Sensorless Direct Torque Control System and Control Method for Permanent Magnet Switched Flux Motor

Abstract

A position sensorless direct torque control system and control method for a permanent magnet switch flux motor relates to the technical field of direct torque control of a permanent magnet switch flux motor. The extended back-EMF module of the present invention detects whether the current speed of the motor is greater than or equal to a set speed threshold. If so, the amplitude of the extended back-EMF is obtained by the extended back-EMF model; if not, a high-frequency signal is injected into the extended back-EMF model until the motor speed is greater than or equal to the set speed threshold, and the amplitude of the extended back-EMF is obtained by the extended back-EMF model; the secondary extended back-EMF module obtains the secondary extended back-EMF according to the extended back-EMF amplitude obtained by the extended back-EMF module; the q-axis component of the reference voltage is obtained by a torque sliding mode controller, and the d-axis component of the reference voltage is obtained by a flux sliding mode controller. The present invention effectively suppresses torque pulsation, realizes full-speed universal position estimation, and improves estimation accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of direct torque control of a permanent magnet switch flux motor, and in particular to a position sensorless direct torque control system and a control method for a permanent magnet switch flux motor. Background Art

[0002] The permanent magnets of traditional permanent magnet synchronous motors are assembled on the rotor, so there are problems such as the need to fix the permanent magnets and irreversible demagnetization caused by heat dissipation or failure. The stator permanent magnet type PMFSM has the advantages of simple rotor structure and high power density.

[0003] Direct torque control and vector control are both classic brushless AC motor control strategies, and are AC speed regulation technologies with high dynamic performance. The structure of direct torque control can be divided into a torque loop and a flux loop. The measured speed is compared with the given speed, and the difference is passed through the PI module to obtain the reference torque value. The difference with the corresponding given value can be obtained by calculating the flux value and torque value, and then sent to the hysteresis loop to select the switch vector to control the torque and flux of the motor. The traditional DTC control strategy adopts Bang-Bang control. According to the hysteresis output of torque and flux, the switch table is directly selected to realize motor control. The structure is simple, but it also makes it difficult to implement the non-position injection technology at low speed. When the hysteresis output reaches the set upper and lower limits, the flux or torque will be adjusted to increase or decrease. This also means that the torque and flux will inevitably fluctuate back and forth within the set tolerance value range. In addition, since the voltage vector selected according to the hysteresis output in the corresponding sector is only one of the six valid vector states, these six voltage vectors are likely not the best choice. These factors lead to the problem of large torque pulsation in traditional direct torque control.

[0004] The acquisition of rotor position in the control system is an important component, but traditional position detection requires external hardware such as photoelectric encoders, Hall or rotary transformers for acquisition and detection, and mechanical detection methods have certain unreliability. Traditional sensorless position estimation within the full-speed range requires switching algorithms at zero low speed and medium and high speeds. This requires not only the construction of position estimation models at low speed and high speed, but also the design of a transition algorithm during switching so that the position estimation module can smoothly transition within the full-speed range. Therefore, traditional full-speed sensorless control is a composite control based on a hybrid algorithm, which makes the overall structure complex.

[0005] Based on this, technicians in this field are committed to providing a direct torque control system and method for a permanent magnet switched flux motor, which not only needs to suppress the torque pulsation problem, but also needs to solve the transition control problem in the full speed range of the motor. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a position sensorless direct torque control system and control method for a permanent magnet switched flux motor, introduces a sliding mode controller in the torque control link and the flux control link, and effectively suppresses torque pulsation; introduces a quadratic extended back-EMF model in the position estimation link, realizes full-speed domain rotor position estimation, and improves estimation accuracy.

[0007] To achieve the above objectives, the present invention provides, on one hand, a position sensorless direct torque control system for a permanent magnet switched flux motor, comprising a speed position estimation unit and a sliding mode control unit;

[0008] The speed and position estimation unit comprises:

[0009] The extended back-EMF module is used to detect whether the current speed of the motor is greater than or equal to the set speed threshold. If so, the amplitude of the extended back-EMF is obtained by the extended back-EMF model; if not, a high-frequency signal is injected into the extended back-EMF model until the motor speed is greater than or equal to the set speed threshold, and the amplitude of the extended back-EMF is obtained by the extended back-EMF model;

[0010] The secondary extended back electromotive force module is used to obtain the secondary extended back electromotive force according to the extended back electromotive force amplitude obtained by the extended back electromotive force module.

[0011] A phase-locked loop module is used to obtain the speed and position of the rotor based on the secondary expanded back electromotive force;

[0012] The sliding mode control unit comprises:

[0013] Torque sliding mode controller to obtain reference torque according to rotor speed and the motor torque Obtain the q-axis component of the reference voltage vector;

[0014] The flux sliding mode controller is used to adjust the reference flux according to the and the magnetic flux of the motor The d-axis component of the reference voltage vector is obtained.

[0015] Furthermore, the torque sliding mode controller is:

[0016] ;

[0017] in, is the torque loop slip variable, .

[0018] Furthermore, the flux sliding mode controller is:

[0019] ;

[0020] in, is the magnetic link sliding variable, .

