Neural based MRAS sensorless techniques for high performance linear induction motor drives

This paper presents a new model reference adaptive system (MRAS) speed observer for high-performance field-oriented control induction motor drives based on adaptive linear neural networks. It is an evolution and an improvement of an MRAS observer presented in the literature. This new MRAS speed observer uses the current model as an adaptive model discretized with the modified Euler integration method. A linear neural network has been then designed and trained online by means of an ordinary least-squares (OLS) algorithm, differently from that in the literature which employs a nonlinear backpropagation network (BPN) algorithm. Moreover, the neural adaptive model is employed here in prediction mode, and not in simulation mode, as is usually the case in the literature, with a consequent quicker convergence of the speed estimation, no need of filtering the estimated speed, higher bandwidth of the speed loop, lower estimation errors both in transient and steady-state operation, better behavior in zero-speed operation at no load, and stable behavior in field weakening. A theoretical analysis of some stability issues of the proposed observer has also been developed. The OLS MRAS observer has been verified in numerical simulation and experimentally, and in comparison with the BPN MRAS one presented in the literature.

This study investigates the behavior of a Five-Phase Permanent Magnet Synchronous Motor (5P-PMSM) when an open-phase fault occurs. It applies a sensorless control approach that combines backstepping control with a Model Reference Adaptive System (MRAS) observer. Using measurable electrical quantities, including stator currents and voltages, the proposed method accurately estimates both rotor position and speed. The design of the backstepping controller follows a structured procedure to achieve global or semi-global stability, making it suitable for applications that demand reliable operation under changing conditions. Lyapunov theory is used to analyze and verify the stability of the combined observer and controller. The main objective of the research is to maintain motor performance and stable operation in the presence of an open-phase fault. Simulations conducted in MATLAB/Simulink compare normal motor performance with performance during the fault to evaluate the effectiveness of the control strategy. The findings show that the backstepping controller, supported by the MRAS observer, enhances fault tolerance and reduces the adverse effects. The simulation results confirm that the proposed approach sustains motor operation with minimal decline in performance, offering a dependable solution for industrial systems requiring resilience to faults.

In order to realize sensorless control for brushless doubly fed induction machine (BDFIM), this paper presents a model reference adaptive system (MRAS) observer, designed based on the error of the control winding current. Furthermore, a phase-locked loop is employed to estimate the current winding position and rotor speed. Consequently, a detailed theoretical derivation proves that the MRAS observer is stable and the dynamic performance is good. Thus, it does not cause any estimated speed error in steady state. Moreover, the estimated position error is bounded and trivial, thus its effects on the sensorless control of the BDFIM are neglected. The correctness, feasibility and robustness of the proposed sensorless control method are verified by means of experimental validation on a 30 kW test rig.

The paper describes Model Reference Adaptive System (MRAS) observers for the speed estimation of an induction motor with direct torque and flux control. The first estimator is a reference frame MRAS (RF-MRAS) and the second estimator is a current based MRAS (CB-MRAS). At first, direct torque controlled induction motor drive with two estimators are implemented on Matlab-Simulink environment. Then, comparison of two observers is done by evaluation of the rotor speed difference. The simulation results confirm that both MRAS estimators are simple to simulate and experiment. By comparison of both observers, the CB-MRAS observer gives higher accuracy of the rotor speed estimation.

An improved finite-set predictive thrust control (FSPTC) is proposed in this paper, in which the fuzzy logic control (FLC) is applied instead of PI control in both speed control loop and adaptation mechanism of the model reference adaptive system (MRAS) with a lower number of membership functions to reduce the computation time. The MRAS observer based on primary flux linkage is implemented to estimate the linear speed. The output of the FLC serves as the reference thrust of the FSPTC while as speed in the MRAS observer. The developed control targets are to improve the characteristics of the drive system, where a significant ripple reduction in thrust and primary flux linkage, fast dynamic response, and lower cost can be guaranteed. Comprehensive simulation results have demonstrated the effectiveness of the proposed method based on one arc induction machine with 3kW rated power.

This paper proposes the theoretical framework and the consequent application of the input-output Feedback Linearization (FL) control technique to Linear Induction Motors (LIM). LIM, additionally to Rotating Induction Motor (RIM), presents other strong non-linearities caused by the dynamic end effects, leading to a space-vector dynamic model with time-varying inductance and resistance terms and a braking force term. This paper, starting from a recently developed dynamic model of the LIM taking into consideration its end effects, defines a FL technique suited for LIMs, since it inherently considers its dynamic end effects. Additionally, it proposes a technique for the on-line estimation of the inductor resistance, based on Model Reference Adaptive System (MRAS) on-line estimator, exploited for adapting on-line the FL control action versus inductor resistance variations leading to undesirable steady-state tracking errors. The stability of the proposed MRAS on-line estimator has been proven theoretically, adopting the Popov’s criterion for hyperstability. The proposed approach has been validated experimentally on a suitably developed test set-up, under both no load and loaded conditions. It has been compared firstly with the simplest control structure, which is the scalar V/f control, secondly, under the same closed-loop bandwidths of the flux and speed systems, with the industrial standard in terms of high performance control technique: Field Oriented Control (FOC).

