Gas outburst predicting method based on EMD and ELM

Abstract

The invention discloses a gas outburst predicting method based on EMD (Empirical Mode Decomposition) and an ELM (Extreme Learning Machine). Firstly, the EDM is utilized for decomposing a gas concentration time sequence into a series of subsequences and residual volumes to reduce the computing scale of the local analyzing of gas concentration signals and improve the predicting accuracy; secondly, the ELM prediction is carried out on the subsequences and residual volumes; finally, the prediction values of the subsequences and the residual volumes are combined to obtain a final prediction value. According to the predicting method, the coal and gas outburst can be accurately predicted, convenience is provided for taking effective prevention measures in a targeted mode, and the prevention capacity of the gas accident is improved.

Description

technical field [0001] The invention relates to the field of coal mine safety production (prediction), specifically a gas outburst prediction method combining Empirical Mode Decomposition (EMD) and extreme learning machine (Extreme Learning Machine, ELM). Background technique [0002] As a primary energy source, coal is an important pillar of my country's national economic development. In my country, coal mines are mostly gas-containing mines, of which high-gas mines account for about 35%. Coal and gas outbursts are the primary problem that threatens the safety of coal mine production. Therefore, it is necessary to accurately predict coal and gas outburst in order to take effective preventive measures and improve the ability to prevent gas accidents. [0003] At present, there are mainly two types of methods for gas outburst prediction research: (1) gas outburst prediction based on the quantitative relationship between various relevant factors (geological and mining factors...

AI Technical Summary

Problems solved by technology

Among them, there are many researches on the gas outburst prediction model based on the neural network, but the traditional neural network needs repeated learning, the training speed is slow, and it is easy to fall into the local minimum point.

Images