kNN (k-nearest neighbor) fault detection method for online upgrading master sample model

Abstract

The invention discloses a kNN (k-nearest neighbor) fault detection method for an online upgrading master sample model and relates to a fault detection method for a batch process. The kNN fault detection method for the online upgrading master sample model comprises the following steps of performing modeling and fault detection by utilizing the kNN fault detection method by taking data of a master sample space as a training set of modeling data, finding out k nearest neighbors of each sample from the master sample space, calculating the quadratic sum of distances of the k nearest neighbors on each sample, permuting the quadratic sums of the k nearest neighbors of all samples in sequence to determine a threshold value of fault detection; finding out k nearest neighbors of a sample x from the master sample space for the new to-be-detected sample x, calculating the quadratic sum of the distances of the k nearest neighbors of the sample x and comparing the quadratic sum of the distances of the k nearest neighbors of the sample x with the threshold value, judging that the sample x is normal if the quadratic sum of the distances of the k nearest neighbors of the sample x is smaller than the threshold value, otherwise, judging that the sample x is faulted. By the selection of the master sample model, data structures of all working conditions are optimized, and the accuracy of a fault detection model is improved.

Description

technical field [0001] The invention relates to a batch process fault detection method, in particular to a kNN fault detection method for online upgrading of a main sample model. Background technique [0002] Batch process is an important chemical production process, which is widely used in the production of high-quality, high value-added products industries, such as: biopharmaceuticals, semiconductor manufacturing, agricultural chemistry, etc. Therefore, the detection and fault diagnosis of the batch process has always been a research hotspot at home and abroad. In batch process fault detection, the collected data often presents the characteristics of non-Gaussian, nonlinear, and multi-working conditions, which puts forward high requirements on the performance of fault detection methods. [0003] With the increasingly extensive and in-depth application of computers in industrial control, the data encountered in industrial process fault detection will increase exponenti...

AI Technical Summary

Problems solved by technology

[0003] With the increasingly extensive and in-depth application of computers in industrial control, the data encountered in industrial process fault detection will increase exponentially, and the influence of noise information contained in it will also increase continuously. Using all known data modeling to carry out Fault detection not only greatly increases the burden of data storage, but also cannot get the most accurate results

The principal component analysis method realizes the dimension reduction processing of the original data, but how to choose the most representative training set for modeling on a large amount of training data, there is no effective method to reduce the modeling space.

Images