Sensorless Position Detection Framework for a Multi-state Switched Reluctance Actuator of a Textile Machine

Sensorless needle position detection for the actuator of a textile machine is challenging because of its multi current levels used and the transient state when the current changes. A local regression method can be used to represent the variation of inductance via the measurement of the current through the actuator while the keys pass the selection area of the actuator. By calculating its correlation with a mother function, the position of the needle can be obtained. A Kalmen filter is added at the end to compensate discretisation errors. The experimental results with different working speed have validated the proposed sensorless detection method.