Abstract:Aiming at the shortcomings of existing fault recognition methods in terms of recognition time and accuracy, especially in the field of medical equipment, a medical equipment fault intelligent diagnosis system has been studied. The system adopts key technologies and methods of LabVIEW, combined with MobileNet-v2 and Single Shot Multi Frame Detector (SSD), to achieve efficient and accurate fault detection. The data acquisition module utilizes DAQmx API and VISA to achieve smooth communication with various hardware, ensuring high-precision data acquisition. The data storage and management module relies on the Database Connectivity Toolkit and File I/O Functions to achieve long-term storage and management of data. The fault detection module optimized the inverse residual structure, linear bottleneck, and depthwise separable convolution design of MobileNet-v2, and combined it with an SSD classifier to construct an SSD-MobileNet-v2 model for fault diagnosis of medical devices at different scales. Through experimental testing, the system achieved a fault recognition accuracy of over 0.97 on 120 medical equipment samples, with a maximum of 0.9843 and an average fault response time of 160.67ms. The average absolute error of the model is concentrated between 0.02 and 0.04, with the highest not exceeding 0.05. The confusion matrix analysis shows that the system has a high overall accuracy in identifying aging faults and loss faults, with 964 and 984 correctly classified samples, respectively. The research results indicate that the system has the characteristics of high accuracy, fast response, and low resource occupation in medical equipment fault detection, which verifies its effectiveness and reliability in practical applications.
