Research on Coal and Gas Outburst Warning from Missing of Samples

Abstract

In order to improve the prediction accuracy of coal and gas outburst in the case of missing data, a coal and gas outburst prediction model based on MICE_NN interpolation algorithm and improved Pathfinder Algorithm (IPFA) optimized Extreme Learning Machine (ELM) is proposed. Firstly, the correlation analysis of various indicators affected by coal and gas outburst is carried out, and the MICE_NN algorithm is used to interpolate the missing values, which is easy to obtain more sufficient information from incomplete data sets and improve the prediction effect and accuracy of the model. Secondly, the Pathfinder algorithm is jointly improved by introducing the evolutionary boundary constraint processing scheme, Levy flight strategy and group fitness variance strategy to improve its global optimization ability, so as to optimize the relevant parameters of ELM and construct the coal and gas outburst prediction model. Finally, the measured data interpolated by MICE_NN are used as samples for experimental verification, and the proposed algorithm is compared with single machine learning and ensemble algorithms. The results show that the data quality based on MICE_NN interpolation is significantly better than the data without interpolation. The classification accuracy, recall rate and   of IPFA_ELM model based on MICE_NN interpolation are significantly higher than those of other comparison models. It provides a new idea and method for coal and gas outburst prediction, and provides a strong reference basis for the next step of gas outburst prevention and control.