Method and system for long-term prediction of engine gas path parameters based on similarity

Abstract

The present invention relates to a long-term prediction method and system of engine gas path parameters based on similarity, wherein the method includes: obtaining the time series of point-by-point distance characteristics between the target trajectory and each historical trajectory through calculation, and using the obtained point-by-point distance The time series of features evaluates the statistical distance between the target trajectory and each historical trajectory. Using the obtained statistical distance and historical trajectory samples, for a single feature element at each prediction time point, each historical trajectory sample generates a target trajectory. The probability density estimation in the form of a hypothetical Gaussian function becomes a hypothetical Gaussian unit set; the obtained hypothetical Gaussian unit set is aggregated by the descending aggregation method to obtain the Gaussian mixture model of the target feature. Compared with the autoregressive moving average, the reverse propagating neural network and the traditional prediction method based on similarity, the present invention has higher prediction accuracy.

Description

technical field [0001] The invention relates to the technical field of performance prediction of aero-engines, in particular to a method and system for long-term prediction of engine air path parameters based on similarity. Background technique [0002] As a type of high-value, high-reliability complex equipment, the performance of aero-engines has been continuously improved with the application of new technologies. Correspondingly, its structural complexity, purchase price and maintenance price have also increased. In order to adapt to the development needs of the aviation industry in the new era, aero-engines are facing pressure from economic and safety aspects. At the same time, today's energy efficiency management and environmental protection standards also put forward higher requirements on the performance level and reliability level of the engine. In order to cope with the multiple pressures on engine maintenance brought by factors such as maintenance difficulty, main...

AI Technical Summary

Problems solved by technology

In other words, due to the inherent characteristics of the weighted average method, the trajectories generated by WLE and KDE must inevitably be within the envelope formed by the typical historical decay trajectories used as the base in the high-dimensional feature space This characteristic will inevitably cause the prediction method to have systematic errors that are difficult to eliminate, thus causing greater damage to the generalization ability of the model.

Images