Search

To achieve high efficiency and dynamics in electric drive applications, it is necessary to have accurate torque control. This is typically accomplished through a current regulator that is fed by references generated by various open-loop control strategies, in order to obtain the desired torque. As an alternative, this work presents a model predictive torque control. Starting from the torque reference, the algorithm generates optimal voltage references to the inverter-fed synchronous motor drive, while working at maximum efficiency and considering the motor current limit. This feature is achieved by combining two different norms in the cost function. According to the paradigm of the more autonomous drives, an important feature is that the algorithm requires only knowledge of the motor model. This means that a tuning procedure for control weight is no longer required as analytically discussed in this work. Experimental validation of the proposed technique are performed on a test rig featuring an anisotropic permanent magnet motor in different dynamic operation, including flipping from the motor nominal working point to the generator one.

The work focuses on a new mathematical formulation for the transfer function of a pulsating injection-based position observer, as the key point for its correct design. The proposed analysis exploits the modulation/demodulation control theory to derive the transfer function of the whole estimator in the Laplace domain. Magnetic iron saturation effects and cross–saturation between the dq-axes are taken into consideration in the observer design. To test the effectiveness of the new transfer function representation of the observer, it is used to synthesise two different regulators, namely by the direct synthesis and the internal model principle. An extensive experimental stage is included pointing to fully verify the accuracy of the proposed model and the theoretical considerations. A comparison with the existing mathematical model of the observer is also reported to show the improvement achievable in the position estimation by the proposed observer synthesis.

By applying rotor angle sensorless control methods the costs of the electrical drive can be decreased while its reliability is increased. Traditionally, the design of both the rotor angle estimator and the drive controller require a detailed motor model and manual tuning leading to significant effort by human experts. In this work, a rotor anisotropy-based sensorless control scheme for interior permanent magnet synchronous motors (IPMSM) is proposed that enables sensorless control with minimum tuning effort and a priori motor knowledge. Due to these characteristics the scheme is particularly suitable for self-commissioning or low-cost drive applications. The scheme contains a finite control set model predictive current controller (FCS-MPCC) with an additional inequality constraint that enables the identification of the motor model in the stator-fixed coordinate system by using the last three measurement samples. Utilizing an automatic system identification procedure, the motor model is determined online in a data-driven fashion. The identified motor model serves both as prediction model of the FCS-MPCC and as baseline for the rotor angle estimation via an eigenvalue decomposition approach. Challenging experimental investigations at standstill up to the medium speed range prove the applicability of the proposed approach. Here, highly dynamic speed and current transients can be handled by the proposed method.

The increasing complexity of modern industrial systems calls for automatic and innovative predictive maintenance techniques. As suggested by the Industry 4.0 process, this demand translates in the need of more‐intelligent drives. Herein, the use of a special kind of neural networks to interpret the data from motor currents for diagnostic purposes is described. The early detection of possible faults in the electrical motor allows programmed maintenance and reduces the risk of unplanned shutdowns. The innovation is in the overall approach to the neural network training, which does not call anymore for a large set of faulty motors. A large training dataset generated using a combination of tuned motor models and some data augmentation techniques is proposed. The result is a comprehensive and effective motor condition monitoring algorithm, whose hearth is a convolutionary neural network trained by a safe and cheap simulation‐based dataset. The details of the design are fully reported here. The method has been implemented in the laboratory and fully tested on both healthy and faulty permanent magnet synchronous motors. The generality of the proposed method also paves the way for the detection of other failures and the application to different electrical motors.

This article deals with a finite-set model predictive current control in synchronous motor drives. The peculiarity is that it does not require the knowledge of any motor parameter. The inherent advantage of this method is that the control is self-adapting to any synchronous motor, thus easing the matching between motor and inverter coming from different manufacturers. Overcoming the flaws of the existing lookup table based parameter-free techniques, the article elaborates the past current measurements by a recursive least-square algorithm to estimate the future behavior of the current in response to a finite set of voltage vectors. The article goes through the mathematical basis of the algorithm till a complete set of experiments that prove the feasibility and the advantages of the proposed technique.

Synchronous reluctance motors are arousing lively interest as a possible alternative to the less efficient induction motors. An open issue is the effective tuning of the inner current loops because of the nonlinearity that cannot be overlooked. The present paper uses a relay feedback approach to perform autotuning without resorting to any parameter knowledge. The tuning is iterated at different working points, to get a uniform current control bandwidth everywhere. Unlike many solutions in the field, the algorithm is truly autonomous, in the sense that it also suggests a correct value for the bandwidth specification. The paper includes both simulation and experimental results, obtained on a laboratory prototype.