This article presents the sensorless control of a PMSM based on the FOC (Field Oriented Control) strategy and on PFC (Power Factor Correction), where the rotor speed is provided by speed observers. It presents the block structure of the PFC together with the numerical simulations achieved in Matlab/Simulink, on the improvement of the PF (Power Factor) of the supply circuit. It also presents the block structures of the FOC strategy and of the back-EMF (back electromotive force) sliding mode, MRAS (Model Reference Adaptive System), and PLL (Phase Locked Loop) type observers. The numerical simulations show the superiority of rotor speed estimation when using the back-EMF sliding mode observer compared to the rotor speed estimate provided by MRAS and PLL observers.

The underground mines, oil and gas industries need remote operation of vector controlled medium voltage variable speed induction motor drives with long feeder.The usage of voltage source inverters in such cases leads to motor over voltage and harmonic problems.The current source inverter (CSI) has no such problems to these applications because of its motor friendly voltages output.This paper proposes "sensor less operation of a CSI fed induction motor drive" and improves the speed estimation.The proposed model uses model reference adaptive system (MRAS) speed estimator requires motor voltage and currents as inputs.Direct measurement of motor voltages and currents are impractical for long feeder drives due to feeder length and safety of environmental precautions.Hence the motor voltages and currents are drawn from the inverter out.In this paper the flux estimation needed for MRAS speed estimation is proposed to observe the performance of the induction motor with sudden change of loads under steady state.This model is simulated in MATLAB/SIMULINK.

A sensorless indirect stator-flux-oriented control (ISFOC) induction motor drive at very low frequencies is presented herein. The model reference adaptive system (MRAS) scheme is used to estimate the speed and the rotor resistance simultaneously. However, the error between the reference and the adjustable models, which are developed in the stationary stator reference frame, is used to drive a suitable adaptation mechanism that generates the estimates of speed and the rotor resistance from the stator voltage and the machine current measurements. The stator flux components in the stationary reference frame are estimated through a pure integration of the back electro-motive force (EMF) of the machine. When the machine is operated at low speed, the pure integration of the back EMF introduces an error in flux estimation which affects the performance torque and speed control. To overcome this problem, pure integration is replaced with a programmable cascaded low-pass filter (PCLPF). The stability analysis method of the MRAS estimator is verified in order to show the robustness of the rotor resistance variations. Experimental results are presented to prove the effectiveness and validity of the proposed scheme of sensorless ISFOC induction motor drive.

This paper proposes a propulsion control algorithm considering End-effect of a linear induction motor (LIM) used in a magnetic levitation train. Indirect field oriented control (IFOC) technique is one of the popular control techniques widely applied to LIM drive control. The main idea of IFOC in a LIM is the decoupling of the flux and torque. Therefore, IFOC has a higher dynamic characteristic than the scalar control [1]. The LIM applied to the magnetic levitation trains should control the thrust force and also take into account the normal force for levitation stabilization. When the LIM is driven, normal force is inevitable. The normal force is a function of the speed, the slip frequency and the current of the train. In the rotary type induction motor, the normal force is canceled due to its symmetrical structure. Therefore, IM could operate in a low slip and high efficiency range. However, in the case of LIM, a high attraction force is generated in a low slip region, which cause the instability of the levitation system. Therefore, the propulsion control for magnetic levitation trains must operate in a higher slip region than the slip used in conventional IFOC. In this paper, the slip characteristics of the LIM are analyzed to improve the efficiency and safety of the magnetic levitation train by FEM analysis. A constant slip frequency propulsion control algorithm is proposed for operation in the analyzed slip region. The proposed algorithm can be controlled by separating magnetic flux and thrust from IFOC. The proposed algorithm is verified in 250 kW test bench. The results of FEM analysis according to speed and slip at constant voltage condition are shown on Fig. 1. The dotted lines of Indirect Field Oriented Control(IFOC) and Constant Slip Indirect Field Oriented Control(CSIFOC) are shown on the figure. The slip of the IFOC is low and that leads to operate near the maximum thrust and efficiency. However, the normal force of LIM is also increase, which could not be allowed in maglev application. On the other hand, when CSIFOC is performed considering normal force, the slip is higher than the conventional algorithm. Although the CSIFOC is not operated in maximum thrust slip, the normal force could be greatly reduced. Fig.2 (a) shows the CSIFOC algorithm considering the end-effect of the linear induction motor. In general IFOC, the d-axis current reference is calculated through the magnetic flux reference value, and the q-axis is calculated based on thrust. However, the CSIFOC for the LIM generates the d-axis and q-axis current references along with the slip frequency and thrust. Fig. 2 (b) shows the waveform of the experimental results based on the general IFOC and the proposed algorithm. Both IFOC and CSIFOC were tested with the same rate pattern of the same load. It is confirmed that CSIVC has larger slip than conventional IFOC and low normal force in all regions. In this paper, a constant slip frequency IFOC algorithm is proposed considering the end effect of LIM used for magnetic levitation trains. The efficiency, thrust and normal force according to the speed and slip through the FEM analysis of the LIM were analyzed and the slip condition suitable for the magnetic levitation train is proposed. In addition, a current reference generation algorithm based on the thrust and slip is proposed in the IFOC considering the end effect to operate under the proposed slip condition. In the full paper, the detailed formulas of the proposed algorithm are developed and will be verified through simulation and experimental results.

