Differences Between Vector Control and Direct Torque Control
Vector Control (VC) and Direct Torque Control (DTC) are two core control strategies in the field of high-performance speed regulation for AC motors—especially induction motors and permanent magnet synchronous motors. Both take “decoupling control” as their core goal, but they differ significantly in control philosophy, implementation approach, and performance characteristics. Below is a systematic comparison of their differences across dimensions such as core principle, control logic, key components, and performance indicators, along with additional recommendations for applicable scenarios.
1. Core Principle: “Indirect Decoupling” vs. “Direct Control”
The fundamental difference between the two lies in their control logic for the motor’s core physical quantities (flux linkage and torque), as detailed in the table below:
| Control Strategy | Core Philosophy | Decoupling Method | Implementation Path for Control Objectives |
|---|---|---|---|
| Vector Control (VC) | It emulates the control logic of DC motors: the stator current of an AC motor is decomposed into an “excitation component (i_d)” and a “torque component (i_q)”. Decoupling of torque and flux linkage is achieved by independently controlling these two orthogonal components. | Indirect Decoupling: Through “coordinate transformations” (e.g., Clark transformation, Park transformation), three-phase AC quantities are converted into DC quantities in a synchronous rotating coordinate system. PID regulation is then applied to these DC quantities to control torque indirectly. | 1. Detect the motor’s current and speed; 2. Obtain i_d and i_q via coordinate transformation; 3. Perform closed-loop control on i_d (to track the specified flux linkage value) and i_q (to track the specified torque value) separately; 4. Generate inverter switching signals through inverse transformation. |
| Direct Torque Control (DTC) | It controls the “stator flux linkage” and “electromagnetic torque” of the motor directly in the stationary coordinate system without the need for coordinate transformation. By judging real-time deviations of flux linkage and torque and selecting the optimal voltage vector, deviations are corrected quickly. | Direct Control: It does not rely on DC quantity decoupling. Instead, a “flux linkage observer” and a “torque observer” are used to calculate the current flux linkage and torque directly. After comparing these values with the specified ones, the inverter’s switching state is selected via a “switching table”. | 1. Detect the motor’s voltage and current; 2. Calculate the actual stator flux linkage and torque using observers; 3. Compare these values with the specified ones to obtain the flux linkage deviation (ΔΨ) and torque deviation (ΔT); 4. Based on the deviations and flux linkage position, query the switching table to output inverter control signals. |
2. Key Technical Differences: From Components to Implementation
2.1 Dependence on Coordinate Transformation
-
Vector Control (VC): It must rely on coordinate transformation.
Its core function is to convert the nonlinear, strongly coupled model of an AC motor into an “excitation-torque” independent model (similar to that of a DC motor) through Clark transformation (three-phase → two-phase stationary) and Park transformation (two-phase stationary → two-phase rotating). Therefore, the accuracy of coordinate transformation directly impacts control performance. -
Direct Torque Control (DTC): It does not require coordinate transformation at all.
It calculates flux linkage and torque directly in the α-β stationary coordinate system, avoiding calculation delays and errors caused by coordinate transformation (e.g., angle errors in Park transformation can lead to incomplete decoupling). This results in a more concise control logic.
2.2 Control Methods for Flux Linkage and Torque
| Dimension | Vector Control (VC) | Direct Torque Control (DTC) |
|---|---|---|
| Flux Linkage Control | Indirect control: The excitation current (i_d) is adjusted to make the rotor flux linkage (for induction motors) or stator flux linkage (for synchronous motors) track the specified value. This is an indirect control method that follows the “current closed-loop → flux linkage closed-loop” sequence. | Direct control: The amplitude and position of the stator flux linkage are calculated using an observer. After comparing these with the specified value, the optimal voltage vector is selected directly to adjust the flux linkage. This is a direct control method based on a “flux linkage closed-loop”. |
| Torque Control | Indirect control: The electromagnetic torque is controlled indirectly by adjusting the torque current (i_q) (torque is proportional to i_q). This process is limited by the adjustment delay of the current loop. | Direct control: The actual torque is calculated using an observer. After comparing this with the specified value, the switching table is used directly to select the voltage vector that changes the torque—resulting in a faster response speed. |
| Closed-Loop Hierarchy | Multi-closed-loop structure: It typically includes a “current loop (inner loop) → speed loop (outer loop)”; in some cases, a flux linkage loop is also added. It has multiple closed-loop levels and a long adjustment chain. | Dual-closed-loop structure: It only consists of a “flux linkage loop + torque loop (inner loop)” and a “speed loop (outer loop)”. It has fewer closed-loop levels and a short adjustment chain. |
2.3 Inverter Switching Signal Generation Method
-
Vector Control (VC): It relies on Pulse Width Modulation (PWM).
After coordinate transformation and PID regulation, the specified voltage values (u_d*, u_q*) in the rotating coordinate system are obtained. Then, either Space Vector Pulse Width Modulation (SVPWM) or Sinusoidal Pulse Width Modulation (SPWM) is used to generate trigger signals for the 6 switching tubes of the inverter. The switching frequency is fixed (usually 2–20 kHz). -
Direct Torque Control (DTC): It relies on a Switching Table.
