Research On Fault Diagnosis Of Manipulator Based On Evidence Theory

The mechanical arm,characterized by complex mechanical structure and coupling properties,is prone to failures under the influence of harsh environments.Therefore,conducting research on fault diagnosis of mechanical arms can enhance their safety and reliability,avoiding losses to personnel and property.Addressing the significant incompleteness resulting from the use of single-source information in current mechanical arm fault diagnosis,this paper investigates the method of integrating multiple information sources based on the theory of information fusion.Furthermore,a fault diagnosis model for mechanical arms based on the improved evidence theory is established,enabling fast and reliable identification of common faults such as gear wear in the context of apple picking scenarios.The main research contents of this paper are as follows:(1)Research on conflict evidence fusion method based on improved evidence theory.To address the issue of poor performance of Dempster’s combination rule in fusing highly conflicting evidence,a feature evidence generation method based on weighted complex basic probability assignment function is proposed.This method takes into account the similarity and reliability of evidence and converts the original evidence into weighted complex basic probability assignment function using comprehensive weights,thereby avoiding the problem of completely discarding the remaining belief during evidence fusion.Simulation results demonstrate that the proposed method effectively identifies erroneous evidence,handles evidence conflicts,and achieves better convergence performance and diagnostic accuracy in the fusion results.(2)Research on fault classification method based on multi-source information fusion.In order to efficiently utilize diverse and complex equipment fault data for equipment fault classification,this paper proposes a fault classification method based on multi-source information fusion.This method preprocesses the original fault data through steps such as data denoising,feature extraction,feature selection,and data dimensionality reduction.Then,a classification model is constructed based on the generation and optimization of the feature reference matrix,and an improved evidence theory is used for multi-source information fusion in classification decision-making.Experimental verification is conducted using the UCI dataset and data samples collected from the rolling bearing fault simulation test rig at Case Western Reserve University,demonstrating the effectiveness of the proposed method.The experimental results show that the proposed evidence fusion method can effectively handle data from different sources,reasonably utilize the classification results generated by different data,thereby improving the overall classification performance and robustness.(3)Manipulator fault diagnosis based on multi-source information fusion.In order to achieve fault diagnosis of a mechanical arm,this study transforms the fault feature information of the mechanical arm into classification evidence based on an optimal fault classification model.By using an improved evidence theory-based multi-source information fusion method,the classification evidence is fused to infer the corresponding fault category.To verify the effectiveness of this method,a fault test platform is constructed,and the corresponding fault data is collected.The results demonstrate that when using the fusion diagnosis method,even with lower classification accuracy based on single-source data,the fusion diagnosis results exhibit higher diagnostic accuracy and fewer false positives.In terms of average accuracy,the fusion diagnosis method improves the accuracy by 11.33%compared to the average results based on single-source data.Therefore,the fusion diagnosis method can effectively improve the accuracy of fault diagnosis and has better robustness,enabling it to adapt to different types of faults and data sources.