基于声发射表征冲击地压的粒子群优化支持向量机算法

Particle swarm optimization support vector machine based on acoustic emission rock burst characterization

为有效预防冲击地压灾害发生,解决人为选取支持向量机参数泛化能力差的问题,引入粒子群算法,将声发射特征信号和煤岩抗压强度作为输入,以综合危险指数作为输出,建立煤岩冲击地压预测模型,通过粒子群算法优化支持向量机参数,预测冲击地压综合危险指数。结果表明,与网格搜索法的支持向量机、遗传算法优化过的支持向量机相比准确率达97.16%,应用粒子群算法优化过的支持向量机预测方法减小了系统误差。该算法具有泛化能力强,对煤岩冲击地压预测的准确性提供参考。

This paper is an attempt to effectively prevent the occurrence of rock burst disasters and address the poor generalization of artificially selected support vector machine parameters.This is rendered possible by introducing the particle swarm optimization algorithm to support vector machine using emission characteristic signal and compressive strength of coal as input and comprehensive hazard index as output,establishing a model behind theprediction of rock burst,optimizing the parameters of support vector machine by particle swarm optimization,and predicting the comprehensive risk index of rock burst using the model.The results show that compared with support vector machine optimized by grid search method and genetic algorithm,the proposed algorithm features the accuracy of 97.16%.and fewer system errors.The algorithm with a stronger generalization ability could provide a reference for the accuracy of rock burst prediction.