摘要
针对矿井瓦斯涌出量多因素预测问题,提出了一种基于量子粒子群优化(QPSO)算法的小波神经网络预测模型.该模型利用了小波神经网络的特征提取能力,并以QPSO算法确定小波网络的最优初始参数.仿真结果表明,优化后的小波神经网络(WNN)预测模型具有收敛速度快、拟合能力强、预测精度高以及预测结果唯一等优点.同时采用数据对比,分析了网络预测结果不稳定的原因、网络训练精度和预测精度之间的矛盾,给出了决定模型预测能力的关键因素以及评价方法.
In order to solve the multifactor mine gas emission prediction problem, a wavelet neural network prediction model is proposed based on quantum particle swarm optimization （QPSO） algorithm. The proposed model utilizes the feature extraction capability of wavelet neural network （WNN）, and applies the QPSO algorithm to determining the optimal initial parameters of the WNN. Simulation results show that the optimized WNN prediction model has the advantages of fast convergence, good fitting ability, high prediction accuracy and the ability to give the only prediction result. Additionally, by comparing the simulation data in the experiments, the reason of instability during the network prediction, the contradiction between the network training and prediction accuracy are analyzed. Then the key decision factors and evaluation methods are given for the prediction ability of the proposed model.
出处
《信息与控制》
CSCD
北大核心
2012年第6期741-746,759,共7页
Information and Control
基金
河北省教育厅科研基金项目(Z2006439)
关键词
量子粒子群优化
小波神经网络
瓦斯涌出量
优化
预测
quantum particle swarm optimization
wavelet neural network
gas emission
optimization
prediction
