论文部分内容阅读
针对工业过程数据的非平稳性、含噪声以及随机性等特点,提出一种改进多尺度主元分析方法用于过程故障监测。首先利用小波阈值去噪的方法,消除原始过程数据中的大部分高频随机噪声,使得数据不受噪声的影响,然后利用小波分解将去噪后的数据分解成逼近系数和细节系数,分别在各个尺度上建立主元分析模型,对各个尺度小波系数消噪并重构得到综合尺度的故障监测模型。将该算法应用于田纳西伊士曼(Tennessee Eastman)过程中进行验证,仿真结果表明,与传统PCA以及MSPCA方法相比,改进的算法减少了误报率和漏报率,提高了过程监测的准确性。
Aiming at the characteristics of non-stationary, noise-containing and randomness of industrial process data, an improved multi-scale PCA method is proposed for process fault monitoring. Firstly, the wavelet denoising method is used to eliminate most of the high-frequency random noise in the original process data, so that the data is not affected by the noise. Then, the wavelet decomposition is used to decompose the denoised data into approximation coefficient and detail coefficient. The principal component analysis model is established on each scale, denoising the wavelet coefficients of each scale and reconstructing the fault monitoring model of comprehensive scale. The proposed algorithm is validated in Tennessee Eastman process. The simulation results show that the proposed algorithm reduces the false alarm rate and false alarm rate, and improves the accuracy of process monitoring compared with the traditional PCA and MSPCA methods Sex.