Research on Fuzzy Control System of Brushless DC Motor Based on DSP

0 Introduction

As a new type of continuously variable speed motor, the brushless DC motor not only has the characteristics of small size, light weight and small inertia of AC motor, but also has the excellent speed regulation performance of DC motor, but without the disadvantages of mechanical commutator, so it has been widely used. It has been increasingly valued in the manufacturing and processing fields such as CNC machine tools and robots, as well as household appliances such as washing machines and computer hard disks. In the past, brushless DC motors were mostly composed of single-chip microcomputers with many kinds of interface devices attached. Not only is it complex, but the speed is also limited, making it difficult to achieve full digital control from position loop to speed and current loop, and it is not convenient to expand. However, the motor servo system implemented by the digital signal processor (DSP) can replace the single-chip microcomputer and various interfaces with only the chip DSP, and due to the fast computing power of the DSP chip, more complex and intelligent algorithms can be implemented; the system can be upgraded and expanded through a high-speed network interface; and full digital control of position, speed and current loops can be achieved.

This paper introduces the use of TI's TMS320IF2407DSP as a controller to form a servo control system for brushless DC motors. First, the principle of brushless DC motors is introduced, then the software and hardware design methods are explained, and finally the experimental conclusions are given.

1 Working principle of brushless DC motor

The brushless DC motor is basically a permanent magnet synchronous motor, with the three-phase stator winding passing through an AC square wave and the rotor being a permanent magnet. Excitation is provided by the permanent magnet of the rotor, and the AC in the three-phase stator winding generates a rotating magnetic field. The armature magnetic potential and the rotor magnetic potential work together to generate electromagnetic torque. Following the characteristics of a DC motor, if the two magnetic fields are always perpendicular, the electromagnetic torque generated is maximum. Since the rotor is rotating, the direction of its magnetic field is also rotating, so it is necessary to change the stator magnetic field to be basically perpendicular to the rotor magnetic field (that is, the torque angle is 90°) by controlling the power-on sequence of the three-phase stator. In fact, the stator commutation logic is to make its average torque angle 90°. First, the current position of the rotor should be known, and then the power-on sequence of the three-phase stator should be determined according to the commutation requirements. This is why a brushless DC motor requires a rotor position sensor. In this experiment, the rotor magnetic pole position is detected by a Hall element applied to the surface of the stator core.

The motor adopts Y-type connection and is driven by three pairs of bridge inverter circuits. It works in a two-phase conduction, three-phase and six-state mode. The three Hall elements give 60° electrical angle position information, that is, they differ by 120° and the pulse width is 180°. The combination of three Hall elements can give 6 states in one cycle (the other two states generally do not appear), that is, a different state is changed every 60°. According to the sensor state information, combined with the commutation logic control, the six power transistors of the PWM inverter module of the three-phase stator are turned on or off, which can meet the torque angle requirements and enable the rotor to continuously obtain stable electromagnetic torque. Since only two phases are turned on at any time, their currents are equal in magnitude and opposite in direction, so it can be considered that the effect is equivalent to DC current. On the whole, the stator current is a square wave. As long as the phase is changed in time according to the magnetic pole position of the rotor, the characteristics of this DC drive can be maintained. Because the commutation is completed by electronic circuits or software instead of brushes, it is called a brushless DC motor.

2 Composition of brushless DC motor experimental control system

TI's TMS320IF2407 is a digital signal processor specifically designed for motion control applications. It contains all the main functional modules required for motor control applications. It not only has a 16-bit fixed-point processor core, but more importantly, it integrates many commonly used interfaces for motor control into a DSP controller. For example, it has two event managers, including a timer and a PWM generator that can drive two motors, an encoder detection circuit that can directly connect to the motor encoder; a standard CAN field bus that can communicate with the outside world at high speed; synchronous and asynchronous serial ports SPI and SCI that can communicate with a variety of standard serial devices; universal bidirectional I/O channels and AD conversion interfaces that directly collect field data; these make the motor control system implemented with DSP simple and modular. The system hardware basically includes a DSP board with TMS320IF2407 as the processing core, a matching power driver board and a PM50 motor.

This system uses the SCI interface to communicate serially with the host PC; the AD conversion interface is used to measure the motor's phase currents ia and ib, and the PWM generator is used to generate the required PWM signals to drive the PWM inverter on the power module; a general timer is used to generate the cycle of current and speed control; the encoder is installed on the motor rotor to measure the motor's position and obtain the motor speed through differentiation.

