A dynamic state prediction method and system based on markov transition matrix
By constructing a state reachability determination matrix based on physical constraints and a baseline transition matrix estimated by Bayesian Dirichlet, and combining the state transition causal graph and conditional Markov transition matrix, the uncertainty and real-time problems of state prediction in the prior art are solved, and efficient dynamic state prediction and control coupling are realized, thereby improving the energy efficiency and stability of building electromechanical systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANYANG NORMAL UNIV
- Filing Date
- 2026-05-07
- Publication Date
- 2026-07-17
AI Technical Summary
Existing state prediction methods for building electromechanical systems based on Markov transition matrices have shortcomings in terms of physical consistency of state space, reliability of transition probability estimation, causal interpretability of correction factors, coupling between prediction and control, real-time online updates, and robustness to uncertain decisions. They are difficult to meet the real-time requirements of high-frequency prediction scenarios and the vulnerability of control decisions in high-uncertainty scenarios.
By establishing a state reachability determination matrix based on physical operation constraint rules, constructing a benchmark transition matrix estimated by Bayesian Dirichlet, and combining it with the state transition causal graph and conditional Markov transition matrix, dynamic correction and online adaptive updating are achieved. Combined with equipment control strategy parameters, state prediction and control are coupled, concept drift is detected, and incremental updates are performed.
It significantly improves the statistical significance and computational efficiency of the transition probability matrix, enhances the robustness of the model in the early stages of equipment commissioning and under abnormal operating conditions, improves the generalization ability and prediction accuracy of the time-varying transition matrix, achieves millisecond-level adaptive matrix evolution and continuous maintenance of long-term prediction accuracy, and enhances the system's fault tolerance and operational safety under complex operating conditions.
Smart Images
Figure CN122414482A_ABST