Measuring Uncertainty in the Negation Evidence for Multi-Source Information Fusion.

Area of Science:

• Artificial Intelligence
• Information Theory

Background:

• Dempster-Shafer evidence theory is a key framework for reasoning under uncertainty.
• Recent advancements include modeling uncertain information using the negation of evidence (negation of basic probability assignment - BPA).
• Quantifying uncertainty within this negation framework remains an open challenge.

Purpose of the Study:

• To propose a novel method for measuring uncertainty in negation evidence.
• To enhance multi-source information fusion by incorporating this uncertainty quantification.
• To address limitations in current Dempster-Shafer evidence theory applications.

Main Methods:

• Adopting and improving Deng entropy, a belief entropy measure.
• Defining a new uncertainty measure based on the negation function of BPA.
• Developing an improved multi-source information fusion method incorporating the new uncertainty measure.

Main Results:

• A new measure effectively quantifies uncertainty in negation evidence.
• The proposed fusion method demonstrates rationality and effectiveness.
• Experimental validation on a numerical example and a fault diagnosis problem.

Conclusions:

• The developed method provides a robust way to handle uncertainty in negation evidence.
• This research advances Dempster-Shafer evidence theory for complex uncertain information fusion.
• The findings have practical implications for applications requiring reliable uncertain information reasoning.