The article deals with the flux linkages estimation of a synchronous motor during either speed or load transients. The nonlinear magnetic model is obtained by a radial basis function neural network (NN), which yields a differentiable function as link among currents and flux linkages. As a distinctive feature, the training of the NN is performed by an innovative algorithm that operates both at steady state and during speed and load transients. The method has been validated experimentally by several tests performed on different synchronous reluctance and interior permanent magnet motor drives.

Stray flux analysis is an interesting source of information for the diagnosis of Induction Motors (IMs). The widespread use of these motors in industry leads to a necessity of additional tools and methods for their predictive maintenance. On the other hand, soft-starters are increasingly used to reduce the high consumption of IMs at start-up. In this work, AI techniques based on convolutional neural networks are applied to detect rotor faults in soft-started motors. The objective is the automatic early detection of broken bars, avoiding the necessity of user intervention to interpret the obtained results. This work proves the potential of the methodology, including a successful set of experimental results.

The performances of a model predictive control algorithm largely depend on the knowledge of the system model. A model-free predictive control approach skips all the effects of parameters variations or mismatches, as well as of model nonlinearity and uncertainties. A finite-set model-free current predictive control is proposed in this paper. The current variations predictions induced by the eight base inverter voltage vectors are estimated by means of the previous measurements stored into lookup tables. To keep the current variations information up to date, the three current measurements due to the three most recent feeding voltages are combined together to reconstruct all the others. The reconstruction is performed by taking advantage of the relationships between the three different base voltage vectors involved in the process. In particular, 210 possible combinations of three-state voltage vectors can be found, but they can be gathered together in six different groups. A light and computationally fast algorithm for the group identification is proposed in this paper. Finally, the current reconstruction for the prediction of future steps is thoroughly analyzed. A compensation of the motor rotation effect on the input voltages is proposed, too. The control scheme is evaluated by means of both simulation and experimental evidences on two different synchronous reluctance motors.

The growth in cost and complexity of modern industrial plants leads to decreasing tolerance for performance degradation or system downtimes. In this context, the paper deals about the automatic condition monitoring of the rotor bars of induction AC motors which is performed through two main components. The former is the extraction of the Short-Time Fourier Transform of the motor stray flux during start ups as a fault-related index. The latter is the automatic classification and recognition of the signal through an Artificial Intelligence-based algorithm: a Convolutional Neural Network. This cutting-edge tool is particularly suitable for knowledge-based image recognition problems and its feasible training is here permitted by some data augmentation techniques.

Climate changes and the lack of running water across vast territories require the massive use of pumping systems, often powered by solar energy sources. In this context, simple drives with high-efficiency motors can be expected to take hold. It is important to emphasise that simplicity does not necessarily lie in the control algorithm itself, but in the absence of complex manual calibration. These characteristics are met by synchronous reluctance motors provided that the calibration of the current loops is made autonomous. The goal of the present research was the development of a current control algorithm for reluctance synchronous motors that does not require an explicit model of the motor, and that self-calibrates in the first moments of operation without the supervision of a human expert. The results, both simulated and experimental, confirm this ability. The proposed algorithm adapts itself to different motor types, without the need for any initial calibration. The proposed technique is fully within the paradigm of smarter electrical drives, which, similarly to today’s smartphones, offer advanced performance by making any technological complexity transparent to the user.

Model predictive control (MPC) requires an accurate system model to achieve favorable performance. Thus, in presence of disturbances, model uncertainties and mismatches, MPC needs tools that provide high degree of robustness to them. Since MPC is, essentially, a proportional control technique, an effective method to deal with the aforementioned issues is the addition of an integrating element to the control scheme. This paper presents a prediction model that introduces an integrator tothe control strategy without increasing the size of the optimization problem. To examine its effectiveness, the sensitivity of the classical and the proposed MPC to parameter deviations are discussed and analyzed, considering a wide range of switching frequencies as well as prediction horizon lengths. The robustness examination is performed based on an industrial case study, namely a medium voltage induction motor drive. acceptedVersion

