Teresa Orlowska‐Kowalska and Marcin Kaminski
The purpose of this paper is to obtain an estimation of not measured mechanical state variables of the drive system with elastic coupling between the driven motor and a load…
Abstract
Purpose
The purpose of this paper is to obtain an estimation of not measured mechanical state variables of the drive system with elastic coupling between the driven motor and a load machine, using neural networks (NN) of different type for the sensorless drive system.
Design/methodology/approach
The load‐side speed and the torsional torque are estimated using multi‐layer perceptron (MLP) and radial basis function (RBF) networks. The special forms of input vectors for neural state estimators were proposed and tested in open‐ and closed‐loop control structure. The estimation quality as well as sensitivity of neural estimators to the changes of the inertia moment of the load machine were evaluated and compared.
Findings
It is shown that an application of RBF‐based neural estimators can give better accuracy of the load speed and torsional torque estimation, especially for the proper choice of the input vector of NN, also in the case of a big change of the load machine time constant.
Research limitations/implications
The investigation and comparison is based on simulation tests and looked mainly at the quality of state variable estimation while the realisation cost in parallel processing devices (FPGA) still need to be addressed.
Practical implications
The proposed neural state variable estimators of two‐mass system can be practically implemented in the control structure of two‐mass drive with additional feedbacks from load machine speed and torsional torque, which results in the successive vibration damping.
Originality/value
The application of RBF neural state estimators for two‐mass drive and their comparison with commonly used MLP‐based estimators, as well as testing of both type of NN in the closed‐loop control structure with additional feedbacks based on state variables estimated by neural estimators.
Details
Keywords
Teresa Orlowska‐Kowalska, Joanna Lis and Krzysztof Szabat
The paper sets out to deal with the off‐line identification of induction motor (IM) parameters at standstill. Determination of values of the IM parameters is essential in…
Abstract
Purpose
The paper sets out to deal with the off‐line identification of induction motor (IM) parameters at standstill. Determination of values of the IM parameters is essential in sensorless drives with regard to accuracy and quality of the control system.
Design/methodology/approach
The presented identification method is based on the reconstruction of stator current response to the forced stator voltage step change; thus the cost function is defined in the classical form of the mean squared error between the computed and experimental data. The identification via evolutionary algorithms (EAs) is presented. The considered problem is continuous and thus a continuous version of EA is suggested as more suitable.
Findings
This approach has been shown to have several advantages over classical optimisation methods like the ability to cope with ill‐behaved problem domains exhibiting attributes such as: discontinuity, time‐variance, randomness, and, what is particularly important in this application, the ability to cope with the signals disturbed by noises. Owing to this ability the EAs could be implemented directly for the identification of IM parameters not only in simulations but also in the industrial applications for the motor control, though the electrical signals acquired from real motor and used as input data in the identification procedures are to a large extent disturbed by the electrical noises.
Originality/value
Two versions of the suggested approach are compared: the EA with hard selection and with soft selection. Both algorithms were tested in a simulation and experimental set‐up using digital signal processor for control and signal processing of the voltage inverter‐fed IM drive. Satisfactory results were obtained for the identification procedure based on the selected EA.
Details
Keywords
Teresa Orlowska‐Kowalska, Mateusz Dybkowski and Grzegorz Tarchala
The purpose of this paper is to obtain an accurate and robust estimation method of the rotor flux and speed for the sensorless induction motor (IM) drive with magnetizing…
Abstract
Purpose
The purpose of this paper is to obtain an accurate and robust estimation method of the rotor flux and speed for the sensorless induction motor (IM) drive with magnetizing reactance variations.
Design/methodology/approach
The sensorless IM drive with sliding mode flux and speed observer (SMO) is presented. Proposed estimation algorithm is extended with the additional magnetizing reactance estimator based on the magnetizing characteristic of the IM. The dynamical and steady‐state properties of the drive system in the low‐speed and in the field‐weakening regions are tested. The simulation results are verified by experimental tests, over the wide range of motor speed and drive parameter changes.
