Coal Mine Safety Evaluation Research Based On Improved BP Neural Network

Not only China produces a huge number of coal; but also is a big consumer of coal. But the safety situation is very serious, in addition to the transportation industry, there are the most number of accidents of coal industry, accounting for more than 60% of industrial fatalities. Most people in our society keep eyes on coal mine safety issues. Timely, accurate, and effective coal mine safety evaluation has benefits to staffs of coal industry, coal mine enterprise, as well as is; good for the stability of society. This paper introduces the internal and external theory of coalmine safety evaluation. After analyzing the common hazard in coal production, a series of coal mine safety evaluation indexes have been built.This paper reviews the theory of coal mine safety at home and abroad and technological research at present; also analyses the strengths and weaknesses of qualitative evaluation and quantitative evaluation. On the base of our country's coal mine safety at present, illustrates the importance of coal mine safety assessment.Artificial neural network is an abstraction and simulation of the human brain and can change the value between the different neurons adaptively via inductive learning. It has good self-adaptability, self-organized and powerful self-study, association of ideas ability. Detailed description of the BP network using error back-propagation algorithm based on artificial neural network.The shortcomings of BP neural network algorithm have been analyzed. In order to overcome the slow convergence of BP network and the learning process easier to achieve the local minimum value, this paper adopts the use of the step factor, increase the momentum term and adaptive learning rate to resolve this problem.At last, after analyzing common hazard and main factors endanger the safety of coal mine production process, coal mine safety assessment model was established. In order to overcome the shortcomings, traditional BP algorithm has been improved. BP network model of assessment indexes (death rate per million tons) has been founded by neural network toolbox of MATLAB7. The paper finished the application research by using the actual data of death rate per million tons of Chinese coal industry, and the final evaluation results are basically consistent with the actual data.