[0021] A second aspect of the present invention provides a position sensorless direct torque control method for a permanent magnet switch flux motor, which is implemented based on the position sensorless direct torque control system of the permanent magnet switch flux motor, and includes the following steps:

[0022] S1, detecting whether the current speed of the motor is greater than or equal to the set speed threshold, if so, obtaining the amplitude of the extended back electromotive force from the extended back electromotive force model; if not, injecting a high-frequency signal into the extended back electromotive force model until the motor speed is greater than or equal to the set speed threshold, and obtaining the amplitude of the extended back electromotive force from the extended back electromotive force model;

[0023] S2, obtaining a secondary extended back electromotive force according to the extended back electromotive force amplitude obtained in step S1, and obtaining the speed and position of the rotor through a phase-locked loop;

[0024] S3, obtaining a reference torque according to the rotor speed obtained in step S2 , reference torque The torque of the motor The q-axis component of the reference voltage vector is obtained from the input torque sliding mode observer, and the reference flux is and the magnetic flux of the motor The d-axis component of the reference voltage vector is obtained by inputting the flux linkage sliding mode observer;

[0025] S4. Obtain a modulation signal according to the q-axis component of the reference voltage vector and the d-axis component of the reference voltage vector.

[0026] Furthermore, after injecting a high-frequency signal into the extended back-EMF model, the amplitude of the extended back-EMF is obtained as follows:

[0027] ;

[0028] in, is the frequency of the injected signal, n represents the harmonic order, the superscript "f" represents the fundamental component, the motor is a double-salient pole structure, .

[0029] Further, the secondary expansion of the anti-electromotive force and for:

[0030] ;

[0031] in, , is the magnitude of the secondary expanded back EMF.

[0032] Furthermore, the secondary extended back electromotive force uses the normalization method to extract the sine and cosine quantities containing position information, and the result is:

[0033] .

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. The present invention realizes full-speed range position sensorless control with the same model in both low-speed and high-speed states through a secondary extended back-electromotive force model, injects high-frequency quantities at low speeds, reduces the use of filters, and reduces the estimation error caused by the phase delay caused by the filters. When the speed increases, the injection can be stopped more conveniently and the back-electromotive force observation can be switched to, thus solving the transition problem. There is no need to design low-speed, high-speed and transition algorithms at the same time, thus simplifying the system algorithm structure.

[0036] 2. The present invention adopts a second-order sliding mode controller to replace the hysteresis loop of the traditional direct torque, which effectively reduces the torque pulsation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the overall structure of a direct torque control system according to a specific embodiment of the present invention;

[0038] Figure 2 It is a block diagram of the position and speed estimation principle of a specific embodiment of the present invention;

[0039] Figure 3 It is a schematic diagram of the overall flow of a direct torque control method according to a specific embodiment of the present invention;

[0040] Figure 4 It is a schematic diagram of a torque circuit diagram of a specific embodiment of the present invention;

[0041] Figure 5 A schematic diagram of a high frequency circuit according to a specific embodiment of the present invention;

[0042] Figure 6 A schematic diagram of voltage vector synthesis according to a specific embodiment of the present invention;

[0043] Figure 7 is a simulation waveform of high frequency injection at low speed in a specific embodiment of the present invention, Figure 7 a is the waveform of the speed mutation, Figure 7 b is the speed error waveform, Figure 7 c is the estimated position waveform, Figure 7 d is the position error waveform;

[0044] Figure 8 The extended back-EMF simulation waveform at low speed of a specific embodiment of the present invention;

[0045] Fig. 9 : is a simulation waveform based on an extended back electromotive force model at high speed according to a specific embodiment of the present invention. Fig. 9 a is the extended back electromotive force waveform at high speed, Fig. 9 b is the position simulation waveform, Fig. 9 c is the position error waveform;

[0046] Fig.10 A weighted transition estimated speed simulation waveform of a transition zone according to a specific embodiment of the present invention;

[0047] Fig.11 is a weighted transition position simulation waveform of a transition zone of a specific embodiment of the present invention, Fig.11 a is the position signal waveform, Fig.11 b is the position error waveform;

[0048] Fig.12 It is a locally amplified waveform of a weighted transition position error in a transition zone of a specific embodiment of the present invention;

[0049] Fig.13 This is a simulation waveform based on a quadratic extended back-EMF model according to a specific embodiment of the present invention. Fig.13 a is the speed curve, Fig.13 b is the actual position curve, Fig.13 c is the estimated position curve, Fig.13 d is the position error curve;

[0050] Fig.14 The back-EMF waveform under low-speed and high-frequency injection in the secondary expansion method of a specific embodiment of the present invention;

[0051] Fig.15 It is a weighted transition position error curve at a speed of 0-500r / min in a specific embodiment of the present invention;

[0052] Fig.16 It is a position error curve of the secondary extended back-EMF model at a speed of 0-500r / min in a specific embodiment of the present invention;

[0053] Fig.17 is a speed curve under no-load in a specific embodiment of the present invention, Fig.17 a is the traditional DTC speed curve, Fig.17 b is the second-order sliding mode DTC speed curve of the present invention;

[0054] Fig.18 is a torque curve in a no-load steady state in a specific embodiment of the present invention, Fig.18 a is the traditional DTC torque curve, Fig.18 b is the second-order sliding mode DTC torque curve of the present invention;

[0055] Fig.19This is a load torque curve of 5 N·m at a speed of 100 r / min in a specific embodiment of the present invention. Fig.19 a is the traditional DTC torque curve, Fig.19 b is the second-order sliding mode DTC torque curve of the present invention;

