Development of an Online Fault-Diagnostic-System Based on STM32 for Actuators

In the process of fault diagnosis for complicated mechatronic equipment such as actuators, the fault diagnosis method is usually calculated on the equipment side. When computing devices are not easy to be installed at the object end, the embedded device can be used for real-time diagnosis, so as to meet the requirement of no network, low bandwidth, low power consumption, and low delay, at the same time improve the availability of the fault diagnosis device. While ensuring the recognition rate and the diagnosis effect, the security, and portability of the equipment are improved. In this paper, an online fault diagnosis system for mechatronic actuators based on the STM32 embedded platform is developed. This system uses a backpropagation neural network and a support vector machine fusion model to diagnose the actuator faults and reduces the memory usage by compressing the model, discretizing the attributes, and reducing the number of input features. Real-time data acquisition, feature extraction, fusion fault diagnosis, and data storage are implemented on this embedded system.