-GRU network is proposed to improve the accuracy of electrical fire detection and recognition.(3)Aiming at the problem that the decision-making conditions of electrical fire detection and alarm are single,and the decision-making output is difficult to fully reflect the severity of electrical fire in the detection area,the existing electrical fire probability and auxiliary decision-making factors in the detection area are used for fuzzy reasoning fusion to improve the rationality of electrical fire alarm decision-making.After simulation and experimental tests,the electrical fire detection system designed in this paper has the advantages of simple node deployment,high reliability and low economic cost.It can meet the requirements of electrical fire detection in places with dense electrical equipment.At the same time,it can effectively realize the accurate detection,identification and hierarchical decision-making alarm of electrical fire,which is feasible and practical.