A noise reduction method for low-speed position sensorless control of dual three-phase permanent magnet motors

Abstract

The present invention discloses a noise reduction method for low-speed, sensorless control of a dual three-phase permanent magnet motor. This method utilizes the relatively independent nature of the two windings in the dual three-phase permanent magnet motor to generate two random numbers using a linear congruential method. This method simultaneously performs random signal selection and winding selection, and injects the selected voltage signal into the corresponding winding. Simultaneously, the high-frequency current responses in the two windings are synthesized to establish a new position estimation coordinate system. A position information extraction method that does not require signal demodulation is designed, ultimately obtaining the motor rotor position and speed, thereby achieving sensorless control of the dual three-phase permanent magnet motor. This method can reduce the component of the injected signal in the response current in sensorless control applications using the signal injection method, effectively reducing the noise introduced by the high-frequency injected signal. This method also reduces system cost and improves reliability, ensuring stable low-speed, sensorless operation of the dual three-phase permanent magnet motor.

Description

Technical Field

[0001] The present invention belongs to the field of permanent magnet synchronous motor drive control applications, and in particular relates to a noise reduction method for low-speed position sensorless control of dual three-phase permanent magnet motors. Background Art

[0002] Permanent magnet motors (PMMs) are attracting increasing attention due to their advantages, such as high efficiency and high power density. Multiphase PMMs inherit these advantages while also offering advantages in low-voltage, high-power drive, fault-tolerant control, and complex control. A dual three-phase PMM, as a multiphase PMM, features two windings spaced 30 degrees apart, offering significant advantages in torque ripple suppression. To achieve high-performance drive control for dual three-phase PMMs, position angle measurement is essential. However, mechanical position sensors and their demodulation circuits pose challenges in terms of system cost, installation, and maintenance, while also reducing reliability in harsh environments. Sensorless control eliminates the need for mechanical position sensors and directly estimates position using motor models or motor saliency. To fully utilize motor saliency, high-frequency injection is often used to achieve sensorless control at low speeds. However, high-frequency signal injection introduces noise, making research on noise reduction in sensorless control of dual three-phase PMMs of this type of motor.

[0003] Domestic and international scholars have achieved considerable success in their research on sensorless control noise reduction methods. The Chinese invention patent, "Permanent Magnet Motor Position Sensorless Control Method Based on Low-Frequency Orthogonal Random Pulse Signal Injection" (Patent No.: CN201910488121.8), discloses a method for controlling a permanent magnet motor position sensorless based on low-frequency orthogonal random pulse signal injection. This method randomly selects the axis system, injection position, and injection sequence of four low-frequency orthogonal injection pulse signals to obtain a low-frequency orthogonal injection pulse sequence. The sampled pulse response current is then processed to obtain a discrete position demodulation signal containing position information. The Chinese invention patent "Permanent magnet motor position sensorless control method based on mixed random signal injection" (patent number: CN201910493266.7) discloses a permanent magnet motor position sensorless control method based on mixed random signal injection. It obtains a high-frequency voltage square wave sequence injected into the motor in a random manner. The phase and frequency of the voltage signal in the sequence are randomly selected and combined. After the sequence is injected into the motor and the response current is extracted, the response current is processed and the position information is extracted by high-frequency random demodulation signal. The above method mainly reduces noise by randomizing the injection signal. How to use the structural advantages of the two sets of windings of the dual three-phase permanent magnet motor to reduce the high-frequency current component caused by the injection signal, thereby reducing the noise caused by the injection signal, and realizing the low-speed position sensorless dual three-phase permanent magnet motor is the main consideration of the present invention. Summary of the Invention

[0004] Purpose of the Invention: To address the noise problem caused by high-frequency signal injection during low-speed, sensorless control of dual three-phase permanent magnet motors, a noise reduction method for sensorless control of dual three-phase permanent magnet motors is proposed. This method maintains sensorless motor control performance at low speeds while attenuating the noise caused by high-frequency injection signals.

[0005] Technical solution: To achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows:

[0006] A noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor comprises the following steps:

[0007] Step 1: Randomly select the injection voltage signal: Determine two sinusoidal voltage signals with different frequencies based on the motor parameters, and use the linear congruential method to generate the first random number between 0 and 1 to select one of them as the injection signal;

[0008] Step 2: Randomly select a winding for signal injection: Use the linear congruential method to generate a second random number between 0 and 1 to determine a set of motor windings for injection. When one cycle of signal injection is completed, return to step 1 and execute the loop.