The stator current detection is carried out by connecting a resistor in series on the lower bridge arm of the inverter. After the stator current is converted into the corresponding voltage with a gain of 0.395 V/A, it is sent to the AD interface of the DSP. Here, only the currents of phases a and b need to be detected, and the current of phase c can be obtained by ia+ih+ic=0. This current detection method is relatively simple, but it requires that the software must ensure that when the command of the PWM inverter is output, the current of the lower bridge arm of the PWM inverter is detected at the same time to ensure the correctness of the current detection.

The DSP main program uses a loop to continuously call the data recording module, the monitoring module for serial communication with the host, etc. During the execution of the main program, t1 interrupts occur continuously. The interrupt service program processes current reading and conversion, encoder reading, speed conversion, etc. More importantly, the calculation of the current control and speed control loops must be completed. Both the current controller and the speed controller use PI control. The brushless DC motor only needs one current regulator like the DC motor, instead of two current regulators like the sine wave permanent magnet synchronous motor. The voltage commutation module completed by the software realizes the calculation of the phase voltage reference value applied to the inverter. In fact, the DSP controller accepts a three-phase reference voltage, and 6 full-comparison PWM outputs the square wave pulses required by the inverter module. At a given position, only two phases are turned on, and only four transistors of the inverter need to be controlled. The system has three closed loops. In actual control, the cycle of the outer loop position and speed control is 1 ms, while the cycle of the inner loop current control is 0.1 ms. This is because the current in the inner loop changes quickly, and a shorter control cycle can reduce torque fluctuations.

3 Fuzzy control method

The position servo system requires fast, accurate, and no overshoot, and conventional PID control is difficult to meet the above control requirements. In particular, some non-deterministic factors in the system, such as the time-varying model and the nonlinearity of the object, make the controller have strong robustness. The fuzzy method does not rely on the object model, has good adaptability, and can use more complex and intelligent control methods. Therefore, fuzzy logic is used for the position controller here, while the speed and current controllers still use PID control. Here, the position error e and the error change ec are used as the input of the position controller, and the output is the speed command value. According to the fuzzy control theory, the input and output are divided into 7 fuzzy subsets, namely nl (negative large), nm (negative medium), ns (negative small), ze (zero), ps (positive small), pm (positive medium), and pl (positive large). For simplicity, the membership function of the input uses a trigonometric function, while the membership of the output is a single-valued function.

During fuzzy reasoning, the input variables are first fuzzified according to the membership function form, and then the adaptability of each rule is calculated by the fuzzy intersection operation of the rule antecedents, and then the fuzzified values ​​of each subset of the output quantity are obtained according to the rule consequences. Since the output membership function is a single-valued function, defuzzification is to find the center of gravity of the fuzzy subset of the output quantity. These complex calculations are written in C language on the PC, and then linked with the current control, PWM output and other modules implemented in assembly language to form a DSP executable file. Finally, it is downloaded to the DSP board through the PC serial port.

According to the above principles and control methods, actual experiments are carried out. The experimental device consists of a permanent magnet synchronous motor, a power module, and a DSP board. The servo motor with a 500-line encoder is used to provide the motor position, and this system differentiates it to obtain speed information. The basic parameters of the motor are: phase resistance 5.25ω, phase inductance 0.46 mh, back electromotive force constant 2.62 V/l 000 r·min-1, rated voltage 19.1 V, rated current 1.16 A, rotor inertia 9×10 kg·m2.

In fact, in fuzzy control, the division of fuzzy subsets is a relatively difficult task. Because in the entire control process, the error and the change value of the error have a certain domain, which is called the domain of the variable. The initial domain is (-5000, 5000). As the error decreases, its possible value range becomes smaller and smaller. At this time, if the original domain is still used for reasoning, although the error can eventually approach zero, a small error will enter the convergence period too early in this case, which may lead to a relatively large positioning error. Therefore, when implementing fuzzy control, we constantly change the domain of the variable according to the actual control process.

4 Conclusion

The superior performance of brushless DC motors has made them widely used. Using DSP to control brushless DC motors is not only cheaper than traditional analog circuits, but also simpler in structure and easier to expand. The fast computing power of DSP can also realize more complex control algorithms, and both the speed loop and the current loop can be realized in digital form to form a fully digital brushless DC motor control system. This paper uses DSP to realize fuzzy control of brushless DC motors. The experiment shows that the position controller of fuzzy control has better positioning accuracy and fast response capability than PID control, especially the fuzzy algorithm with a changing domain can obtain better control performance.