基于多模态融合的电能质量扰动智能识别方法
By employing a multimodal fusion method that combines adaptive morphological filtering, local gradient enhancement, and dynamic time warping with collaborative attention interaction and composite loss function, the problem of insufficient adaptation and lack of modal correlation modeling in traditional power quality disturbance processing is solved, achieving high-precision and robust identification of power quality disturbances.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHANGCHUN INST OF TECH
- Filing Date
- 2026-05-12
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional power quality disturbance processing methods suffer from problems such as fixed parameters, insufficient adaptability, lack of modal correlation modeling, and poor class imbalance in the processes of denoising, enhancement, multimodal fusion, and classification loss. These problems lead to loss of disturbance details, artifacts, feature misalignment, and classification bias, resulting in insufficient overall recognition accuracy and robustness.
A multimodal fusion-based intelligent identification method for power quality disturbances is adopted. Through adaptive morphological filtering, contrast enhancement guided by local gradient, dynamic time warping, and collaborative attention interaction fusion, combined with physical prior features and composite loss function, the method can accurately identify power quality disturbances.
It effectively suppresses noise, preserves perturbation details, improves the clarity of time-frequency images, corrects modal time shifts, and enhances the accuracy and robustness of composite perturbation recognition, solving the problems of insufficient adaptability and lack of modal correlation modeling in traditional methods.
Smart Images
Figure CN122196931B_ABST