Simulation Credibility Evaluation Based on Multi-source Data Fusion

Yuchen Zhou,Ke Fang,Ping Ma,Ming Yang
DOI: https://doi.org/10.1007/978-981-13-2853-4_2
2018-01-01
Abstract:Real-world system experiment data, similar system running data, empirical data or domain knowledge of SME (subject matter expert) can serve as observed data in credibility evaluation. It is of great significance to study how to incorporate multi-source observed data to evaluate the validity of the model. Generally, data fusion methods are categorized into original data fusion, feature level fusion, and decision level fusion. In this paper, we firstly discuss the hierarchy of multiple source data fusion in credibility evaluation. Then, a Bayesian feature fusion method and a MADM-based (multiple attribute decision making) decision fusion approach are proposed for credibility evaluation. The proposed methods are available under different data scenarios. Furthermore, two case studies are provided to examine the effectiveness of credibility evaluation methods with data fusion.
What problem does this paper attempt to address?