The paper investigates the improvement in sensorless vector control of induction machines through the use of a non-linear feedback integrator embedded in a standard rotor flux model reference adaptive system (MRAS)-based observer. The feedback integrator compensates for drift and dc-offset and is capable of open loop flux estimation down to frequencies below 0.1 Hz. The paper presents experimental results of sensorless speed controlled induction motor exploiting a MRAS observer. The sensorless system shows good performance down to frequencies of 0.2 Hz.

In this paper, a sliding mode observer (SMO) for the speed-sensorless direct torque vector control of linear induction motor(LIM) is proposed, considering the influence of dynamic end effects. The proposed SMO estimate the motor speed, the rotor flux, the angular position of the rotor flux, and the speed dependent parameter w, only by using the measured stator voltages and stator currents. The nonlinear sliding mode technique guarantees very good performance for different situations of motor speed. The proposed SMO does not use complicated observer gains and robust to the parameter variation over the full-speed range. Moreover, the convergence of SMO is proven by using the Lyapunov theory. Furthermore, LIM Drive state observability is presented with detailed analysis. The performance of the proposed SMO is verified by hardware-in-the-loop test (HIL) system. Furthermore, a comparative study with and Luenberger observer(LO) and model reference adaptive system(MRAS) is presented.

In this study an adaptive recurrent-neural-network controller (ARNNC) is proposed to control a linear induction motor (LIM) servo drive. First, the secondary flux of the LIM is estimated with an adaptive flux observer on the stationary reference frame and the feedback linearization theory is used to decouple the thrust force and the flux amplitude of the LIM. Then, an ARNNC is proposed to control the mover of the LIM for periodic motion. In the proposed controller, the LIM servo drive system is identified by a recurrent-neural-network identifier (RNNI) to provide the sensitivity information of the drive system to an adaptive controller. The backpropagation algorithm is used to train the RNNI on line. Moreover, to guarantee the convergence of identification and tracking errors, analytical methods based on a discrete-type Lyapunov function are proposed to determine the varied learning rates of the RNNI and the optimal learning rate of the adaptive controller. The effectiveness of the proposed control scheme is verified by both the simulated and experimental results. Furthermore, the advantages of the proposed control system are indicated in comparison with the sliding mode control system.

In this paper, a new active damping method of LCL filter without capacitor voltage sensors is proposed for 3 phase PWM Inverter. Normally, L filter or LCL filter is used as an output filter of grid connected PWM inverter. An LCL filter has more excellent performance than L filter to reduce harmonic current, so the small inductance value can be used. However, the resonance problem in LCL filter is happen due to the zero impedance by the addition of LC branch. To solve the resonance problem, the various active damping method has been proposed so far. Generally, the virtual resistor active damping methods is required to additional hardware sensors for measurement of capacitor voltage and current. In this paper, the new active damping method is proposed without any capacitor voltage or current sensors. In the proposed method, the resonance component of the capacitor voltage of LCL filter can be observed by a simple MRAS(Model Reference Adaptive System) observer without additional hardware sensors, and this component is suppressed by feedforward compensation. The validity of the proposed method is proven by simulation and experiment on the 3-phase PWM inverter system.

In this paper, a new grid voltage estimation algorithm without voltage sensors is proposed for the three-phase vienna rectifier. Generally, input voltage sensor circuits increase size and cost of the PWM rectifier In order to reduce the cost and size and in order to increase reliability from the electrical noise, grid voltage estimation scheme without input voltage sensor is highly required. In this paper, the grid voltage estimation algorithm is proposed by a simple MRAS(Model Reference Adaptive System) observer without input voltage sensors. The validity of the proposed method is proven by simulation and experiment on the three-phase vienna rectifier system.