A Geometric Method for Health Prediction of Rotating Machinery Based on Single Branch Truncated Data

Abstract

The invention discloses a geometric method for forecasting healthy state of rotary machinery based on single truncation data. The geometric method comprises the following steps: monitoring vibration of the rotary machinery in real time to obtain multiple truncation data points; carrying out time continuity-based dimension reduction processing on multiple truncation data point high-dimension features of the multiple truncation data points to obtain a truncation data healthy streamline space with the time distribution-based features of the multiple truncation data points; carrying out line connection processing on the features of the multiple truncation data points according to a time relation between the features of the multiple truncation data points in the truncation data healthy streamline space to form healthy state evolution tracks of the truncation data based on the time; and fitting the healthy state evolution tracks of the truncation data to obtain a life value of the truncation data of the vibration of the rotary machinery. The geometric method solves the technical problem of damage to a 'neighborhood structure' and an 'overall structure', so that the healthy state of the rotary machinery can be better forecast by using the truncation data.

Description

technical field [0001] The invention relates to the technical field of prediction, in particular to a geometric method for predicting the health state of rotating machinery based on single truncated data. Background technique [0002] With the progress and development of society, people not only need to know the current state of the monitored object, but also need to know the future state of the monitored object, and then carry out follow-up work to effectively control risks. However, the essence of prediction is to seek the dynamic evolution law in the process of the degradation of things. In recent years, when a large amount of failure data is available, many data-driven methods have achieved better prediction results, such as: artificial intelligence networks, hidden Markov models. In addition, some methods try to solve the prediction problem under the conditions of sparse data or truncated data. However, these research works are often based on a relatively large number...

AI Technical Summary

Problems solved by technology

[0013] 1) 'Neighborhood structure' destruction: The addition of the new data point set has fundamentally changed the "neighborhood" constructed by the original truncated data set. As a result, the data points that were in the "neighborhood" relationship in the original truncated data may no longer maintain the original "neighborhood" 'relationship, which will inevitably lead to the redistribution of local geometric relations, and then change the structure of the original truncated manifold;

Images