基于PCA-PDBNs的故障检测与自学习辨识

如何提高工业过程故障识别的准确性及其算法训练的效率一直是故障检测与辨识研究领域的重点和热点。将深度学习方法引入该领域,结合粒子群优化(PSO)算法和深度信念网络(DBNs),提出了一种基于PSO的DBNs辨识方法(即PSODBNs,PDBNs),使用该方法对复杂函数的拟合进行了数值仿真。实验结果表明,相比于基本的DBNs模型,经PSO算法对网络参数优化后的DBNs模型获得了更好的函数逼近效果,具有更高的辨识精度。为验证该方法在实际工业过程故障检测中的可行性,结合主元分析(PCA),提出了一种PCA-PDBNs模型,并将此应用于田纳西-伊斯曼(TE)过程的故障检测中,结果表明,基于PCA-PDBNs方法降低了故障检测模型的复杂度,进一步提高了对未知故障类型的辨识精度,取得了较好效果。

Fault detection and self learning identification based on PCA PDBNs

How to improve the accuracy of industrial process fault recognition and the algorithm training efficiency is always the hot spot in fault detection and identification field. In this paper, the deep learning method is introduced to this field; the particle swarm optimization (PSO) algorithm and the deep belief networks (DBNs) are combined; an identification method (PDBNs) based on PSO and DBNs is proposed. Using this method, the complex function fitting is numerically simulated. The experiment results show that compared with the basic DBNs model, the PDBNs model with the network parameters optimized achieves better function approaching performance, and has higher identification precision. To verify the effectiveness of the new algorithm in the actual industrial process fault detection, integrating the characteristics of PCA and PDBNs, a new identification model called PCA PDBNs is proposed; and was applied in the fault detection of Tennessee Eastman process. The results indicate that the method based on PCA and PDBNs decreases the complexity of the fault detection model, further improves the identification precision for unknown fault, and achieves good fault detection result.