Calculation of prediction parameters by probability integral method of improved GWO-BP algorithm

Aiming at the defects of BP neural network model in calculating the parameters predicted by probability integration method, a BP neural network parameter prediction model IGWO-BP based on improved gray wolf optimization (GWO) is proposed in this paper.By improving the convergence factor a and speed update formula of gray wolf algorithm, its optimization performance is better.The improved gray wolf optimization is used to optimize the initial weight and threshold of BP neural network, and the optimal initial weight and threshold are used to train and predict the model, so as to obtain the prediction results of the parameters of probability integral method.In order to verify the advantages of the algorithm in the parameter calculation of probability integral method, it is compared with the prediction results of SVR model and PLS model.From the calculation accuracy of the three methods from the three evaluation indexes of relative error, average absolute percentage error and consistency index, it is found that the maximum relative error of IGWO-BP algorithm.The average absolute percentage error and consistency index are better than the other two prediction methods, which shows that IGWO-BP algorithm can be applied to mining subsidence prediction by probability integral method.