[0056] Fig. 20 This is the current waveform of a specific embodiment of the present invention at a speed of 100 r / min and a load of 5 N·m. Fig. 20 a is the traditional DTC current waveform, Fig. 20 b is the second-order sliding mode DTC current waveform of the present invention;

[0057] Fig.21 This is a load torque curve of 20 N·m at a speed of 100 r / min in a specific embodiment of the present invention. Fig.21 a is the traditional DTC torque curve, Fig.21 b is the second-order sliding mode DTC torque curve of the present invention;

[0058] Fig. 22 This is the current waveform of a specific embodiment of the present invention when the speed is 100r / min and the load is 20N·m. Fig. 22 a is the traditional DTC current waveform, Fig. 22 b is the second-order sliding mode DTC current waveform of the present invention;

[0059] Fig.23 This is a load torque curve of 5 N·m at a speed of 500 r / min in a specific embodiment of the present invention. Fig.23 a is the traditional DTC torque curve, Fig.23 b is the second-order sliding mode DTC torque curve of the present invention;

[0060] Fig.24 This is a current waveform of a load of 5 N·m at a speed of 500 r / min in a specific embodiment of the present invention. Fig.24 a is the traditional DTC current waveform, Fig.24 b is the second-order sliding mode DTC current waveform of the present invention;

[0061] Fig.25 This is a load torque curve of 20 N·m at a speed of 500 r / min in a specific embodiment of the present invention. Fig.25 a is the traditional DTC torque curve, Fig.25 b is the second-order sliding mode DTC torque curve of the present invention;

[0062] Fig.26 This is the current waveform under a load of 20 N·m in a specific embodiment of the present invention. Fig.26 a is the traditional DTC current waveform, Fig.26 b is the second-order sliding mode DTC current waveform of the present invention;

[0063] Fig. 27 is a speed curve under a torque mutation in a specific embodiment of the present invention, Fig. 27 a is the traditional DTC speed curve, Fig. 27 b is the second-order sliding mode DTC speed curve of the present invention;

[0064] Fig.28 is a torque curve under a torque mutation in a specific embodiment of the present invention, Fig.28 a is the traditional DTC torque mutation curve, Fig.28 b is the second-order sliding mode DTC torque mutation curve of the present invention. DETAILED DESCRIPTION

[0065] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0066] like Figure 1 As shown, in a specific embodiment, a position sensorless direct torque control system of a permanent magnet switched flux motor includes a speed position estimation unit, a sliding mode observation unit and a SVPWM modulation unit;

[0067] like Figure 2 As shown, the speed position estimation unit includes an extended back electromotive force module, a secondary extended back electromotive force module and a phase-locked loop module.

[0068] The speed sensorless control method in the prior art is usually divided into two parts: low-speed control and high-speed control. This is because at high speeds, the speed is generally estimated and calculated through the related quantities generated by fundamental wave excitation, but at low speeds, it is difficult to distinguish between the required signal and noise, so it is difficult to apply; similarly, high-frequency injection can be used at low speeds to obtain accurate position estimation, but as the speed increases, the estimation error of the high-frequency injection method will become larger and larger, so it cannot be applied to high-speed conditions. Therefore, the sensorless control method for the full speed domain generally adopts the high-frequency injection method for low-speed applications, and various methods based on fundamental wave excitation are used for estimation at high speeds. This means that two estimation models need to be used and switched, usually by a linear weighted method to transition between the two algorithms. To this end, in a specific embodiment, speed and position estimation for the full speed domain is achieved by a speed and position estimation unit.

[0069] The extended back-EMF module is used to detect whether the current speed of the motor is greater than or equal to a set speed threshold. If so, the amplitude of the extended back-EMF is obtained by the extended back-EMF model; if not, a high-frequency signal is injected into the extended back-EMF model until the motor speed is greater than or equal to the set speed threshold, and the amplitude of the extended back-EMF is obtained by the extended back-EMF model;

[0070] The voltage equation of the motor in the stationary coordinate system is:

[0071] ;

[0072] R is the stator resistance, , For stator winding , Shaft inductance, , is the stator current , Axis component, is the rotation speed.

[0073] , To extend the back EMF The components under the axis system are:

[0074] ;

[0075] in, , are the d and q axis components of the stator current, is the rotor position angle, is the permanent magnet flux of the rotor, and p is the differential operator.

[0076] The amplitude of the extended back EMF under the high-speed operation condition of the motor is obtained as:

[0077] .

[0078] The extended back electromotive force at high speed is expressed as:

[0079] ;

[0080] In a specific embodiment, when the current speed of the motor is less than the set speed threshold, the motor is operating in a low-speed condition. In the low-speed condition, high-frequency excitation is injected into the q-axis so that the q-axis current becomes a value containing high-frequency quantities to increase the extended back-EMF amplitude. When the motor is operating in a low-speed condition, the high-frequency current is added to the extended back-EMF model, and the amplitude of the extended back-EMF is obtained as follows:

[0081] ;

[0082] in, is the frequency of the injected signal, n represents the harmonic order, the superscript f represents the fundamental component, and the motor is a double-salient pole structure. .