[0009] Step 3: Perform power spectral density analysis on the high-frequency response current caused by the injected signal: derive the power spectral density expression of the response current, separate the discrete spectral density expression of the current, and determine whether there is a power spectral density peak based on whether there is a discrete spectrum;

[0010] Step 4: Demodulate the position information in the high-frequency current response: First, establish a high-frequency inductance model of the motor at low speed, solve the high-frequency current in the stator coordinate system, synthesize the high-frequency current responses in the two sets of windings, establish a synthetic position estimation coordinate system, and transform it into the demodulation coordinate system. After low-pass filtering and normalization, finally use an orthogonal phase-locked loop to estimate the position angle and speed.

[0011] Furthermore, the specific steps of step 1 include:

[0012] Define two sinusoidal voltage signals as u inj1 and u inj2 , whose amplitude and angular frequency are V1, V2 and w1, w2 respectively, where w1>w2. At the same time, in order to ensure the ideal signal-to-noise ratio, the amplitude-frequency ratio of the two signals remains equal, that is, V1 / w1=V2 / w2; a random number P between 0 and 1 is generated by the linear congruential method. When P is less than 0.5, the injection voltage signal is selected as u inj1 , otherwise the injection signal is selected as u inj2 .

[0013] Furthermore, the specific steps of step 2 include:

[0014] Define the ABC winding and DEF winding of the dual three-phase permanent magnet motor as winding 1 and winding 2, respectively. According to the linear congruential method, generate the first random number P and the second random number Q between 0 and 1. Use the second random number to select one of the two windings of the dual three-phase motor for signal injection. When Q is greater than 0.5, inject the voltage signal determined in step 1 into winding 2. Otherwise, inject the selected voltage signal into winding 1. When one cycle of injection is completed, update the two random numbers P and Q, and return to step 1 to start the loop.

[0015] Furthermore, the specific steps of step 3 include:

[0016] According to the form of the injected signal, the current response of the random signal can be expressed as

[0017]

[0018] in, represents a random frequency signal, Represents a single-cycle signal, t is the time, t k is the kth moment; when E[e j2πfT]=1, the power spectrum density of the random frequency signal can be expressed as

[0019]

[0020] Where E[] represents the mathematical expectation factor, I(f) is the Fourier transform of the current signal within one cycle, and T represents one cycle. When the following equation holds true, the discrete spectrum in the current can be eliminated, thereby reducing the noise;

[0021] E[|I(f)|]=0

[0022] Furthermore, the specific steps of establishing the high-frequency inductance model of the motor at low speed in step 4 include:

[0023] Since the motor angular velocity is low at zero speed or low speed, the back EMF term and cross-coupling term in the voltage equation can be ignored. At the same time, the resistance R can be ignored at high frequencies. s Therefore, the model of permanent magnet synchronous motor under high frequency signal excitation can be equivalent to a pure inductance model, and the voltage equation can be simplified as follows:

[0024]

[0025] Among them, u dh 、u qh 、i dh 、i qh They are the motor dq axis voltage and current, L dh and L qh is the high-frequency inductance of the motor dq axis.

[0026] Furthermore, the specific steps of solving the high-frequency current in the stator coordinate system in step 4 include:

[0027] Since the two sets of winding parameters of the dual three-phase permanent magnet motor are the same, they only differ by 30° in space, that is, winding 2 lags 30° relative to winding 1. Taking the first set of windings as an example, according to the high-frequency inductance model of the motor, when the injected voltage is as follows

[0028]

[0029] in, and are the estimated high-frequency injection voltages of the d and q axes in winding 1, V h Indicates the injected voltage u inj The amplitude, ω h represents the angular frequency of the injected voltage, “^” represents the estimated coordinate axis or estimated component;

[0030] The Park transformation matrix from the real rotation coordinate system to the estimated coordinate system is defined as the Park inverse transformation matrix (Anti-Park)

[0031]

[0032]

[0033] Among them, the position estimation error θ e1 is the true position angle of winding 1, is the estimated position angle of winding 1, Δθ e1 is the position estimation error;

[0034] Therefore, the high-frequency current in the rotating coordinate system can be calculated as

[0035]

[0036] Among them, s is the differential operator, i d1h and i q1h are the high-frequency response currents of the d and q axes in winding 1, L dh and L qh is the high-frequency inductance of the motor dq axis;

[0037] At the same time, the high-frequency current of the stator coordinate system can be further calculated as

[0038]

[0039] Similarly, the high-frequency current of winding 2 can be expressed as

[0040]

[0041] Among them, θ e2 is the position angle of winding 2.

