Improved sensorless operation of a CSI-based induction motor drive: Long feeder case

Various applications, like in underground mines and oil and gas industries, require remote operation of vector-controlled medium-voltage variable speed drives via a long motor feeder. The use of voltage source inverters in such cases leads to motor overvoltage and harmonic quality problems. The current source inverter (CSI) is ideally matched to these applications because of its motor-friendly voltage output. Speed sensorless operation is mandatory due to the long motor feeder. Although the model reference adaptive system (MRAS) is a powerful and proven speed estimation tool, its implementation in long motor feeder drives faces many challenges. Among them, and addressed in this paper, are inherent dc offset in its stator model, the need for actual motor voltage and current values, and oscillations in the estimated speed due to errors in the motor current measurement signals. In this paper, a sensorless CSI vector-controlled drive, suitable for long motor feeder applications, is studied. Improved speed estimation is achieved by proposing 1) a modified dc-offset eliminator for an MRAS speed estimation and 2) a compensation technique for motor current's measurement errors. Intensive experimental results, for a low-voltage scaled model, along with simulations validate the effectiveness of the proposed technique.