A multi-source information fusion method for multi-level performance index evaluation

Maintaining equipment performance at specified levels is crucial for optimizing operational effectiveness, necessitating mult-level performance index evaluations. Currently, performance evaluations often confront challenges related to data insufficiency, necessitating the comprehensive utilization of multi-source data and consequently leading to issues in multi-source information fusion for performance assessment. This paper addresses these challenges by analyzing perf ormance index systems and proposing a multi-source information fusion framework, which encompasses four stages: data collection, quantification, assessment, and fusion. Utilizing a cognitive cross-entropy approach, the paper determines data value weights, setting forth multi-level evaluation criteria and a quantitative reference for the long-term performance of stored equipment. Additionally, it introduces the minimum cross-entropy solution method and the statistical characteristic parameter method for scenarios characterized by sufficient and insufficient data, respectively, based on these data value weights. To demonstrate and validate the feasibility of this approach, a case study involving bolt torque reduction is presented.