Fault detection method based on particle swam optimization kernel independent-component analysis model

Abstract

The invention discloses a fault detection method based on a particle swam optimization kernel independent-component analysis model. Kernel learning skills and a particle swam optimization algorithm are combined and used in the method, a traditional independent component analysis method is expanded into a modeling method capable of directly processing non-linear process data, and a corresponding fault detection model is established on the basis. The fault detection method specifically comprise the following steps: firstly, an original training data matrix is converted into a kernel matrix through a kernel function, and centralization processing is carried out; secondly, non-linear independent components are solved by utilizing iteration of the particle swam optimization algorithm and are ranked according to non-Gaussian sizes; and, finally, the non-linear fault detection model is established, and online fault detection is implemented. Compared with a traditional method, the method of the invention prevents a whitening pretreatment process so that a condition that original data information loses or is distorted cannot occur. Furthermore, the method of the invention is not limited to establish the fault detection module, but can be applied to other fields related to non-linear data signal source separation.

Description

technical field [0001] The invention relates to an industrial process fault detection method, in particular to a fault detection method based on particle swarm optimization kernel independent element analysis model. Background technique [0002] In recent years, with the rapid development and wide application of computer technology, modern industrial processes have gradually moved towards the "big data era". Industrial processes, especially process industrial processes, have abundant production data resources, which provide a solid data foundation for data-driven fault detection research. Generally speaking, the data-driven fault detection method does not require an accurate mechanism model of the process object. It only needs to analyze the sampling data of the normal operation of the process and define a region that describes the normal data fluctuation range to implement online fault detection. Multivariate statistical analysis algorithms (such as principal component ana...

AI Technical Summary

Problems solved by technology

Although this PSO-ICA algorithm overcomes the shortcomings of the Newton iterative method, it is more time-consuming than the Newton method.

Fortunately, when PSO-ICA is used for fault detection, since it only involves solving independent elements in the offline modeling stage, the timeliness of the algorithm will not limit its application to fault detection

However, the PSO-ICA algorithm is still a linear transformation algorithm, which cannot effectively mine useful information of nonlinear process data.

Although the traditional KICA algorithm can handle nonlinear data, in terms of the KICA modeling process, it actually uses the kernel principal component analysis to whiten the data, and then implements FastICA iteratively to obtain the independent elements.

If the PSO-ICA algorithm is simply used to process the whitened data after nuclear principal component analysis, on the one hand, the extracted independent elements are not directly derived from the original data, and on the other hand, the corresponding calculation time will be increased

Therefore, this nonlinear expansion method is not advisable, and it also violates the original intention that the PSO-ICA algorithm can be directly used to analyze the original data

Images