Abstract:
Small low-frequency data is difficult to use for permanent magnet synchronous machine drive inverter fault diagnosis without additional hardware and affecting accuracy. This problem often wastes system memory, computes unnecessary data, increases hardware requirements and realization difficulty, and reduces practicability.
A novel multi-switches fault diagnosis algorithm is proposed to improve this problem, make fault diagnosis design, debugging, and implementation easier and cheaper, and improve algorithm practicability.
First, a second low-frequency processing method extracts small low-frequency data from controller feedback signals. These low-frequency data contain key switch state features. Second, the single extremum normalization method, which uses state data symmetry, normalizes these small low-frequency data. These processed data preserve inverter fault state asymmetry. Third, the distortion part and envelope change of processed small low-frequency data extract the main fault components and features. Fourth, echo state network and extracted features implement intelligent classification.
Hidden layer networks are simpler than traditional neural networks and train faster. Thus, network debugging is easy. The proposed algorithm is simple and inexpensive for multi-switches fault diagnosis. The experiment proves its efficacy.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.