[0042] Furthermore, the specific steps of establishing the synthetic position estimation coordinate system in step 4 include:

[0043] In order to adjust the high-frequency currents in windings 1 and 2 to a new coordinate system, the high-frequency coordinate system of winding 2 needs to be adjusted to coincide with the high-frequency coordinate system of winding 1. In order to make the stator coordinate system of winding 2 advance by 30°, the following transformation can be performed:

[0044]

[0045] Among them, i d2h and i q2h are the high-frequency response currents of the dq axes in winding 2, T represents one cycle, θ e2 is the position angle of winding 2, and are the high-frequency response currents of the α and β axes after phase shift in winding 2, respectively, and the high-frequency components in winding 1 and winding 2 are synthesized;

[0046]

[0047] Among them, i αh and i βh They represent the synthesized αβ axis high frequency response current, i α1h and i β1h are the high-frequency response currents of the dq axes in winding 1, respectively. Therefore, the synthetic position estimation coordinate system can be obtained, and its high-frequency current can be expressed as

[0048]

[0049] Furthermore, the specific steps of transforming to the demodulation coordinate system in step 4 include:

[0050] The synthetic position coordinate system can be transformed into the demodulated coordinate system through the following matrix:

[0051]

[0052] Among them, s is the differential operator, u inj is the injection voltage, and They represent the high-frequency response current of the dq axis in the demodulation coordinate system, Δθ e is the position estimation error of the synthetic coordinate system, L dh and L qh is the high-frequency inductance of the motor dq axis, L dif is the differential inductance, and L dif =(L dh -L qh ) / 2, and process the rotation component current of the demodulated coordinate system to obtain

[0053]

[0054] Then, through inverse transformation, we can get the stator component current of the demodulated coordinate system:

[0055]

[0056] in, and They represent the high-frequency response current of the α and β axes after demodulation, and

[0057]

[0058] Furthermore, the specific steps of low-pass filtering and normalization processing in step 4 include:

[0059] After filtering out the high-frequency signal with a low-pass filter LPF, normalization is performed to remove the influence of parameter changes, and we can get

[0060]

[0061] in, and They represent the high-frequency response current of the α and β axes output after demodulation and the low-pass filter, respectively. e is the true position angle of the synthetic coordinate system.

[0062] Furthermore, the specific steps of estimating the position angle and the rotation speed by the orthogonal phase-locked loop in step 4 include:

[0063] By multiplying the current component output by the demodulated coordinate system with the sine and cosine values ​​of the position angle, the position error can be obtained as

[0064]

[0065] Where ε represents the position error in the synthetic coordinate system, To estimate the position angle of the synthetic coordinate system, the proportional integral controller (PI) is used to track the position error to obtain the motor angular velocity, and the estimated position angle can be obtained through the integrator.

[0066] Beneficial effects of the present invention:

[0067] 1) The random signal and random winding sinusoidal signal injection method proposed in the present invention can effectively diffuse the power spectrum density peak, thereby reducing the noise caused by the injected signal;

[0068] 2) The present invention adopts a high-frequency signal injection method to achieve position sensorless control at low speed, eliminating the need for a mechanical position sensor;

[0069] 3) The present invention uses a synthesized high-frequency current to extract position information without the participation of a demodulated signal, thus eliminating the influence of the digital control system delay on position estimation;

[0070] 4) The present invention adopts a dual dq control method, which has a better control margin;

[0071] 5) The present invention is applicable to both surface-mounted and embedded dual three-phase permanent magnet synchronous motors. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 The overall block diagram of the dual three-phase permanent magnet motor position sensorless noise reduction control

[0073] Figure 2 Select flow chart for random signal and random winding

[0074] Figure 3The waveform of the random signal in the winding

[0075] Figure 4 Block diagram for high frequency signal demodulation

[0076] Figure 5 is the high frequency response current in the two windings

[0077] Figure 6 Power spectrum density waveforms of different injection methods

[0078] Figure 7 Comparison diagram of the true position angle and the estimated position angle

[0079] Figure 8 Comparison chart of actual speed and estimated speed DETAILED DESCRIPTION

[0080] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0081] like Figure 1 As shown, the present invention proposes a noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor.

