X-ray high voltage power supply fault prediction method based on wavelet decomposition and LSTM

Abstract

The invention belongs to a test device for electrical performance, relates to the technical field of electrical fault detection devices, and discloses a X-ray high voltage power supply fault prediction method based on wavelet decomposition and LSTM. The state change data {y1, y2,. . . , yt-1, yt} of an X-ray high voltage power supply is used as an initial sequence, respectively, the wavelet decomposition is conducted on the sequence before the first t-1 moments, and the subsequences {Dj, j = 1, 2,. . . , n} of the wavelet decomposition at different scales are obtained; the data of the t-1 moments of the n subsequences is used for establishing the LSTM model for training and prediction, and the prediction result of each subsequence at t moment is obtained; linearly superimposing is conducted on the prediction results of each subsequence at t moment to obtain a predicted value at t moment of a power state sequence; the relative error between the prediction result and a true value is calculated, and the prediction result is verified and evaluated. The method utilizes the advantages of the LSTM model by the wavelet decomposition, improves the prediction precision, has a high generalization ability, and has a great social value and practical significance.

Description

technical field [0001] The invention belongs to the technical field of electrical performance testing device and electrical fault detection device, and in particular relates to an X-ray high-voltage power supply fault prediction method based on wavelet decomposition and LSTM. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: with the development of the information age, industrial development is becoming more and more modernized, and the performance of various types of large and complex equipment that plays a major role in various fields is constantly improving and the complexity of system composition The continuous increase makes the safety reliability, maintainability, fault prediction and diagnosis of large and complex equipment more and more important in various fields such as aviation, aerospace, communication and industrial applications. Therefore, based on massive data, the prediction of equipment failure ...

AI Technical Summary

Problems solved by technology

[0003] To sum up, the problems existing in the existing technology are: the long-term prediction accuracy of the traditional fault prediction technology is very poor; the expert system requires a lot of practice and accumulation, and the development cycle is very long

Images