改进粒子群与小波神经网络在冲击地压预测中的应用研究.docVIP

摘 要 冲击地压是当前世界范围内煤矿及岩石工程领域中遇到的最具破坏性的自然灾害之一，已经成为国际采矿工程和岩石力学界急需研究解决的科学难题。冲击地压不仅影响因素众多，而且发生机理相当复杂，到目前为止尚未形成公认的理论，在其预测、预防技术方面还有较大的改进空间。因此，对煤矿冲击地压的影响因素及预测方法进行系统的研究具有重要的理论意义和实际意义。 冲击地压是一种非线性的、复杂的动力学现象，是由煤矿开采等地下工程具有快收敛速度高收敛精度强鲁棒性特点Abstract Rockburst is a worldwide serious natural disaster encountered in coal mining and rock engineering. It has been a scientific difficulty which should be solved urgently in the field of mining engineering and rock mechanics. The mechanism of rock burst is extremely complex and there is no recognized theoretical until now. Prediction Method of rock burst has greatly improved space, too. Therefore, it has important practical significance to research the influence of rock burst and its prediction method. Rockburst is a complex nonlinear dynamic phenomena, is a kind of induced geological disasters by human activities such as coal mining. In order to solve this problem, this paper based on rockburst disaster in tunnel excavation as the research object, puts forward with SDPSO and wavelet neural network into the method of combining the impact earth pressure on the prediction research. First of all paper analyzes the basic theory of wavelet and the classification of network and the structure characteristics, first proposed by wavelet neural network prediction of rockburst, establish the forecast model and simulated by the coal mine monitoring historical data, in view of the neural network low convergence precision, easy-to-precocious and easy to fall into the local extremum shortcomings, using particle swarm optimization algorithm of wavelet neural network to optimize the structure and parameters. So the particle swarm optimization algorithm is introduced into the wavelet neural network for parameter optimization, but as a result of particle swarm algorithm in iterative later period, easy to fall into local optimal solution, this limits the performance of the algorithm, and also makes the optimization process, more difficult to find global opt