A new combination method based on Pearson coefficient and information entropy for multi-sensor data fusion

Context: When confronted with greatly contradictory evidence, the Dempster-Shafer evidence theory may exhibit certain constraints that lead to fused results which are inconsistent with common understanding. Within the existing Internet of Things landscape, there are occasions when a small number of sensors may malfunction and contradict each other.Objective: This study addresses contradictory information by processing the bodies of evidence beforehand. Additionally, an enhanced fusion technique for conflicting evidence is introduced, which employs Pearson correlation coefficient and information entropy.Methods: We propose a novel combination approach for multi-sensor data fusion based on evidence theory. Firstly, the credibility for each piece of evidence is computed through amalgamating correlation measurements with evidence distance between two pieces of evidence. Next, based on the information volume, the credibility is adjusted, resulting in the final weighting factor for the evidence. The reasonable weighted average evidence is then created using the weighting factor of each piece of evidence. Finally, the combined result is obtained by applying Dempster's combination rule, which combines the weighted average evidence N -1 times.Results: Upon comparing the fusion results, it has been observed that the performance of the proposed method surpasses that of other methods. Our method can effectively minimize the ramifications of profoundly conflicting evidence in the fusion process, resulting in more logical fusion results than other methods.Conclusions: The outcomes of numerical examples expose that the technique put forward in this manuscript can manage highly conflicting evidence, thereby yielding fusion results that are more precise and conducive to making sound decisions.