Model predictive control (MPC) lacks an integrating element. Thus, parameter mismatches can deteriorate its steady-state performance. To address this issue and enhance the robustness of MPC, analternative formulation of the prediction model is discussed in this paper. This model introduces an integrator to the optimization problem without increasing its size and consequently its computational complexity. An in-depth analysis of the effect of parameter mismatches on the control performance is performed when both the conventional and the proposed prediction model are used. Specifically, the aforementioned analysis is carried out for a range of switching frequencies as well as prediction horizon lengths, while a permanent magnet synchronous motor (PMSM) drive is used as a case study. acceptedVersion

The high-frequency signals normally used for sensorless operations in anisotropic permanent-magnet synchronous motors cause unwanted torque ripple, noise, and vibrations. The target of the present research is to extend downwards the speed range in which the sensorless drive can work without any signal injection. Specifically, this paper proposes a new back-electromotive force observer that eliminates the instability that may arise in conventional observers when the motor drive operates in generating mode. Two different versions of the observer are proposed and discussed, to include motors with high harmonic content in the back-electromotive force. The detailed mathematical analysis of the proposed observer is supported by simulations and the practical feasibility is proven through experimental results.

The general trend toward more intelligent energy-aware ac drives is driving the development of new motor topologies and advanced model-based control techniques. Among the candidates, pure reluctance and anisotropic permanent magnet motors are gaining popularity, despite their complex structure. The availability of accurate mathematical models that describe these motors is essential to the design of any model-based advanced control. This paper focuses on the relations between currents and flux linkages, which are obtained through innovative radial basis function neural networks. These special drive-oriented neural networks take as inputs the motor voltages and currents, returning as output the motor flux linkages, inclusive of any nonlinearity and cross-coupling effect. The theoretical foundations of the radial basis function networks, the design hints, and a commented series of experimental results on a real laboratory prototype are included in this paper. The simple structure of the neural network fits for implementation on standard drives. The online training and tracking will be the next steps in field programmable gate array based control systems.

The large majority of technological devices can be seen as the sum of components of heterogeneous nature (electrical, mechanical, thermal, software, etc.) so that their analysis and design calls for the use of tools from different engineering disciplines. Integration among the different tools is particularly relevant at control system design level, since it is at this stage that it is required to analyze the behavior of both each single component and the system as a whole, with different levels of detail. In this paper mechatronic systems are considered, that exhibit a strong interplay between mechanical and electrical components, and the issue of modeling and designing embedded control algorithms and architectures for such systems is addressed. In particular, an integrated virtual prototyping approach for analyzing the system behavior down to embedded software level is proposed, that can be used in a wide number of situations that can represent the actual real-world operational conditions of consumer products. This approach can be used for system design (in particular at control systems level), embedded software design, and virtual testing so as to optimize and reduce the costs of late stage software development, physical prototypes, and their testing. The case study is based on some recently derived software tools that perform the co-simulation of the firmware execution and the multi-physical controlled system dynamics. The actual control software implemented in the final product can be entirely developed and tested inside the virtual prototype. To prove the validity and potentialities of the proposed approach, a real case study is presented, regarding a very common, though complex to simulate, mechatronic system such as an electric sliding gate. It turns out that the proposed environment goes beyond hardware-in-the-loop tools, since it does not require the use of specific hardware (the hardware itself is simulated in detail) but allows to analyze in detail the status of the microprocessors and peripherals at arbitrary time-scales and allows the designer to study at the earliest design stages the dynamics between the multi-physical controlled system and the firmware, without committing to a given hardware structure.

This paper presents an Artificial Neural Network (ANN) model for predicting the dynamic viscosity of oxide nanoparticles suspension in water and ethylene glycol. The model accounts for the effect of temperature, nanoparticle volume fraction, nanoparticle diameter, cluster of nanoparticles average size, and base fluid properties. The model was trained on a set of data obtained by the present authors and tested on data coming from other authors. The model shows a fair agreement in predicting experimental data: the mean absolute percentage error (MAPE) is 4.15%. The characteristic parameters of the ANN model are reported in details in the paper.

The paper deals with the flux linkages estimation of a synchronous motor during speed or load transients. The proposed method is based on the radial basis function neural network, which returns continuous functions between currents and magnetic flux linkages, considering all the nonlinearities. The training of the neural network weights is realised by an innovative training algorithm able to cope with both steady state and variable working conditions, conversely to most of the identification techniques. The method was validated throughout several tests performed on a synchronous reluctance motor electric drive.

