Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
Estimation of Shaft Speed and Load Inertia Applied to Induction Motor Using Kalman Filter for Unknown Inputs
Submission Author:
Daniel de Lemos Santos , BA
Co-Authors:
Daniel de Lemos Santos, Thiago Chagas, Gildson de Jesus, Vinícius Madureira, Matheus Soares
Presenter: Daniel de Lemos Santos
doi://10.26678/ABCM.COBEM2021.COB2021-0084
Abstract
The three phase induction motors became very common because of their easy adaptability to different loads and the low costs in repairing and construction. The measurement of its operation data, specially the motor speed and currents, allows to expand their uses into safer and more complex mechanical systems. However, this data extraction is commonly made by using sensors which are physically complicated to implement and more expensive as they are more accurate. Instead of a tachometer and inertial sensors, using current and voltage sensors to get motor data is a suited solution because they are cheaper, less intrusive and can operate at higher sampling periods. This work presents a methodology of three phase induction motor speed and load inertia estimation from its nominal operating point, based on a modified Kalman filter for unknown input (KFUI) by computer simulation. The FKUI uses a linear discrete state space system model and real-time measurement of some state variables and inputs to estimate the state and the unknown inputs. Thus, modeling, linearization and discretization steps were done to get a system model compatible with the KFUI, where the currents and voltage applied in the motor stator are the measured state variables and input, respectively. Rotor speed and load inertia are the estimated state variable and the unknown input, respectively. This approach allows yet the estimation of stator and rotor currents, besides the motor torque. Simulations using the original KFUI showed high noises for load inertia estimates, so a low pass butterworth filter, adjusted by genetic algorithm, was implemented into the KFUI to improve its filtering capability and the precision of the unknown input estimation. Since the motor model was obtained using the dq0 axis reference-frame, which showed high sensibility with the abc system, the simulations were done considering specific measurement and system noises. The analyses were based on two operation cases: the first situation consists of load changes on the motor shaft with constant voltage supply and the second one includes voltage disturbances in the previous load change situation. Load type was selected to be generic in both situations but in practice the system’s estimation can be adjusted by optimization to any type of target load. Results showed maximum nRMSE of 0.0006% for the speed estimate and 0.0142% for the load inertia estimate. Therefore, shaft speed and inertia estimation, using the modified KFUI, consists of a cheaper and precise alternative to sensors.
Keywords
currents estimation, filtering, Induction Motors, Optimization, sensorless, torque estimation
