Engine residual life prediction method of multi-period rule cascade memory network

Abstract

The invention provides an engine residual life prediction method based on a multi-period rule cascade memory network, and the method combines an automatic expansion cascade long and short term memory network with a plurality of domain knowledge rules, and is used for predicting the residual life of an engine. According to the method, an automatically-expanded cascaded long-short-term memory network is used for preliminarily predicting the residual life of the engine. A plurality of domain knowledge rules in the method are used for further correcting the preliminary prediction result so as to reduce the final prediction error. The engine residual life prediction method based on the multi-period rule cascade memory network can effectively solve the problem that a traditional prediction method is not high in prediction precision, the function of accurately predicting the engine residual life is achieved, the structure of a traditional model and the prediction method are optimized, and the prediction efficiency and prediction precision of a traditional method are improved.

Description

technical field [0001] The invention belongs to the field of engines, and relates to a multi-period rule cascade memory network prediction method, which is suitable for predicting the remaining life of an engine. Background technique [0002] Currently, predictions about remaining engine life rely mainly on time series data and long short-term memory networks. Although the long-term and short-term memory network has outstanding advantages in the learning of long-term series data due to the special design of the gate structure, the upper limit of the prediction accuracy of the long-term and short-term memory network itself makes it difficult for researchers to increase the number of network layers or training times. To break through the prediction accuracy bottleneck of the remaining life of the engine. [0003] In addition, although the proposal of a series of variant models related to long-term and short-term memory networks has effectively improved the prediction accuracy...

AI Technical Summary

Problems solved by technology

Although the long-short-term memory network has outstanding advantages in the learning of long-term sequence data due to the special design of the gate structure, the upper limit of the prediction accuracy of the long-term short-term memory network itself makes it difficult for researchers to increase the number of network layers or the number of training times. To break through the bottleneck of the prediction accuracy of the remaining life of the engine

[0003] In addition, although the existing series of variant models related to the long-term short-term memory network have effectively improved the prediction accuracy of the remaining life of the engine to a certain extent, the prediction accuracy of the remaining life of the engine using these variant models of the long-term short-term memory network is The resulting results always fluctuate greatly and cannot accurately fit the real remaining life of the engine

[0004] Therefore, a method for predicting the remaining life of an engine based on a multi-period regular cascaded memory network is proposed. The method adopts an automatically expanded cascaded long-short-term memory network that relies on the cascaded structure of multiple sub-modules to break through the limitations of the long-term short-term memory network itself. Predict the bottleneck of accuracy, and at the same time reduce the volatility of engine remaining life prediction results through the formulation of multiple domain knowledge rules under different time periods, effectively solving the difficult problem that the current mainstream methods are still unable to accurately predict the remaining life of the engine

Images