[0083] The secondary extended back electromotive force module is used to obtain the secondary extended back electromotive force according to the extended back electromotive force amplitude obtained by the extended back electromotive force module. In a specific embodiment, it is used to further expand the extended back electromotive force into a new secondary extended back electromotive force model containing position information, which is:

[0084] ;

[0085] in, , is the magnitude of the secondary extended back EMF;

[0086] The main noise source in motor drives is voltage distortion caused by inverter nonlinearity. It is approximately proportional to the rotor speed, so at high speeds, observability is evident, while if HF injection is used, observability at low speeds is also guaranteed.

[0087] After observing the secondary extended back EMF, a normalization method is used to remove The size of , and the sine and cosine quantities with position information are retained and extracted, as shown in the following formula:

[0088] ;

[0089] The phase-locked loop module obtains the speed and position of the rotor according to the secondary expanded back electromotive force;

[0090] The transfer function of the phase-locked loop is:

[0091] ;

[0092] in, , Determines the bandwidth of the PI and .

[0093] The sliding mode control unit includes a torque sliding mode controller and a flux sliding mode controller. The sliding mode controller is used to replace the hysteresis loop in the traditional direct torque control. The flux and torque errors are input and the voltage vector under the AC and DC axes is output through the sliding mode controller.

[0094] The torque sliding mode controller is used to obtain a reference torque according to the rotor speed. and the motor torque Obtain the q-axis component of the reference voltage vector;

[0095] The flux sliding mode controller is used to adjust the reference flux according to the reference flux. and the magnetic flux of the motor The d-axis component of the reference voltage vector is obtained.

[0096] Furthermore, the torque sliding mode controller is:

[0097] ;

[0098] in, is the torque loop slip variable, .

[0099] Furthermore, the flux sliding mode controller is:

[0100] ;

[0101] in, is the magnetic link sliding variable, .

[0102] The SVPWM modulation unit includes a sector judgment module, a vector action time module and a switch tube conduction time calculation module.

[0103] In a specific embodiment, a method for direct torque control of a permanent magnet switched flux motor without a position sensor is provided, which is implemented based on the direct torque control system of the permanent magnet switched flux motor without a position sensor, and includes the following steps:

[0104] S1, detecting whether the current speed of the motor is greater than or equal to the set speed threshold, if so, obtaining the amplitude of the extended back electromotive force from the extended back electromotive force model; if not, injecting a high-frequency signal into the extended back electromotive force model until the motor speed is greater than or equal to the set speed threshold, and obtaining the amplitude of the extended back electromotive force from the extended back electromotive force model;

[0105] The amplitude of the extended back EMF under high-speed operation of the motor is:

[0106]

[0107] When the motor is running at low speed, the amplitude of the extended back electromotive force is:

[0108] .

[0109] S2, obtaining a secondary extended back electromotive force according to the extended back electromotive force amplitude obtained in step S1, and obtaining the speed and position of the rotor through a phase-locked loop;

[0110] In a specific embodiment, the injected high-frequency signal is a sinusoidal voltage, and the total high-frequency voltage consists of two parts: the injected voltage The high-frequency voltage generated by the sliding mode controller flowing through the torque link and the structure diagram of the q-axis current loop are shown in the figure below. Figure 4 As shown in (4-7), due to the reference torque is generated by the speed PI regulator without HF component, so the q-axis torque loop is Figure 5 (4-8), where only HF signals are considered.

[0111] Depend on Figure 5 It is expressed as a transfer function:

[0112] ;

[0113] ;

[0114] in, is the transfer function of the torque sliding mode controller, .

[0115] When a high-frequency sinusoidal voltage is injected, the high-frequency voltage generated and the induced high-frequency current on the q-axis are obtained from the transfer function and can be expressed as:

[0116] ;

[0117] ;

[0118] For high frequency current The derivative is:

[0119] ;

[0120] Define the amplitude as , the phase is defined as It can be seen that the injection of high frequency will change the high frequency voltage when there is no filter effect. amplitude and phase.

[0121] The motor voltage equation under high frequency excitation is:

[0122] ;

[0123] According to the motor voltage equation under high-frequency excitation, the corresponding high-frequency current generated by the q-axis can be derived. When a high-frequency sinusoidal quantity is injected, the high-frequency current can be expressed as:

[0124] ;

[0125] ;

[0126] When the motor reaches high speed and runs steadily under this condition, the injection is removed, HF disappears, and the magnitude of the secondary extended back EMF at high speed is:

[0127] .

[0128] The main noise source in motor drives is voltage distortion caused by inverter nonlinearity. It is approximately proportional to the rotor speed, so at high speeds, observability is evident, while if HF injection is used, observability at low speeds is also guaranteed.

[0129] After observing the secondary extended back EMF, a normalization method is used to remove The size of , and the sine and cosine quantities with position information are retained and extracted, as shown in the following formula:

[0130] ;

[0131] The estimator based on the secondary extended back electromotive force in this embodiment can perform position estimation in the full speed range. At low speed, high-frequency voltage is injected on the q-axis, and high-frequency current can be induced on the q-axis. When the rotor speed increases, high-frequency injection is stopped to release the voltage margin, and position estimation can still be performed through the observer. In addition, the method does not need to use a digital filter in the feedback link of the flux and torque and the position estimator, which essentially avoids the bandwidth reduction and position error caused by the filter.

