The Safety Evaluation Study Of The Coal Mine Based On The Artificial Neural Network Measure

As the pillar industry of our national economy,the coal industry plays a vitally important part in the development of it. However, our country has been confronted with a not optimistic situation of the coal mine safety so far. And the casualty incidence of colliery accident comes first in the world.Thus the safety evaluation has become an essential link in the security management of colliery enterprises. The text adopts the method of artificial neural network, which features highly nonlinear, fault tolerance and self-organization, to carry out the safety evaluation.The primary research is as follows:(1)It is indicated that traditional safety evaluation has the flaw of linearity, locality and certainty by comparison and analysis while the neural network evaluation has the characteristics of self-organization,self-learning and highly nonlinear approximation.And the latter is convenient for associating, synthesizing and extending.Thus these advantages make the neural network to have the practicability to be used for coal safety evaluation.(2)Basing on fully understanding the features of coal production system, we can set up the colliery safety evaluation index system with five aspects, which consists of inherent risk factor, population risk factor, equipment risk factor,management risk factor and environmental risk factor, according to the design principle of colliery safety risk index system.(3)It explores the selection on the numbers and nodes of hidden layers and establishes the network structure of back propagation (BP) neural network. It works on the platform of MATLAB7.0by using MATLAB software programming. The colliery is evaluated by momentum back propagation (MOBP), resilient back-PROPagation (RPROP) and levenberg-marquardt (L-M) separately, the result shows that they all can finish the evaluation.But when it comes to MOBP, the operation speed is slow and the steps are too many. As for RPROP, it is deficient in approaching target value.L-M has both fast operation and few steps. What is more, it is provided with the precision to which the former two could not equal.(4)It introduces radial basis function (RBF) neural network which is superior to BP neural network regarding precision approximation ability,classification recognition and the network training speed, If we use RBF neural network to evaluate the mine,the training speeding will be faster than the common back propagation neural network.Not only the time is short but also the precision is high. And it can be equal to the L-M of BP neural network and finish the evaluation commendably. Therefore, it is necessary to popularize it energetically.The research of this text suggests that artificial neural network has obvious advantages compared with the traditional safety evaluation.The effective operability and accurate result make the safety evaluation of colliery enterprises more efficient.