论文部分内容阅读
将小波包分析与Bayes判别分析法相结合的方法应用于矿用通风机故障诊断问题中。利用小波包分析技术提取了矿用通风机不同工作状态的特征向量,选用此特征向量作为Bayes判别分析模型的判别因子,以矿用通风机故障实测模拟数据作为学习样本进行训练,通过分析计算,建立了相应线性判别函数,并利用回代估计方法进行检验。研究结果表明:这种新模型判别能力强,交叉确认估计的误判率为0,不需要优化网络结构,是解决矿用通风机故障诊断的一种有效方法。
The combination of wavelet packet analysis and Bayes discriminant analysis method is applied to mine ventilator fault diagnosis. The wavelet packet analysis technology was used to extract the eigenvectors of mine ventilators in different working conditions. This eigenvector was selected as the discriminant of Bayesian discriminant analysis model. The simulation data of mine ventilator fault was used as training samples. Through the analysis and calculation, The corresponding linear discriminant function is established and tested by the back-estimation method. The results show that the new model has a good ability of discriminating and the false positive rate of cross-validation estimation is zero, which does not need to optimize the network structure. It is an effective method to solve the fault diagnosis of mining ventilator.