A Novel DMK Fusion Model for Full-Cycle Data and Its Application in Electronic Equipment

The significant data dispersion in multi-enterprise collaborative models has hindered the efficiency of fault diagnosis in electronic equipment. To address this issue, we propose a new fusion model and a fault diagnosis framework. Specifically, design and measured data, computational mechanisms, and fault knowledge are integrated to construct a Data-Mechanism-Knowledge (DMK) fusion model, which is a full-cycle fusion system with capabilities for both qualitative fault reasoning and quantitative calculation tuning. Subsequently, we introduce a novel dual-channel Kolmogorov Arnold Networks (KAN) graph convolutional network (DC-KANGCN), which is trained using knowledge from the DMK fusion model to enhance fault diagnosis reasoning. In the experiment, typical electronic equipment such as phased array antennas and LC microwave filters were selected to validate the proposed method. The qualitative fault reasoning accuracy of DC-KANGCN improves by 3-15% compared to other methods. And the quantitative calculations and tuning are similar to the actual measured results. All functions have been validated in the provided software. The DMK fusion model holds significant potential in intelligent manufacturing. The code of DC-KANGCN is available at https://github.com/linqiangqiang/DC-KANGCN-for-DMK-model.