[0132] S3, obtaining a reference torque according to the rotor speed obtained in step S2 , reference torque The actual torque of the motor The q-axis component of the reference voltage vector is obtained from the input torque sliding mode observer, and the reference flux is and the actual flux linkage of the motor The d-axis component of the reference voltage vector is obtained by inputting the flux linkage sliding mode observer;

[0133] In this embodiment, the actual flux linkage and actual torque of the motor are respectively:

[0134] ;

[0135] ;

[0136] In the sliding mode observer of the flux linkage, it is hoped that the flux linkage will eventually be equal to the expected value, so the sliding variable is designed as Similarly, the sliding variable of the torque loop observer is designed as , and the gain , Meet stability requirements.

[0137] In this embodiment, the torque sliding mode controller is:

[0138] ;

[0139] In this embodiment, the flux sliding mode controller may be:

[0140] .

[0141] The q-axis component of the reference voltage vector is obtained by the torque sliding mode controller , the d-axis component of the reference voltage is obtained through the flux sliding mode controller .

[0142] S4. Obtain a modulation signal through SVPWM according to the q-axis component of the reference voltage vector and the d-axis component of the reference voltage vector.

[0143] After the reference voltage vector is determined in step S3, the sector in which it is located is determined, and the voltage vector to be used for synthesis is determined according to the sector in which it is located. Define three variables X1, X2, and X3, and let:

[0144] ;

[0145] when , let A=1; otherwise A=0.

[0146] when , let B=1; otherwise B=0.

[0147] when , let C=1; otherwise C=0.

[0148] Define variables based on A, B, C ,The corresponding relationship between variable N and sector is shown in Table 1.

[0149] Table 1 Relationship between sectors and variable N

[0150]

[0151] The target vector in a cycle is equivalent to the corresponding time of the two non-zero voltage vectors. Determining the sector also determines which two of the six basic non-zero vectors need to be used. In order to obtain the target vector, it is also necessary to determine the respective action time, such as Figure 6 (3-6) shown.

[0152] The role of the zero voltage vector is to compensate for the remaining time after removing the action time of two non-zero vectors in a cycle. The relationship between the action times is as follows, taking sector I as an example:

[0153] ;

[0154] The reference voltage vector is placed in the axis system Decomposed below, according to the mathematical relationship we can get:

[0155] ;

[0156] The magnitudes of the six non-zero reference voltage vectors are equal, all , so the solution of the simultaneous equations is:

[0157] ;

[0158] The sum of the action time of the non-zero vector should not be greater than the switching period , if it occurs, the action time should be redistributed proportionally.

[0159] ;

[0160] Defining variables , , They are:

[0161] ;

[0162] Then the action time of the two basic voltage vectors of the sector is set as , , which can be summarized in Table 2.

[0163] Table 2 Voltage vector action time table

[0164]

[0165] The conduction time of the switch tube can be obtained by the action time obtained above. Assuming the switch moments are respectively: , , :

[0166]

[0167] The conduction time of each phase switch is shown in Table 3.

[0168] Table 3 Sector modulation three-phase conduction time

[0169]

[0170] The second-order sliding mode control adopted in this embodiment has better control accuracy than the traditional hysteresis regulation, and combined with the SVPWM technology, it has better performance than the traditional DTC system. In addition, the design and application of the flux linkage and torque sliding mode controller make it possible to obtain the dq axis components of the reference voltage vector, which solves the input voltage vector of the SVPWM modulation module.

[0171] The direct torque control method of the present application is further verified by simulation experiments. In a specific embodiment, the low speed, high speed, weighted transition and quadratic extended back EMF methods are simulated and analyzed.

[0172] 1) Verification of high-frequency injection position estimation method at low speed:

[0173] In order to fully demonstrate the advantages and disadvantages of the high-frequency injection method, the high-frequency injection method is used alone for position estimation, and position estimation simulations are performed at speeds of 20-50r / min and 800r / min respectively.

[0174] Figure 7 a is the speed and position estimation waveform of the speed change at low speed. The motor first increases from zero speed to 20r / min, and then increases the speed to 50r / min after stabilization. It can be seen that the high-frequency injection method has a good position estimation tracking effect at low speed. The speed error is as follows Figure 7 b, position signal and position error such as Figure 7 c and Figure 7 d. Since there is a certain phase error in the rotor position estimation, the position signal is usually compared with The remainder is converted into a triangular wave observation, so when the remainder of one is already zero, the amplitude of the other remainder is approximately So the error graph will show a value of A brief pulse of time.

[0175] 2) Position estimation method based on extended back-EMF at high speed:

[0176] The above simulations have shown that the high-frequency injection method is good at low speeds, but the dynamic performance of the position estimation and system operation of this method at high speeds is not ideal, and the injection of high frequency will increase the burden on the inverter and increase the loss, and the overall efficiency of the system will decrease accordingly. The injected pulsating high-frequency voltage component will increase the modulation of the inverter and reduce the effective adjustment range of the fundamental component. At the same time, the high-frequency current response generated will further increase the loss of the inverter, making the efficiency of the entire drive system lower. Therefore, after the speed increases, the high-frequency injection method is abandoned and other methods are used for position detection to release the voltage margin and improve the output efficiency. In order to achieve sensorless control in the full speed range, in this embodiment, a position detection method based on an extended back electromotive force model is used at high speeds. When the speed is set to 10r / min, although there are estimated values ​​of speed and position in the simulation, the speed cannot track the given value, and the amplitude of the extended back electromotive force is very low. The position estimation effect is not ideal, and it is difficult to apply at low speeds in practice. At this time, the extended back electromotive force waveform is as follows, and the amplitude is only about 0.6V, such as Figure 8 shown.

