Research On Soft-sensing Model Of Gas Concentration Based On Chaotic LogWOA-ESN Network

The most important work is to reduce the accident rate of gas disaster.Because the mechanism of coal and gas outburst is complex,and the gas concentration has the characteristics of nonlinear dynamics,the traditional method to predict the gas concentration often has a large error.Neural network has strong fitting ability in nonlinear and time-varying system.Therefore,building a soft-sensing model by neural network method to predict gas concentration has become an effective means of coal and gas disaster prevention.In order to further improve the prediction effect of gas concentration,this paper uses the improved whale optimization algorithm to optimize ESN,to build a soft measurement model of-gas concentration based on chaotic LogWOA-ESN.The main research contents are as follows:First of all,in order to improve the ability of whale optimization algorithm(WOA),a whale optimization algorithm based on chaotic logarithmic nonlinear contraction factor is proposed,which is called chaotic LogWOA algorithm.There are two major improvements to this algorithm.In order to improve the search efficiency of the algorithm,the chaotic optimization technology is used to carry out Logistic chaotic mapping during the population initialization,so as to make the initial value distribution even and improve the quality of the solution.For the linear shrinkage factor in the original algorithm,a logarithmic nonlinear shrinkage factor model is proposed to improve the performance of whale optimization algorithm.By comparing the trig function,exponential function,power function and logarithmic function,the logarithmic contraction factor is selected as the best model to replace the linear shrinking factor.Experimental results show that the chaotic LogWOA algorithm proposed in this paper achieves good convergence accuracy and stability by using 4 unimodal functions,3 multimodal functions and 3 fixed dimensional functions as test functions.Secondly,aiming at the problem that the traditional echo state networks may produce pathological solution when it outputs weights,the chaotic LogWOA algorithm is used to optimize the process of its output weights,and the soft-sensing model of gas concentration of chaotic LogWOA-ESN is constructed.The empirical mode decomposition algorithm is used todenoise the actual gas concentd tion data in the mine,and the coordinate delay method isused to teconstruct the gas concentration time series.The core of model construction is adopted chaotic Low WOA algorithm to optimize the network output weight solution method and thw network error function is used to replace the original objective function for solution,and then output of prediction results.The experimental results show that the chaotic Log WOA-ESN soft-sensing model proposed in this paper is more accurate than other models,which proves the effectiveness of the model.