Findings
It is shown that the sensorless induction motor drive can work stable in wide speed range using the Sliding‐Mode Observer with additional magnetizing reactance estimator.
Research limitations/implications
The investigation looked mainly at the speed estimation methodology with additional motor parameter estimator.
Practical implications
The proposed SMO can be easily implemented on digital signal processors. The implementation was tested in an experimental setup with DS1103 card. The fixed‐point realisation needs to be developed to obtain the practical application in the industrial drive systems.
Originality/value
The SMO with an additional magnetizing reactance estimator based on magnetizing characteristic of the IM is tested. This method of the speed and flux reconstruction can be applied in different electrical drives working in wide speed range, including very low‐speed region and field‐weakening region, too. The proposed solution is not sensitive to magnetizing reactance variations and is simple in practical implementation in the real‐time system.
Details
Keywords
Teresa Orlowska‐Kowalska and Joanna Lis
The purpose of this paper is to obtain a preliminary off‐line identification of induction motor (IM) parameters at standstill in a reasonable calculation time, which will be…
Abstract
Purpose
The purpose of this paper is to obtain a preliminary off‐line identification of induction motor (IM) parameters at standstill in a reasonable calculation time, which will be useful for the initial adjustment of controllers and state observer parameters in the sensorless drive system.
Design/methodology/approach
The identification procedure of electrical parameters of IM equivalent circuit is performed at standstill and is based on the reconstruction of the stator current response to the forced stator voltage using evolutionary algorithms (EAs) with hard selection and different mutation schemes.
Findings
It is shown that an application of the EA with adaptive mutation mechanism based on simulated annealing method gives very good accuracy of parameters identification and the shortest execution time of the identification procedure as well in simulation as in the experimental tests.
Research limitations/implications
The investigation looks mainly at the minimization of the execution time of the identification algorithm and on the identification accuracy performance, taking into account the good approximation of the measured stator current response.
Practical implications
The proposed EA with the improved adaptive mutation scheme can be easily realised using modern digital signal processor (DSP), which is usually applied for control purposes of the sensorless IM drive system with vector control. The implementation is tested in experimental setup with floating point DSP used as the system controller.
Originality/value
The application of adaptive mutation with simulated annealing in the EA with hard selection for the fast, off‐line preliminary identification of the IM parameter at standstill.
Details
Keywords
T. Orlowska‐Kowalska and M. Dybkowski
This paper aims to obtain an accurate and robust estimation method of the rotor flux and speed for the sensorless induction motor (IM) drive.
Abstract
Purpose
This paper aims to obtain an accurate and robust estimation method of the rotor flux and speed for the sensorless induction motor (IM) drive.
Design/methodology/approach
The reduced order observer has been used as an online tuned rotor flux model in the model reference adaptive system (MRAS) concept applied for the IM speed estimation. The output of this observer was used also as a feedback signal required in the direct field‐oriented control (DFOC) structure of the IM.
Findings
It is shown that a new rotor flux and speed estimator are more robust to motor parameter changes in comparison with the classical MRAS estimator and can work stably in the DFOC structure, in the wide speed range, even for relatively high (50 per cent) identification errors of equivalent circuit parameters of the IM.
Research limitations/implications
The investigation looked mainly at the estimation accuracy performance and whole system stability while economic issues will still need to be addressed.
Practical implications
The proposed new improved MRAS speed estimator can be easily realised using modern digital signal processors. The implementation was tested in an experimental set‐up with floating point DSP used as the system controller. The fixed‐point realisation needs to be developed to obtain the practical application in the industrial drive systems.
Originality/value
The application of the reduced order flux observer as a tuned flux model in the MRAS type speed estimator instead of the simple, but very sensitive to motor parameter uncertainties, current flux model, enables much better accuracy and stability of the rotor speed estimation in the complex DFOC structure than in the case of classical MRAS estimator.