This paper proposes an enhanced strategy for direct torque control (DTC) combining artificial intelligent (AI) and predictive algorithms. The advantages of this merge are in the solution of closed-loop controlled induction machine (IM) problems. Predictive DTC (P-DTC) methods reduce the high torque ripple and improve the performance at both starting condition and low mechanical speed operation. However, P-DTC depends on the IM parameter's knowledge. The approach here is the introduction of fuzzy logic control with dynamic rules based on the P-DTC law's to reduce the parameter dependency and improve the performance of P-DPC. Additional comparative performance study of eight modulation strategies under the proposed fuzzy-predictive DTC (FP-DTC) is conducted. It results that the space-vector modulation (SVM) is the most suitable scheme with the best combination of criteria such stator current total harmonic distortion, switching losses and dynamic behavior. The parameter dependency of the FP-DTC is tested by a sensitivity analysis which corroborates the robustness of the proposed control. For verification purposes, simulations of the DTC, P-DTC, and FP-DTC were conducted and compared. Experimental results for the three controllers and two modulations (pulse width modulation and SVM) confirm the expected performance of the proposed control algorithm and modulation assessment study.
24 January 2017 in IEEE Transactions on Energy Conversion
A parameter estimation method for deep-bar induction machines is presented. The parameters are estimated using two instantaneous voltage and current waveforms during a direct startup. The instantaneous input impedance is used as a stator indicator to solve a constrained nonlinear optimization problem that outputs the model's parameters. For such purposes, a novel analytical expression for the instantaneous input impedance is introduced. The method is validated in two distinct National Electrical Manufacturers Association (NEMA) design type induction machines (designs A and B), and the accuracy of the obtained parameters is determined by comparing the instantaneous input impedance magnitude and angle errors between the deep-bar and single-cage models with experimental data. The two tested motors showed an improvement when implementing the deep-bar model with the estimated parameters. The error decrease is more significant for the NEMA design B motor which corresponds to a deep-bar rotor construction. Finally, the single-cage and deep-bar models are simulated and their outputs are compared to experimental waveforms. The deep-bar model with the estimated parameters outperforms the single-cage model, showing excellent agreement between the experimental and simulated mechanical speed, stator currents, and electromagnetic torque. The results endorse the accuracy of the method and its applicability for transient studies.
Electric motors play a fundamental role in various industries, and their relevance is strengthened in the context of the energy transition. Having efficient tools and techniques to detect and diagnose faults in electrical machines is crucial, as is providing early alerts to facilitate prompt decision-making. This study proposes indicators based on the magnitude of the space vector stator current for detecting and diagnosing incipient inter-turn short circuits (ITSCs) in induction motors (IMs). The effectiveness of these indicators was evaluated using four machine learning methods previously documented in the literature: random forests (RFs), support vector machines (SVMs), the k-nearest neighbor (kNN), and feedforward and recurrent neural networks (FNNs and RNNs). This assessment was conducted using experimental data. The results were compared with indicators based on discrete wavelet transform (DWT), demonstrating the viability of the proposed approach, which opens up a way of detecting incipient ITSCs in three-phase IMs. Furthermore, utilizing features derived from the magnitude of the spatial vector led to the successful identification of the phase affected by the fault.
In this work a parameter estimation method of the induction machine model based on the instantaneous impedance is presented. The proposed method is based on instantaneous voltage and current measurements during start-up of the induction motor at rated voltage. The proposed parameter estimation method uses an space vector description of the transient machine equations, referred to the stator frame of reference, during the dynamic conditions presents at start-up. The method adjusts the dynamic model parameters using an error function obtained by comparing the instantaneous impedance obtained from the space vector representation of the machine and the corresponding obtained from the measured variables at start-up. The method gives a very precise off-line parameter estimation. The solution is obtained using a non-linear regression method that can be used to improve the control behavior in several predictive DTC strategies.
This paper proposes a control scheme for a wind turbine using a DFIG electromechanical converter, implemented through an NPC three-level back to back converter, that keeps a unitary power factor injection to the grid and maximizes the energy harvested from wind, using a maximum power point tracking algorithm (MPPT). A predictive direct power control (DPC) strategy drives the grid-side converter, to maintain the DC bus reference voltage. Whereas, a predictive direct torque control (DTC) strategy drives the machine-rotor-side converter, to control the power extraction, the power factor and balancing of the DC bus capacitors. The developed model allows to study the wind energy conversion system (WECS) for energy harvesting, rotor dynamics and power quality analysis. Simulation results endorse the effectiveness of the advanced control techniques for the whole wind spectrum, including the pitch angle control. A sensitivity analysis shows that the predictive DTC control strategy is robust with an uncertainty up to 20% of the induction machine parameters.