No PWM is required; instead, the corresponding inverter switching state (there are 8 basic voltage vectors in total) is selected directly from a preset “switching table”. This selection is based on two core deviations (flux linkage deviation ΔΨ: usually classified into three levels: “+”, “0”, “-“; torque deviation ΔT: usually classified into three levels: “+”, “0”, “-“) and the sector where the stator flux linkage is located (there are 6 sectors in total). The switching frequency is not fixed (it changes dynamically with load and speed).
2.4 Observer and Parameter Sensitivity
-
Vector Control (VC):
It requires a “rotor flux linkage observer” (especially for induction motors). This observer typically calculates the rotor flux linkage position based on the motor’s mathematical model (e.g., voltage model, current model) to achieve angle synchronization in the Park transformation. Therefore, it is highly sensitive to motor parameters (such as rotor resistance r_r and inductance L_d/L_q): when the motor heats up and causes r_r to change, the observer’s accuracy decreases, leading to decoupling failure and torque fluctuation. -
Direct Torque Control (DTC):
It requires a “stator flux linkage observer” and a “torque observer”. These observers typically calculate values based on the instantaneous values of stator voltage and current (e.g., Ψ_α = ∫(u_α – i_α r_s)dt). It is sensitive to stator resistance r_s but not to rotor parameters (e.g., r_r) (since it does not involve rotor flux linkage calculation). However, at low speeds, the voltage drop (i_α r_s) accounts for a large proportion of the total voltage, which increases flux linkage observation errors—requiring a low-speed compensation strategy.
3. Performance Indicator Comparison: From Response to Stability
| Performance Indicator | Vector Control (VC) | Direct Torque Control (DTC) |
|---|---|---|
| Dynamic Response Speed | Slow: The current loop has an adjustment delay (usually tens of milliseconds), and torque response is constrained by this current loop delay. | Extremely fast: The torque response delay is only 1–2 switching cycles (usually a few milliseconds), making it suitable for scenarios that require fast dynamics (e.g., elevator start-stop, servo positioning). |
| Steady-State Torque Ripple | Small: With a fixed switching frequency and SVPWM modulation, the current waveform is close to a sine wave, and torque ripple is minimal (usually < 5%). | Relatively large: Due to the non-fixed switching frequency and “bang-bang” control of voltage vectors (only the optimal vector is selected), torque ripple is noticeable at low speeds (usually 5%–15%). Special optimization (e.g., virtual vectors, Model Predictive DTC) is required to mitigate this. |
| Low-Speed Performance | Good: Through optimizations such as rotor resistance identification and field weakening control, the current remains stable at low speeds, and torque ripple is small. | Poor: At low speeds, the stator resistance voltage drop accounts for a large proportion of the total voltage, leading to significant flux linkage observation errors and a higher risk of “flux linkage drift”—which increases torque ripple. |
| Parameter Robustness | Poor: It is sensitive to parameters like rotor resistance and inductance; changes in these parameters can easily degrade control performance. | Good: It is only sensitive to stator resistance and not to rotor parameters, giving it stronger adaptability to motor aging and load changes. |
| Computational Complexity | High: It requires multiple coordinate transformations, PID regulation, and SVPWM modulation, which places high demands on the controller’s computing power. | Low: It does not require coordinate transformation—only flux linkage/torque observation and switching table queries—resulting in a small computational load. This makes it suitable for low-cost controllers. |
4. Summary of Applicable Scenarios
| Control Strategy | Advantageous Scenarios | Disadvantageous Scenarios | Typical Application Cases |
|---|---|---|---|
| Vector Control (VC) | Scenarios requiring high steady-state accuracy and low-speed performance; scenarios that demand smooth torque output. | Scenarios requiring extremely high dynamic response; scenarios where motor parameters change frequently. | CNC machine tool spindles, precision servo systems, electric vehicle drives (partial), large wind turbines. |
| Direct Torque Control (DTC) | Scenarios requiring extremely high dynamic response; scenarios where motor parameters change frequently; scenarios using low-cost controllers. | Scenarios requiring low low-speed torque ripple and high steady-state accuracy. | Elevator traction machines, port cranes, general-purpose frequency converters, hybrid electric vehicles. |
5. Extension: Modern Improvement Directions
With the development of control theory, the boundary between the two strategies has gradually blurred, and integrated technologies have emerged:
- Improvements to Vector Control: Model Predictive Control (MPC) is introduced to replace PID, enhancing dynamic response; parameter adaptive algorithms are added to improve robustness.
- Improvements to Direct Torque Control: “Model Predictive Direct Torque Control (MP-DTC)” is adopted to replace the switching table, reducing torque ripple; virtual voltage vectors are introduced to optimize the switching frequency.
In conclusion, the choice between VC and DTC depends on the application’s priority trade-offs between “dynamic response”, “steady-state accuracy”, “cost”, and “parameter robustness”—rather than an absolute distinction between advantages and disadvantages.