Monitoring of load torque variations by using an Electrical Drive with Direct Torque Control
MetadataShow full item record
- Institutt for elkraftteknikk 
This work analyses the accuracy of the estimated mechanical speed and electromagnetic torque in a Direct Torque Controlled (DTC) induction motor drive with no speed measurement. A literature study has been performed in order to reveal new areas where the DTC technology can expand to, and this knowledge was compared with the accuracy data to determine the most feasible areas of expansion. The most promising new technique is to combine the DTC technique, with its estimation of speed, torque and power, with a model based method of estimating load state variables. If the load is observable using the DTC estimates, and the estimates are considered accurate enough for control purposes, this could potentially replace load process measurements such as mass flow or process pressure. A model based method could also increase the knowledge available about the load, leading to more efficient control methods and reduced energy consumption. A MATLAB Simulink model of the DTC drive was implemented using a DTC drive block from the SimPowerSystems library as basis. The original block speed controller relied on a speed measurement feedback, but a new speed, torque and flux estimator using only current and voltage measurements was implemented, and the speed measurement feedback replaced with a speed estimate feedback. Lab measurements from a preliminary project were analyzed and compared to the simulation setup results. A simplified transfer function from measured speed to DTC estimated speed is suggested. It is equal to the speed estimate low pass filter transfer function. Several analyses show this to be a good approximation. The unfiltered speed estimation is shown to have a very high accuracy of better than 0.3% error for applied load torque frequencies of up to 10Hz, when the estimated data is used as input to frequency analyses. But the unfiltered speed signal also has a very high ripple, explaining the need for the low pass speed filter. The filtered speed estimate has some measurement errors, due to the damping in the speed filter. The estimated electromagnetic torque is very accurate, which is one of the strengths of DTC. The different parameters in the DTC setup, and their impact on the speed and torque estimates, were analyzed. The speed filter setting and the DTC speed sampling time was found to be the most important settings considering the accuracy of the speed estimation. Some improvements in the speed estimate calculation in the simulation model are suggested to improve the speed estimate. Improvements are also suggested for the ACS800 DTC drive in the lab setup. Where increasing the memory available for sampling would significantly improve the frequency analyses.