[0177] To verify the position estimation of the PMFSM motor at high speed and the dynamic performance of the system, the system is simulated at 500r / min, and the speed is suddenly increased to 800r / min at 0.2s. The simulation results are shown in Fig. 9,Comparing the position and position error of the motor, the error of the position estimation is generally stable, indicating that the speed change process is relatively smooth, and a relatively ideal speed and position estimation can be achieved.,It can be seen that at medium and high speeds, the position estimation method based on the extended back-EMF can estimate the rotor position and speed more accurately, and the tracking effect is good.

[0178] 3) Simulation analysis of transition area:

[0179] In order to smoothly transition between high-frequency injection and back-EMF observation methods, a composite control simulation with weighted transition is first performed. The transition interval is set to 100r / min to 200r / min. When it is lower than 100r / min, the high-frequency injection method is used. In the transition zone, the weighted average method of the estimated values ​​of the two methods mentioned above is used for estimation. When it is higher than 200r / min, the back-EMF observation estimation is used. In order to better observe the transition zone, the speed is set to increase to 100r / min within 0.02s, run smoothly until the speed starts to increase at 0.2s, rise to 200r / min after 0.1s, and maintain 200r / min until the end of the simulation at 0.5s. The estimated speed is as follows Fig.10 As shown, the estimated position is Fig.11 shown.

[0180] In the transition zone, it can be seen that the speed fluctuation increases slightly during the transition process, and the position estimation error pulse in the transition zone is wider than that in the steady state, that is, the position estimation fluctuates compared with the steady state. When the position estimation coincides with the actual error, if there is no phase advance or lag, the amplitude of the position error in the corresponding period is The pulse will be zero.

[0181] In order to observe the rotor position error, the signal before taking the remainder of the rotor position is compared with the actual value, such as Fig.12 As shown in the figure, it can be seen that the position error does not change significantly during the speed increase, but in the enlarged figure, the position error can be found to increase first and then decrease in the transition zone. The maximum error amplitude is 0.2rad, about 11.5°. After the speed reaches the upper limit of the transition zone, the extended back-EMF estimation method is used, and the position error gradually converges and stabilizes. There is no sudden increase in error, and the transition is relatively smooth.

[0182] 4) Simulation analysis of secondary extended back EMF estimation method:

[0183] Under the same simulation conditions, the transition zone simulation of the position estimation method of the secondary extended back EMF described in the article is carried out, and the speed and position errors are as follows Fig.13As shown in the figure, the speed rises to 800r / min in 1s. The speed is still relatively stable in the transition zone, and the position tracking effect is good. The estimated position error does not have a large mutation during the simulation process, and can smoothly transition between low speed and medium and high speed stages. After reaching 200r / min, the method of canceling the injection position estimation is similar to the method of extending the back EMF, and the position error tends to be stable.

[0184] Different from the original extended back EMF, which is too low to be observed at low speed, the secondary extended back EMF estimation method introduces high-frequency injection at low speed, so that the low-speed back EMF can be observed, and its amplitude changes with time. When a 40V high-frequency voltage is injected, the following is generated: Fig.14 The back EMF waveform with an envelope is shown, and Figure 8 In comparison, the back EMF waveform changes more regularly and obviously. By squaring the extended back EMF to expand the observed value and then normalizing it, we can get an amplitude of 1. , and then obtain the rotor position information through a phase-locked loop.

[0185] In order to better compare the weighted transition method with the secondary extended back-EMF method, the two methods were simulated under the same conditions. To avoid errors caused by sudden speed changes, the speed was set to increase from 0r / min to 500r / min at a fixed slope within 1s. This process includes the low-speed stage, transition zone, and medium- and high-speed stages. In order to more intuitively observe the change in position error, the position error curve was enlarged for local observation. The position error is shown in Figure 2. Fig.15 As shown in the figure, the position estimation method of the quadratic extended back-EMF eliminates filters such as low-pass filtering and band-pass filtering during the low-speed injection process. Compared with the pulse injection, the phase delay caused by the filter is reduced, and the error is reduced accordingly, as shown in the figure. Fig.16 As shown, and because the same model is applied at low and high speeds, the transition is smoother than the weighted transition.

[0186] In a specific embodiment, a conventional direct torque control (DTC) system and the direct torque control system of the present application are compared, and simulation comparisons are performed under different loads, different speeds, and speed mutations and torque mutations. The conventional DTC inner loop adopts a hysteresis controller, and the torque hysteresis loop width is set to 0.1, and the flux hysteresis loop width is set to 0.002; the inner loop of the DTC system of the present invention adopts a second-order sliding mode control. In addition to the switching function and the square root of the absolute value of the input value, the controller also includes a proportional integral link, in which K P Set to 1, K I Set to 100.

