Rotary machinery digital twin modeling method based on mechanism-data heterogeneous information fusion

Abstract

The invention belongs to the technical field of engineering machinery, and discloses a rotary machinery digital twin modeling method based on mechanism-data heterogeneous information fusion. The method comprises the following steps: establishing a geometric model based on a physical entity, and establishing a dynamic white box model under a damage coupling mechanism of a rotary machinery mechanism; collecting multi-source dynamic data in real time in a physical space; projecting discrete data collected in a physical space to a historical evolution time dimension by considering relevance, and establishing a black box model based on big data statistics and a network system; and carrying out heterogeneous fusion on the white box model and the black box model, establishing association mapping characteristics from a physical space to an information space, and forming a complete digital twin grey box model. According to the method, the mapping relation from the physical space data to information space life prediction of the rotating machinery is established from the two aspects of mechanism and data, and the method has practical significance in guaranteeing reliable and efficient operation of engineering machinery equipment and preventing major accidents.

Description

technical field [0001] The invention belongs to the technical field of construction machinery, and in particular relates to a digital twin modeling method of a rotating machinery based on fusion of mechanism-data heterogeneous information. Background technique [0002] The transmission system of construction machinery has been in service for a long time under the complex operating environment of heavy load, large impact, and variable speed. Various loads form non-stationary, non-periodic, and nonlinear excitations on rotating parts, and surface damage failures, deformation failures and Fracture failure and other problems will lead to abnormal temperature, power loss, performance degradation, and even reduce the reliability and safety of the entire transmission chain, causing major safety accidents and causing huge economic losses. Therefore, it has important research significance for the model simulation of the mechanical transmission system under complex working conditions,...

AI Technical Summary

Problems solved by technology

[0003] Taking the hydraulic torque converter in rotating machinery as an example, the hydraulic torque converter is a cross-energy domain nonlinear strong coupling system. During the transmission process, in addition to the fluid pressure load generated by the internal circulation and eddy current, the hydraulic torque converter The impeller is also affected by the joint action of external load and centrifugal load, and there are complex working conditions such as eccentric load, variable load and poor lubrication, which make the internal flow field, temperature field and deformation field between the impellers interact; in addition, due to damage Existence, changing one of the characteristic parameters will cause some or all other parameters to change, and will affect the frequency spectrum structure, amplitude, and stable characteristics of the vibration signal, while the traditional finite element model has low solution accuracy, and the liquid Torque converters are highly dynamic and uncertain, and it is extremely difficult to accurately describe the vibration mechanism and operating mechanism of hydraulic torque converters

[0004] On the one hand, the mechanism model of the hydraulic torque converter does not fully estimate the actual working conditions. In view of the complex working conditions of construction machinery, using the "white box model" for life prediction is not only time-consuming, but also the evaluation accuracy is not high

On the other hand, due to the large discreteness of the degradation degree and failure time of the hydraulic torque converter, and the different damage forms and the scarcity of test data, the value density of the common damage characteristics calculated by using large sample data is low and the data quality varies. Qi, and it is difficult to build a large enough "black box model" for training

Images