Coal Spontaneous Combustion Prediction System Based On Improved Particle Swarm Wavelet Neural Network

In the past,the prediction of coal spontaneous combustion relied only on the data collected by a single or a small number of sensors.The accuracy of this type of method is low.The error of the data of the sensor caused by the influence of complex environment usually leads to misoperation.Multi-sensor technology is similar to the process in which the human brain processes multiple sensory information to obtain a consistent target state description.The accuracy of coal spontaneous combustion prediction system to predict the dangerous state of coal spontaneous combustion can be improved.The environment in which the sensor works is harsh and requires the design of a dedicated sensor measurement circuit.Therefore,this paper provides a dedicated measurement circuit for the indicator gas sensor.At the same time,environmental disturbances,circuit start-stop and other electromagnetic interference will bring noise to the measurement signal,so it is necessary.Noise reduction is performed on the signal containing.The traditional Fourier transform denoising method is suitable for non-time-varying signals,it is not suitable for time-varying signals or fast-changing signals.Processing time-varying signals can use wavelet de-noising.For the noise of time-varying signals,Wavelet transform noise reduction analysis processing is selected in this paper.The noise-reduced signal is sent to the network for training in the network.The data fusion algorithm used in this paper is based on the improved particle swarm optimization algorithm.The paper compares four models,namely the improved wavelet neural network model,SVM model and the original wavelet neural network model.The results show that the improved particle swarm neural network has the highest classification accuracy,and the training speed of the model is the fastest.This method overcomes the shortcomings of premature convergence in the traditional greedy algorithm.The nonlinear relationship between the various detection gases and the degree of spontaneous combustion of coal can be approximated by wavelet neural network based on wavelet analysis theory,which overcomes the blindness of traditional neural network model structure selection.In this paper,by using MATLAB simulation software,each model is trained and used in sequence.According to the experimental results,the training speed of wavelet neural network based on particle swarm and the superiority of classification accuracy using network can be clearly seen.Therefore,the research in this paper plays an important role in the prediction of coal spontaneous combustion in the safety process.