SIGNAL BIOPSY-BASED PARAMETER ESTIMATION AND END-TO-END DIAGNOSIS SYSTEM FOR ELECTRICAL MACHINES.

Abstract

The invention relates to a signal biopsy-based system and method that performs online parameter estimation and end-to-end diagnosis using current and / or voltage signals obtained from electrical machines and / or associated power electronics systems. Within the system, electrical data are continuously sampled and stored in circular buffer memory, and an event-dependent signal biopsy is generated, including pre-event, event / triggering and post-event data sections, upon identification of an event according to physical, spectral and / or algorithmic triggering criteria. The same signal biopsy generated is processed in parallel in the Physical Parameter Estimation Line (15) and the Artificial Intelligence-Based Diagnosis Line (17).In the physical parameter estimation line, at least one of the parameters suitable for the topology of the electric machine is estimated online, while in the AI-based diagnostic line, anomaly information and / or fault hypothesis is generated from the electrical properties obtained from the signal biopsy. The difference between the measured and / or derived magnitude and the corresponding magnitude predicted by the physical model is determined by the Model Residual Value Calculation Unit (19), and the AI-based diagnostic result, the predicted physical parameter behavior and the model residual value are evaluated together in the Physically Validated Cross-Validation Unit (18). The estimated parameters intended to be used in motor control are validated by the Parameter Security and Validation Filter (20), and the parameters found to be reliable are transferred to the Motor Control Unit (21).