This paper deals with the non-linear modelling of synchronous machines by using the flux linkage as a state variable. The model is inferred from a conventional set of measurementswhere the relation between the currents and the flux linkages in the rotating reference frame (also known as dq reference frame) are known by measurements or estimated through finite-element simulations. In particular, the contribution of this paper is twofold: first, it proposes a method to extract the non-linear model information which can be easily implemented in electric drives, without the need of offline post-processing of the data. Second, it mathematically demonstrates that the method converges to the final result in a stable way. An example based on experimental measurements of the current-to-flux look-up tables of an 11-kW synchronous reluctance machine is shown, proving the feasibility of the proposed method. QCR 20180919

In modern ac drives the reliability issue is pushing towards sensorless control. Extended Kalman filter is a valid candidate for the estimation of rotor position in permanent magnet synchronous motor drives. The traditional weak point is the tuning of its covariance matrices, which is still a loose thread to tie up after many years. This work proposes an innovative automatic procedure for the covariance matrices tuning based on both the rated values of the motor and the result of signal injection tests. The output of the automatic procedure was tested on an experimental motor prototype.

The MTPA algorithms are attractive solutions for reaching very high energy efficient drives. The paper presents a general formulation of the MTPA problem statement, including also the magnetic saturation in the formulation. A classification and a review of the MTPA techniques is proposed in this paper. A distinctive feature of the proposed classification is the highlighting of the most important characteristics (e.g. needs of motor parameters, signal injection, self-commissioning tests, etc.) of each technique. Also, offline and online techniques are thoroughly discussed. All synchronous motors, with or without permanent magnets, are considered.

As neodymium and other rare earth materials become a critical commodity, the exploitation of reluctance torque in synchronous motors is certainly an interesting option. Alternative motor topologies span from internal permanent magnet motors to pure reluctance machines. Anyway, any anisotropic structure suffers from some magnetic nonlinearities that call for more sophisticated models to get an efficient torque control. Neural network-based algorithms are good candidates for modeling the current-to-flux linkages curves of synchronous reluctance (SynR) motors, but so far their use was limited by the inherent complexity and the computational burden. This paper proposes the use of a special kind of neural networks, namely, the radial basis function networks, to get the magnetic model of any synchronous motor, including saturation and cross-coupling effects. Through experimental evidence, it will be shown that the structure is light enough to be implemented, trained, and self-updated online on standard high-end ac drives. The model is used to track online the maximum torque-per-ampere working point of a SynR motor drive.

This paper presents an Artificial Neural Network (ANN) model for predicting refrigerant boiling heat transfer coefficients inside Brazed Plate Heat Exchangers (BPHE). The model accounts for the effect of plate geometry, operating conditions and refrigerant properties. The model shows a fair agreement with a database of 1760 data points comprising 15 plate geometries and 16 refrigerants (including 4 natural refrigerants and 6 other low-GWP refrigerants). The Mean Absolute Percentage Error (MAPE) of the model predictions is 4.8%. The ANN model exhibits a better predictive capability than most of the state-of-the-art analytical-computational procedures for boiling inside BPHE available in the open literature. The characteristic parameters of the ANN model are fully reported in the paper.

This paper presents an Artificial Neural Network (ANN) model for predicting refrigerant condensation heat transfer coefficients inside herringbone-type Brazed Plate Heat Exchangers (BPHE). The model accounts for the effect of plate geometry, operating conditions and refrigerant properties both in the saturated and in the superheated vapour condensation regimes. The model predictions demonstrate good agreement with a database of 1884 data points comprising 12 plate geometries and 16 refrigerants (including 4 so-called natural refrigerants and 6 other low-GWP refrigerants). The Mean Absolute Percentage Error (MAPE) of the model predictions is 3.6%. The results demonstrates that the ANN model presented herein is capable of better predictive capability than most of the state-of-the-art BPHE analytical-computational models presented in the open literature. The characteristic parameters of the ANN model are reported in the paper.

Model Predictive Control scheme performances are strongly influenced by an accurate knowledge of the plant model. A Model-Free predictive control approach permits to overcome all the approximations due to parameters variations or mismatches, model non linearities or inadequacies. A Finite-Set Model-Free Current Predictive Control is thus proposed in this paper. The current variations predictions related to the eight feeding voltages are performed by means of the previous measurements stored into look-up tables. To keep the current variations information up to date, the three current reactions related to the three most recent feeding voltages are combined together to reconstruct all the others. The reconstruction is performed by taking advantage of the relationships between three different state voltage vectors. In particular, 210 possible combination of three state voltage vectors can be found, but they can be gathered together in six different groups. A novelty introduced in this work is a light and computationally fast algorithm for the current variation reconstruction based on the state voltage vectors group identification. Finally, the current reconstruction for the prediction at future steps is thoroughly analysed. A compensation of the motor rotation effect on the input voltages $dq$ projections is proposed, too.

