| Sealed electronic components are basic components in aerospace systems,and their internal excesses affect the reliability and stability of the overall system.The particle collision noise detection method is currently the mainstream redundant object detection technology.However,it has shortcomings in identifying special interference signals and component signals.To solve above problems,this paper uses the Sakoe-Chiba constraint-based dynamic warping algorithm to optimize the template and threshold in the special interference signal training set,and designs LB_Hust,LB_Keogh,Sakoe-Chiba-DTW hierarchical filtering algorithms to identify and eliminate special interference signals accurately and efficiently;to solve the problem about local information of missing signal when Fourier transform decomposes component signals and redundant signals,wavelet packet transform is used to extract energy features,and PCA is used to reduce the dimensionality of the extracted energy features.LDA is used to make up for the lack of category information of PCA in the dimensionality reduction process;by using classification of the Boosting algorithm as a new feature added to the original sample,this paper designs a serial fusion method based on the classification result and uses Cat Boost algorithm and XGBoost algorithm.The Logistic algorithm is used as the basic classifier to build the fusion model.From the experimental results,it can be seen that the fusion model is stable at 98% in the five sets of redundant signal and component signal data samples.This paper realized effective elimination of special interference signals and accurate identification of component signals,laying the foundation for the follow-up redundant object positioning and redundant object quality evaluation technologies. |