[0187] 1) No-load start simulation experiment:

[0188] No-load startup waveform Fig.17As shown, it starts at a speed of 400r / min, and when it runs smoothly to 0.2s, the speed suddenly changes to 800r / min. It can be seen that under no-load conditions, the speed first rises rapidly and has an overshoot process during startup and speed mutation, but can respond quickly and reach the desired speed. Among them, the speed overshoot reaches 497.4r / min during the startup process in the traditional method, which is basically stable at 0.04s, and the overshoot reaches 897.65r / min when the speed suddenly changes, and reaches a stable 800r / min at 0.24s. For better comparison, the DTC under the sliding mode control of the present invention uses the same speed loop PI parameters as the traditional DTC control, and the overall trends of the two are basically the same. In contrast, the startup overshoot of the sliding mode control is 494.2r / min, which reaches stability at 0.0375s, and the overshoot reaches 891.93r / min when the speed suddenly changes, and reaches stability at 0.23s. Under the same speed PI parameters, the system with a sliding mode controller instead of a hysteresis loop has slightly improved performance compared to the traditional DTC under no-load.

[0189] The torque waveform of the motor in a stable state under no-load is as follows: Fig.18 As shown in the figure, it can be seen that the torque waveform in the traditional DTC control method fluctuates more than that in the sliding mode control, which is reflected in the curve graph as the difference in curve thickness. The traditional DTC torque fluctuates between -0.54N·m and +0.51N·m, and the second-order sliding mode DTC torque fluctuates between ±0.18N·m. Fig.17 , 18 It can be clearly seen that the no-load starting lower sliding mode DTC has relatively good dynamic and static performance.

[0190] (2) Comparative simulation of different load torques at low speed

[0191] The motor speed is set to 100r / min, and the motor is running with load. The torques of 5N·m and 20N·m are compared. Fig.19 The figure shows the torque comparison under steady-state load of 5N·m. The torque under the traditional method fluctuates around 5N·m by about 0.5N·m, while the torque under sliding mode control is around 4.69N·m-5.28N·m. Compared with the traditional DTC method, the torque fluctuation is reduced by nearly 40%.

[0192] At this time, the current waveform is as follows Fig. 20 As shown in the figure, due to the high number of motor pole pairs, the current waveform frequency is high. In order to better observe and compare the current waveform, only the current waveform during part of the stable time is intercepted. It can be seen that the sinusoidality of the current waveform of the two methods is relatively good, but the current fluctuation is more obvious under the traditional DTC. Since the traditional DTC inner loop adopts hysteresis control, the current fluctuation varies with the loop width. The larger the loop width, the greater the current fluctuation, and the smaller the loop width, the smaller the current fluctuation, but the device switching frequency will be higher.

[0193] Fig.21 The torque comparison chart under a load of 20 N·m shows that under 20 N·m, the sliding mode control still has a small torque pulsation. The torque of the traditional DTC fluctuates between 19.4 N·m and 20.5 N·m, and the torque fluctuation of the sliding mode control DTC is about 0.2 N·m. Fig.19 In comparison, under different loads, the torque fluctuation of traditional DTC is limited by the loop width, with a fluctuation amplitude of about ±0.5N·m, while the torque fluctuation amplitude of second-order sliding mode DTC is ±0.2N·m.

[0194] Current waveform Fig. 22 As shown in the figure, the current waveform amplitudes of the two methods under the same load are the same. The current fluctuation of the traditional method still exists. Since the overall amplitude has increased compared with 5N·m, the fluctuation is not as obvious as that of 5N·m. However, compared with the sliding mode control current waveform under 20N·m, it can still be seen that there is a relatively obvious fluctuation, which is the same as the comparison with the torque. Fig. 22 As shown in the figure, the current fluctuation amplitude of the traditional DTC under the two methods is significantly greater than that of the second-order sliding mode DTC. Fig. 20 In comparison, under different loads, the current fluctuation amplitude of the traditional DTC is also limited by the loop width and the variation is still large; the current fluctuation amplitude of the second-order sliding mode DTC is relatively stable.

[0195] 3) Comparative simulation of different load torques at high speed:

[0196] The steady-state torque comparison was carried out at 500r / min with loads of 5N·m and 20N·m. The simulation results are shown in Fig.23 As shown in the figure, the torque under traditional direct torque control fluctuates between 4.4N·m and 5.5N·m, while the torque under sliding mode control fluctuates between 4.7N·m and 5.27N·m. The fluctuation range of the torque is significantly reduced, with a reduction of nearly 50%.

[0197] The current waveform at 500r / min is as follows Fig.24 As shown, it can be seen that the frequency of the current waveform increases accordingly compared to 100r / min, and the amplitude remains unchanged. In the two methods, the traditional DTC current waveform fluctuates more obviously than the sliding mode control, which indirectly proves that the torque fluctuation effect has been improved.

[0198] When the load is set to 20N·m, the steady-state torque is Fig.25 As shown in the figure, the torque under traditional DTC fluctuates between 19.39 N·m and 20.51 N·m, while the torque under sliding mode control fluctuates by about 0.3 N·m. Under a load of 20 N·m, the torque fluctuation of sliding mode control is reduced by about 50% compared with that of traditional DTC.

[0199] Current waveform Fig.26As shown in the figure, compared with the current waveform frequency at 100r / min, the current waveform amplitude is basically the same when compared with the same load, and the current waveform under sliding mode control is smoother.

