The matrix converter (MC) is often projected as a substitute to voltage source converter due to its numerous advantages, especially in converter fed drive systems, where features like bidirectional power flow, common mode voltage (CMV) elimination, improved source, and load side current quality, are highly desirable. This article presents the speed encoderless control of an induction motor, with the exclusive use of only six switching states of MC that generate zero CMV. The use of only these six states, contrary to the 27 (or 21) states used otherwise, also ensures reduced processor burden. For the control of both source, and load side parameters, multiple objective finite set model predictive control (MO-FS-MPC) is used. MO-FS-MPC is conventionally associated with cumbersome weight tuning, which is eliminated by undertaking a fuzzy decision making approach. Furthermore, a full-order predictive model of an induction motor is used for the prediction of flux, and stator currents, allowing control down to zero motor speed. For the estimation of motor speed, the mechanical model of the motor is used, such that parameter drift is significantly overcome through suitable compensation of the error between predicted, and sensed stator currents.

The Direct matrix converter (DMC) offers a compact and robust solution over typical AC-DC-AC converter, by enabling direct AC-AC power conversion without any energy storage requirements. Moreover, it offers enhanced source current quality, innate bi-directionality and power factor correction capabilities. Further, its industrial suitability is advocated by the elimination of common mode voltage through the direct matrix converter topology. Space vector modulation and model predictive control based approach to common mode voltage elimination in matrix converter, are previously proposed. This paper proposes delta sigma modulation for common mode voltage elimination in matrix converter, which is computationally less cumbersome and performs better as compared to previously documented methods. This method utilizes only six most optimum states of direct matrix converter that yield zero common mode voltage, and performs load voltage and source current control by computing each error variable (delta), integrating it (sigma) and further minimizing it within a hysteresis band. Source current control necessitates load current estimation, which is done through predictive model of the load, hence eliminating load current sensors.

As research in matrix converters (MCs) has progressed over time, space vector modulation has emerged as the most popular technique for their modulation. Lately, model predictive control (MPC) has also been explored for the modulation and control of MCs, primarily gaining popularity due to the ease in realization of multiple objectives, as desired in MCs. However, multi-objective MPC employs assignment of weights to each control objective, wherein weight tuning is either empirical in nature or computationally cumbersome. This paper proposes a new modulation technique for MCs, inspired from delta sigma modulation for voltage source converters, that can be used for load voltage and source current control of MCs. Since source currents are samples of the load current envelope, it is required that this envelope be known for the control of source current. To eliminate load current sensors, predictive model of the load is used in envelope estimation. Further, independent switching schemes are generated for each objective, and an error minimization approach is used in integrating the two schemes into a final state selection, hence eliminating weighing factors. Unlike MPC, error predictions for all states of MCs, are not needed, hence facilitating easier implementation without significant processor burden, besides ensuring the quality of response.

Multiobjective model predictive control (Mo-MPC), owing to its simplicity and adaptability to comply with diverse constraints, has gained popularity in the control of direct matrix converters (DMCs). However, the MPC strategy makes the converter operate at an uncontrolled and variable switching frequency, which may inhibit high power density applications due to increased power losses. This article proposes a method for controlling the switching frequency of the converter and simultaneously regulating the common mode voltage in the DMC fed system while maintaining sinusoidal load and source currents at near unity input power factor. Furthermore, in order to streamline the control while simultaneously dealing with these objectives, a suitable normalization technique is implemented for scaling the errors in each objective. This method is validated through simulation and experimental results.

Finite control set model predictive control (FCS‐MPC) has lately received noteworthy attention in the control of power converters. Such converters have a finite number of switching states, which ensure a small sample space of predictions and minimum computational burden. Although unanimously popular in most converters, multi‐objective FCS‐MPC (Mo‐FCS‐MPC) is a particularly attractive choice in the control of matrix converters (MCs), as it enables attainment of multiple objectives with relative ease. Conventionally, a weighing factor‐based approach is undertaken in the implementation of Mo‐FCS‐MPC wherein a cost‐function is framed such that each of its constituent objectives, is assigned a relative weight according to its significance. However, the tuning of weights is empirical in nature and hence tedious. This study proposes an improvised technique for implementing Mo‐FCS‐MPC in MCs while simultaneously meeting a number of objectives such as load current, source current, and input power factor control. Sector information from space vectors of reference output voltages and reference input currents, coupled with a fuzzy decision‐making criterion is used to make the final switching decision, hence eliminating the conventional weighing factor‐based approach. The inclusion of space vector modulation into predictive control enhances the quality of both loads as well as source current waveforms.

Predictive control has emerged as a popular tool in the modulation of power converters and control of sensorless motor drives. However, it is associated with inherent limitations, such as increased processor burden and heuristic weight tuning, when dealing with multiple objectives. Moreover, in applications, such as modulation of matrix converters, predictive control necessitates input filter modeling, besides requiring repeated reruns of load (and input filter) models for securing the most optimum state selection. This article deals with the restricted use of predictive control as an estimation technique only, for speed sensorless control of an induction motor through delta sigma modulated matrix converter. Delta-sigma technique permits conveniently implementable modulation for the power converter, whereas predictive estimation of flux, torque, and speed ensures high precision even at near zero motor speed. Both predictive control and delta-sigma modulation work in tandem, to provide a robust drive solution, as demonstrated through a wide range of simulation and experimental results.