Hidden Behavior Prediction of Complex Systems Based on Hybrid Information

Area of Science:

• Engineering Systems Analysis
• Behavioral Prediction Modeling
• Uncertainty Quantification

Background:

• Predicting hidden behaviors in complex engineering systems is challenging.
• Existing methods struggle to integrate hybrid, uncertain information (qualitative and quantitative).
• Belief Rule Base (BRB) models are effective for observable behaviors but not directly for hidden ones.

Purpose of the Study:

• To propose a novel BRB-based model for predicting hidden behaviors in engineering systems.
• To address limitations in existing methods regarding hybrid information utilization.
• To enhance the accuracy and applicability of forecasting models for unobservable system dynamics.

Main Methods:

• Development of a new BRB-based model tailored for hidden behavior prediction.
• Integration of a parameter estimation algorithm based on the maximum likelihood algorithm.
• Training the forecasting model parameters using hybrid information with uncertainties.

Main Results:

• The proposed BRB model effectively utilizes hybrid information with uncertainties.
• The integrated parameter estimation algorithm refines model parameters for improved accuracy.
• A case study demonstrates the model's capability in enhancing forecasting precision.

Conclusions:

• The novel BRB-based model, combined with parameter estimation, offers an effective approach for predicting hidden behaviors.
• This integrated method improves forecasting precision by leveraging uncertain, hybrid data.
• The model shows significant potential for applications in complex engineering systems analysis.