One-dimensional complex hysteresis relation construction and structure simulation method based on deep learning

Abstract

The invention provides a one-dimensional complex hysteresis relation construction and structure simulation method based on deep learning, and relates to the technical field of civil structure engineering and computers. The method comprises the following steps: firstly, selecting a target object with a one-dimensional stress characteristic and establishing a corresponding physical model; selecting an external effect, and carrying out simulation by utilizing a physical model to form a generalized displacement-generalized force data set; training a selected deep neural network by using the data set to obtain a data-driven one-dimensional hysteretic relationship model; and finally, embedding the data-driven one-dimensional hysteretic relationship model into physically-driven classical simulation software to realize instant data fusion and interaction between the data-driven one-dimensional hysteretic relationship model and the physically-driven classical simulation software so as to form a data model and physical model collaborative structure simulation method, and carrying out structure response simulation according to the data model and physical model collaborative structure simulation method. According to the method, the one-dimensional complex hysteresis relationship can be accurately constructed, and high-precision and high-efficiency structure simulation can be carried out, so that the efficiency of a structure simulation task is remarkably improved under the condition that the precision is ensured.

Description

technical field [0001] The invention relates to the fields of civil structural engineering and computer technology, in particular to a deep learning-based one-dimensional complex hysteresis relationship construction and structure simulation method. Background technique [0002] The hysteretic characteristic simulation of engineering structures and other target objects under external effects is an important link and key technology for engineering practice and research, and its essence is to construct the mapping relationship between external effects and target object responses. Among them, the one-dimensional hysteresis relationship is the most basic hysteresis relationship, which widely exists in the test and engineering design of supports, ordinary beams and columns, and can be further extended to many important application scenarios such as layer models. Traditionally, based on physical or empirical methods, through careful mathematical and mechanical analysis, research, a...

AI Technical Summary

Problems solved by technology

However, the above methods have the following disadvantages: 1) Due to the complexity of modeling and analysis, large-scale modeling and simulations are carried out manually starting from the microscopic scale, which is too heavy in engineering practice and lacks feasibility; 2 ) Explicitly condensing key features is also highly dependent on professional knowledge and skills, and often needs to introduce more simplifications and assumptions; 3) When dealing with complex and abstract cross-scale problems, there is a lack of general automation methods, and it is difficult to improve accuracy; 4 ) Complex physical models are less efficient in carrying out simulations, and are limited to the basic structure of the model, making it difficult to further improve

Images