Multifeature fusion-based SSA-SVM for precision screw pump fault diagnosis

With the advancement of new energy vehicles, flat-wire motors are widely used for their superior performance. As a critical component in the stator drip painting process of flat-wire motors, the operational status of precision screw pumps directly impacts the production efficiency and product quality of motor stators. Traditional manual fault diagnosis methods suffer from hysteresis and low accuracy, highlighting the need for intelligent solutions. This study proposes an SSA-SVM fault diagnosis model using two triaxial vibration sensors to collect data. Nineteen time-frequency domain features are extracted, reduced via the mRMR algorithm, and fused as inputs. The sparrow search algorithm optimizes the SVM, achieving a 94.5% accuracy, outperforming traditional models like RF, KNN, DT, and SVM, thus enhancing screw pump fault diagnosis accuracy.