Wind turbine bearing fault diagnosis method based on PCA and KNN density algorithm

Abstract

The invention provides a wind turbine bearing fault diagnosis method based on PCA and KNN density algorithm. The method comprises the following steps: obtaining vibration signals of a wind turbine bearing under different working states; carrying out pretreatment on the vibration signal data; calculating time domain and frequency domain statistical parameters of each sample, and constructing a characteristic matrix of the wind turbine bearing signals; carrying out dimensionality reduction on the multi-characteristic matrix of the wind turbine under different working states by utilizing a PCA algorithm, extracting characteristic input and serving the input as a training sample set of a fault diagnosis model; carrying out modeling on training samples through a support vector machine (SVM); carrying out optimization on parameters of the support vector machine through the KNN density classification algorithm; and displaying the final diagnosis result in a human-computer interaction interface. The method can accurately carry out classification on fault types, thereby improving wind turbine bearing fault classification precision; and the method provides guarantee for safe and reliable operation of the wind turbine, so that power grid dispatching can be optimized, and safe, stable and economical operation of the power grid is realized.

Description

technical field [0001] The invention relates to the field of bearing fault diagnosis algorithms, in particular to a method for wind turbine bearing fault diagnosis based on PCA and KNN density algorithms. Background technique [0002] The traditional KNN algorithm is a widely used fault diagnosis and classification algorithm. The traditional KNN algorithm first finds the K nearest neighbors with the closest distance to the sample to be classified, and then adopts the decision rule that the minority obeys the majority according to the fault category of the K neighbors. Determine the fault category to which the sample to be classified belongs. [0003] Traditional wind turbine bearing fault diagnosis methods are classified according to the state quantities of bearing vibration signals. The complete wind turbine bearing fault diagnosis process mainly includes: ①Signal acquisition: according to the working environment of the wind turbine bearing, select and acquire signals that...

AI Technical Summary

Problems solved by technology

[0004] In the traditional wind turbine bearing fault diagnosis technology, in ① signal acquisition, the vibration signal is not preprocessed, and the signal irrelevant to the data greatly increases the workload of the diagnosis system; in ② feature selection, the feature input is extracted The workload is heavy, which increases the time for fault diagnosis; in ③ pattern recognition, due to the uneven distribution of different fault sample data, the fault data samples are misclassified when using the traditional KNN algorithm, which leads to the failure classification accuracy Decline

Images