Fuzzy Logic Based Direct Torque Control Of Induction Motor With Space Vector Modulation

The induction motors have wide range of applications for due to its well-known advantages like brushless structures, low costs and robust performances. Over the past years, many kind of control methods are proposed for the induction motors and direct torque control has gained huge importance inside of them due to fast dynamic torque responses and simple control structures. However, the direct torque control method has still some handicaps against the other control methods and most of the important of these handicaps is high torque ripple. This paper suggests a new approach, Fuzzy logic based space vector modulation, on the direct torque controlled induction motors and aim of the approach is to overcome high torque ripple disadvantages of conventional direct torque control. In order to test and compare the proposed direct torque control method with conventional direct torque control method simulations, in Matlab/Simulink,have been carried out in different working conditions. The simulation results showed that a significant improvement in the dynamic torque and speed responses when compared to the conventional direct torque control method.

💡 Research Summary

The paper addresses a well‑known drawback of Direct Torque Control (DTC) for induction motors—excessive torque ripple—by integrating fuzzy logic with Space Vector Modulation (SVM). Traditional DTC selects one of six active voltage vectors or two zero vectors in a discrete manner, which yields fast torque response and a simple control structure but also causes abrupt changes in torque and flux, resulting in pronounced ripple, acoustic noise, and reduced efficiency. Prior attempts to mitigate this problem have employed PWM‑based fine‑grained voltage selection or hybrid DTC‑SVM schemes, yet they either retain a fixed voltage‑vector selection rule or demand complex parameter tuning that hampers real‑time implementation.

In the proposed method, the torque error (e_t) and its derivative (Δe_t) are fed into a fuzzy inference system. Triangular membership functions define linguistic variables such as “Negative Large,” “Zero,” and “Positive Small.” A rule base of nine IF‑THEN statements maps the error conditions to desired voltage‑vector magnitude and angle adjustments. The fuzzy output is not a discrete vector; instead, it determines a reference voltage vector that is synthesized by SVM as a weighted combination of the two adjacent active vectors. This continuous synthesis eliminates the abrupt switching inherent in conventional DTC, thereby smoothing the torque waveform.

The authors validate the approach using a MATLAB/Simulink model of a 4 kW, 1500 rpm three‑phase induction motor. Three operating scenarios are examined: (1) steady‑state speed at rated value, (2) variable load from 0 to 5 Nm, and (3) rapid acceleration/deceleration from standstill to rated speed and torque. For each case, the conventional 6‑vector DTC is compared with the fuzzy‑SVM DTC.

In steady‑state, the conventional DTC exhibits a torque ripple of approximately 0.12 Nm, whereas the fuzzy‑SVM scheme reduces it to about 0.065 Nm—a 45 % improvement. Speed overshoot drops from roughly 6 % to 3.8 %. Under variable‑load conditions, the conventional controller’s torque error spikes to 0.25 Nm during sudden load steps, while the fuzzy‑SVM controller limits the error to 0.13 Nm, demonstrating superior disturbance rejection. During rapid acceleration, the fuzzy‑SVM controller produces smoother torque and speed trajectories, curtails current peaks by roughly 15 %, and cuts overall power loss by about 12 % compared with the classic DTC.

A sensitivity analysis explores the impact of rule‑base size and membership‑function width on performance and computational load. Increasing the rule count from 7 to 13 yields only marginal ripple reduction (≈1 %) but raises the execution time linearly, indicating that a nine‑rule configuration offers the best trade‑off for real‑time deployment. Widening the membership functions improves robustness to measurement noise but can slow the controller’s response if made excessively broad.

The study concludes that fuzzy‑logic‑driven SVM within a DTC framework effectively transforms the inherently discrete voltage‑vector selection into a continuous, error‑adaptive process, thereby mitigating torque ripple without sacrificing the fast dynamic response that makes DTC attractive. The resulting controller retains a relatively simple architecture, making it suitable for industrial inverters, electric‑vehicle drives, and robotic actuators where precise torque control is critical. Future work is suggested to implement the algorithm on digital signal processors or FPGA platforms for hardware‑in‑the‑loop testing and to extend the methodology to high‑power, high‑speed motor applications.