Methods for fusing uncertain results obtained from different models in accident reconstruction.

Area of Science:

• Engineering
• Data Science
• Risk Analysis

Background:

• Existing methods for fusing reconstructed results often rely on subjective expert opinions.
• The fusion of probabilistic and mixed uncertainty results remains an underexplored area.
• A need exists for objective and automated methods to combine diverse data sources.

Purpose of the Study:

• To propose and evaluate two novel methods for fusing uncertain reconstructed results.
• To address the challenge of fusing probabilistic and mixed uncertainty data.
• To provide objective and statistically sound fusion techniques.

Main Methods:

• The Monte Carlo Method (FMCM) involves generating sample points from individual result distributions and performing statistical analysis to derive the fused cumulative distribution function.
• The Sub-Interval Technique (FSIT) establishes a fusion interval set using lower and upper bounds and a defined sub-interval length, followed by weighted matrix calculation for statistical analysis.

Main Results:

• Both FMCM and FSIT successfully fuse uncertain results, producing a cumulative distribution function for the final fused outcome.
• The methods were validated using three real-world accident case studies.
• Demonstrated practical applicability and effectiveness in complex scenarios.

Conclusions:

• FMCM and FSIT offer viable, objective alternatives to expert-based fusion methods.
• These techniques effectively handle probabilistic and mixed uncertainty in data fusion.
• The proposed methods show strong performance in practical applications.