Data-driven complex system mechanism automatic learning method, system and device

Abstract

The invention belongs to the field of big data and machine learning, and specifically relates to a data-driven automatic learning method, system and equipment for complex system mechanisms, aiming to solve the problem that the existing system modeling technology is not capable of predicting behavior trends from field observation data and reconstructing The mechanism model does not match the physical observation data, and the robustness is poor. The invention includes: acquiring historical multi-modal data and real-time multi-modal data, constructing a time-series long-range associated hypergraph model through spatial cyclic memory coding, performing normalized combination of hypergraph models through neural differential equation model knots, and performing continuous game network The automatic iterative search of the structure obtains the continuous dynamic model of the system mechanism, and performs the biological evolution simulation to obtain the causal model, and then recalculates the correlation weight to obtain the active early warning system. The invention realizes the description and characterization of the special properties of the nonlinearity, emergence, balance step, adaptability and feedback loop of the complex system, and improves the prediction accuracy of the model.

Description

technical field [0001] The invention belongs to the field of big data and machine learning, and in particular relates to a data-driven automatic learning method, system and equipment for complex system mechanisms. Background technique [0002] With the rapid development of information technology, human production activities increasingly rely on various complex systems, such as smart manufacturing, the Internet, smart cities, smart medical care, energy Internet, smart transportation and ecosystems, etc. These complex systems are a whole formed by several units or subsystems through interrelationships. According to human wisdom, these systems are endowed with symbiosis and symbiosis. These complex systems continue to accumulate a large amount of dynamic multi-modal data, which contains valuable knowledge such as the working mechanism behavior, state and regulation optimization of the system from the micro to the macro. Machine learning plays an important role in the mechanism...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the above-mentioned problems in the prior art, that is, the existing system modeling technology is difficult to predict the behavior trend from the field observation data, and the reconstruction mechanism model does not match the physical observation data, and the risk robustness is poor. The present invention provides a A data-driven complex system mechanism automatic learning method, the method comprising:

Images