[0200] (4) Torque mutation simulation

[0201] In order to better compare the torque pulsation size and the dynamic performance of the control system, the waveforms under the two controls are set to the same coordinate size. It can be clearly seen from the steady-state torque waveform under no-load in the previous article that the torque fluctuation under sliding mode control is smaller, indicating that sliding mode control can indeed reduce the torque pulsation of DTC control.

[0202] At a given speed of 400r / min, the load is started with load and the load changes suddenly. The load is set to change suddenly from 5N·m to 20N·m in 0.2s. The speed waveform is as follows Fig. 27 As shown in the figure, when the load suddenly increases, the speed first decreases at the moment of the mutation and then recovers to the given speed. The speed overshoot of the traditional DTC reaches 483.6r / min during load startup, and it stabilizes in about 0.05s. During the torque mutation, the speed drops to a minimum of 327.52r / min and recovers to the given speed in 0.24s. Under sliding mode control, the overshoot of DTC during load startup reaches 480.9r / min, which is less than that of the traditional DTC, and it stabilizes in 0.04s. During the torque mutation, the speed drops to a minimum of 328.84r / min and recovers to the given speed in 0.235s. Under the same speed PI, the dynamic performance of the sliding mode control DTC during the load mutation is slightly improved compared with the traditional DTC.

[0203] It can be seen that when the motor is started with load, the speed still has a good response speed, and it can quickly return to stability when the load torque suddenly changes. Fig.28 As shown in the figure, the torque response speed is slightly improved under sliding mode control, the dynamic performance is good, and the torque fluctuation is always relatively low.

[0204] It can be seen that in the case of sudden load changes, the dynamic performance of the system is good, and the torque pulsation is significantly reduced during the overall simulation process. In addition, the introduction of sliding mode controller and SVPWM not only reduces torque pulsation, but also makes it possible to achieve sensorless control at low speed.

[0205] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A sensorless direct torque control system for a permanent magnet switched flux motor, characterized in that: It includes a speed and position estimation unit and a sliding mode control unit; The speed and position estimation unit comprises: The extended back-EMF module is used to detect whether the current speed of the motor is greater than or equal to the set speed threshold. If so, the amplitude of the extended back-EMF is obtained by the extended back-EMF model; if not, a high-frequency signal is injected into the extended back-EMF model until the motor speed is greater than or equal to the set speed threshold, and the amplitude of the extended back-EMF is obtained by the extended back-EMF model; The secondary extended back electromotive force module is used to obtain the secondary extended back electromotive force containing position information according to the extended back electromotive force amplitude obtained by the extended back electromotive force module and for: ； in, , is the magnitude of the secondary extended back EMF; , is the component of the extended back EMF under the αβ axis; is the rotor position angle; To expand the amplitude of the back EMF; A phase-locked loop module is used to obtain the speed and position of the rotor based on the secondary expanded back electromotive force; The sliding mode control unit comprises: Torque sliding mode controller to obtain reference torque according to rotor speed and the motor torque Obtain the q-axis component of the reference voltage vector; The flux sliding mode controller is used to adjust the reference flux according to the and the magnetic flux of the motor The d-axis component of the reference voltage vector is obtained.

2. According to claim 1, the position sensorless direct torque control system of the permanent magnet switched flux motor is characterized in that: The torque sliding mode controller is: ； in, is the torque loop slip variable, , is the q-axis component of the reference voltage vector.

3. The position sensorless direct torque control system of the permanent magnet switched flux motor according to claim 1 is characterized in that: The flux sliding mode controller is: ； in, is the magnetic link sliding variable, , is the d-axis component of the reference voltage vector.

4. A method for direct torque control of a permanent magnet switch flux motor without a position sensor, characterized in that: The sensorless direct torque control system of the permanent magnet switched flux motor according to any one of claims 1 to 3 is implemented, comprising the following steps: S1, detecting whether the current speed of the motor is greater than or equal to the set speed threshold, if so, obtaining the amplitude of the extended back electromotive force from the extended back electromotive force model; if not, injecting a high-frequency signal into the extended back electromotive force model until the motor speed is greater than or equal to the set speed threshold, and obtaining the amplitude of the extended back electromotive force from the extended back electromotive force model; S2, according to the extended back electromotive force amplitude obtained in step S1, obtain the secondary extended back electromotive force, and obtain the speed and position of the rotor through the phase-locked loop. and for: ； in, , is the magnitude of the secondary extended back EMF; , is the component of the extended back EMF under the αβ axis; is the rotor position angle; To expand the amplitude of the back EMF; S3, obtaining a reference torque according to the rotor speed obtained in step S2 , reference torque The torque of the motor The q-axis component of the reference voltage vector is obtained by inputting the torque sliding mode controller, and the reference flux is and the magnetic flux of the motor The d-axis component of the reference voltage vector is obtained by inputting it into the flux sliding mode controller; S4. The q-axis component of the reference voltage vector and the d-axis component of the reference voltage vector are modulated by SVPWM to obtain a modulation signal.

5. The method for position sensorless direct torque control of a permanent magnet switched flux motor according to claim 4, characterized in that: After injecting a high-frequency signal into the extended back-EMF model, the amplitude of the extended back-EMF is: ； in, is the frequency of the injected signal, n represents the harmonic order, and the superscript f represents the fundamental component. ; is the rotation speed; i d 、i q are the d-axis and q-axis components of the stator current; p is the differential operator; is the permanent magnet flux of the rotor.

Images