Fault Diagnosis and Evaluation Based on Visual Cognitive Computing

In essence, the fault diagnosis and performance of an electromechanical system are the processes of recognizing the fault mode and recognizing the performance status of the system, respectively. Therefore, cognitive ability directly affects the fault diagnosis effect and performance evaluation. However, due to the limitation and one-sidedness of cognition, people cannot be aware of the state of the system from the original signal. At present, a variety of algorithms have been developed using the existing basic theories and science, including model-based methods and data-driven approaches to recognize the fault modes and performance statuses of electromechanical systems. However, due to basic tool limitations, model establishment accuracy problems, and external environment disturbance, it is difficult to apply the existing fault diagnosis and performance evaluation methods to actual engineering applications, which leads to a series of engineering application problems. The demand for guaranteeing the safety and reliability of system operations urges people to explore a more effective fault recognition method for electromechanical products to realize reasonable and effective fault diagnosis and performance status evaluation.