Abstract: With the widespread adoption of variable-frequency speed regulation technology, drives are extensively used in the petrochemical industry and have become a core piece of equipment for energy-saving upgrades and intelligent transformation in this sector. However, various faults are prone to occur during actual operation, leading to economic losses and even casualties. Therefore, it is essential to conduct in-depth research on fault diagnosis methods for drives. As a leading enterprise in the drive industry, Danfoss is committed to applying artificial intelligence methods for drive fault diagnosis research. For the first time, experiments on fan fouling faults in drives were conducted, and a Support Vector Machine-based fault diagnosis model was constructed. The diagnostic results show that the model significantly improves the accuracy of diagnosing fan fouling faults.

Key words: Drive, Fault Diagnosis, Machine Learning, Data-Driven, Support Vector Machine