A finite control set model predictive control strategy for the control of the stator currents of a synchronous reluctance motor driven by a three-level neutral point clamped inverter is presented in this paper. The presented algorithm minimizes the stator current distortions while operating the drive system at switching frequencies of a few hundred Hertz. Moreover, the power electronic converter is protected by overcurrents and/or overvoltages owing to a hard constraint imposed on the stator currents. To efficiently solve the underlying integer nonlinear optimization problem a sphere decoding algorithm serves as optimizer. To this end, a numerical calculation of the unconstrained solution of the optimization problem is proposed, along with modifications in the algorithm proposed in [1] so as to meet the above-mentioned control objectives. Simulation results show the effectiveness of the proposed control algorithm. acceptedVersion

Sensorless drives for permanent magnet (PM) and reluctance synchronous motors are quite interesting options in modern ac drives. Fundamental-frequency observers provide a rotor position estimate, to be used instead of that measured by delicate sensors. The estimation algorithms are based on either the extended electromotive force or active flux concepts. They feature easy implementation, although their reliability may be undermined by both corrupted measurements and parameter variations, especially in the case of anisotropic rotor structures. Through an analytical sensitivity analysis, this article investigates the influence of model uncertainty on the observed position. The substantial relationships between cause (parameter uncertainty or measurement error) and effect (position error) are validated by experimental results, which also highlights the effects of the q -axis inductance and resistance values on the electric measurement sensitivities. The resulting functions and graphic plots are useful to draw interesting considerations about the motor selection and the control design of the sensorless drive.

This study presents an Artificial Neural Network (ANN) model for predicting the dynamic viscosity of H2O/KCOOH (potassium formate) solution. The model accounts for the effect of temperature and concentration in salt and it covers the concentrations typical for brine (0–50%) and desiccant (60–80%) applications, including also pure water. The model shows a fair agreement in predicting experimental data: the mean absolute percentage error (MAPE) is 0.92%. The characteristic parameters of the ANN model are fully reported in the paper.

In the last decade, interior permanent magnet and synchronous reluctance motors are receiving attention from both academic and industrial environments, that appreciate the considerable advantages in terms of cost reduction and power density. However, these motors call for more advanced control techniques, due to the rotor anisotropy and the nonlinearity of the magnetic circuit. Reliability and cost also push towards sensorless solutions. Often, the fundamental-frequency based techniques are preferred to the high frequency injection, for the noise and vibrations that the latter implies. Unfortunately, the former cannot work down to zero speed. The aim of this paper is the study and development of a simple and effective start-up technique for sensorless synchronous motor drives. Merged with fundamental-frequency based sensorless control, it yields a full-range ac drive that fits for many applications that do not require steady state operation at low/null speed. In particular, the paper paves the way for integrating the motor model in the start-up design, highlighting pitfalls of the standard methods and showing how to improve the transition to the closed-loop control.

Solar water pumping systems provide cost-effective water supplies whenever the mains supply is unreliable. This green solution well matches with synchronous reluctance motor drives, characterised by high efficiency and lighter ecological footprint, compared to the rare-earth permanent magnet counterpart. The challenge is to drive the water pump extracting the maximum power from the photovoltaic panel, at varying sunlight conditions. The proposed algorithm is based on an accurate model of the synchronous reluctance motor, obtained by the offline training of a low-complexity neural network, specifically developed to the purpose. The maximum power point tracking is obtained by modifying the speed of the water pump impeller, and the working point of the solar panel accordingly, to meet the instantaneous solar radiation conditions.

This paper shows how the problem of estimating the initial rotor position in permanent-magnet sensorless drives can benefit from the availability of a direct measurement of motor phase voltages. Two different estimation methods are presented for this purpose. The first is a readjustment of a classic procedure based on the detection, by means of the injection of voltage test pulses, of inductance variations due to motor saliencies, from which it is then possible to infer the position of the rotor. Since the application of irregular test pulses may increase the estimation uncertainties, the available voltage measurement is exploited to implement a closed-loop amplitude control of the test pulses. The second method is introduced both to overcome an issue related to the digital measurement approach adopted in this paper and to allow position estimation even in a motor with no relevant saliency, a case for which the first method is inappropriate. The effectiveness of the proposed solutions is validated by several experimental tests, which are carried out on two motors with different saliency properties.