基于异步迭代算法的冲击地压预测

Rockburst forecast based on asynchronous iterative algorithm

采用三层BP神经网络方法对冲击地压建立了数学模型。网络的训练算法采用具有松弛因子的动态异步迭代法,该算法在克服网络的麻痹现象及局部极小问题上都优于目前常用的训练方法,因而,采用此算法对网络进行了训练及震级的预报。基于新汶矿务局华丰煤矿1999～2000年的冲击地压现场监测数据,把冲击地压的能量、产生波的幅值、频次做为输入数据,相应期间的最大震级为输出数据,组成神经网络的训练样本及测试样本,对原始数据进行了数学预处理,网络结构采用了输入层3个结点,中间层7个结点,输出层1个结点的前向神经网络;网络最终的训练误差为0.06,预测结果的相对误差率平均为 9.2 %,预测效果比较理想。

A rockburst mathematical model is established by three layers neural network; and a new algorithm which is called asynchronous iterative algorithm with relaxation factor is trained the model. The algorithm has a better speed and avoids local minima and paralysis in some degree. As a results, maximum magnitude of rockburst is forecasted. The paper adopts 19992000抯 field rockburst data of Huafeng Coal Mine of Xinwen Mining Company and turns rockburst energy, wave amplitude and frequency into neural network input, turns related maximum magnitude of rockburst with above data into neural network output. So, we set up three layers forward neural network with three input nodes, seven hiden layers nodes and one output node. The paper selects 35 groups from 40 groups in 19992000 year and turns it into training groups, the remains is regarded as testing groups. The training program is designed by Matlab 6.1. After 50 000 iterations, the network error is 0.06 and the value is under the network抯 error precision. So, the network is convergence. At last, average error between the network forecast value and testing value is 9.2﹪,the result verifies that the method is successsful.