[0082] The specific implementation steps of the proposed noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor include:

[0083] Step 1: Randomly select the injected voltage signal

[0084] Define two sinusoidal voltage signals as u inj1 and u inj2 , whose amplitude and angular frequency are V1, V2 and w1, w2 respectively, where w1>w2. At the same time, in order to ensure the ideal signal-to-noise ratio, the amplitude-frequency ratio of the two signals remains equal, that is, V1 / w1=V2 / w2. A random number P between 0 and 1 is generated by the linear congruential method. When P is less than 0.5, the injection voltage signal is selected as u inj1 , otherwise the injection signal is selected as u inj2 .

[0085] Step 2: Randomly select windings for signal injection

[0086] Define the ABC winding and DEF winding of the dual three-phase permanent magnet motor as winding 1 and winding 2 respectively. According to the linear congruential method, generate the first random number P and the second random number Q between 0 and 1 at the same time. Use the second random number to select one of the two windings of the dual three-phase motor for signal injection. When Q is greater than 0.5, inject the voltage signal determined in step 1 into winding 2, otherwise inject the selected voltage signal into winding 1. When one cycle of injection is completed, update the two random numbers P and Q, and return to step 1 to start the loop execution, as shown in the following example. Figure 2 shown.

[0087] Step 3: Perform power spectrum density analysis on the high frequency response current caused by the injected signal

[0088] According to the form of the injected signal, the current response of the random signal can be expressed as

[0089]

[0090] in, represents a random frequency signal, Represents a single-cycle signal. When E[e j2πfT ]=1, the power spectrum density of the random frequency signal can be expressed as

[0091]

[0092] Where E[] represents the mathematical expectation factor, I(f) is the Fourier transform of the current signal within one cycle, and T represents one cycle. When the following equation holds, the discrete spectrum in the current can be eliminated.

[0093] E[|I(f)|]=0

[0094] Step 4: Demodulate the position information in the high-frequency current response

[0095] Since the electrical angular velocity is low at zero speed or low speed, the back electromotive force term and cross-coupling term in the voltage equation can be ignored, and the resistance R can be ignored at high frequencies. s Therefore, the model of the permanent magnet synchronous motor under high-frequency signal excitation can be equivalent to a pure inductance model, and the voltage equation can be simplified as:

[0096]

[0097] Among them, u dh 、u qh 、i dh 、i qh They are the motor dq axis voltage and current, L dh and L qh is the high-frequency inductance of the motor dq axis.

[0098] Since the two sets of winding parameters of the dual three-phase permanent magnet motor are the same, they only differ by 30° in space, that is, winding 2 lags 30° relative to winding 1. Taking the first set of windings as an example, according to the high-frequency inductance model of the motor, when the injected voltage is

[0099]

[0100] in, and are the estimated high-frequency injection voltages of the d and q axes in winding 1, V h Indicates the injected voltage u inj The amplitude, w h represents the angular frequency of the injected voltage, and “^” represents the estimated coordinate axis or estimated component.

[0101] The Park transformation matrix from the real rotation coordinate system to the estimated coordinate system is defined as the Park inverse transformation matrix (Anti-Park)

[0102]

[0103]

[0104] Among them, the position estimation error θ e1 is the true position angle of winding 1, is the estimated position angle of winding 1, Δθ e1 is the position estimation error;

[0105] Therefore, the high-frequency current in the rotating coordinate system can be calculated as

[0106]

[0107] Among them, s is the differential operator, i d1h and i q1h are the high frequency response currents of d and q axes in winding 1 respectively;

[0108] At the same time, the high-frequency current of the stator coordinate system can be further calculated as

[0109]

[0110] Similarly, the high-frequency current of winding 2 can be expressed as

[0111]

[0112] Among them, θ e2 is the position angle of winding 2.

[0113] like Figure 3The following table shows the injection signals of the two windings. In order to adjust the high-frequency currents in windings 1 and 2 to a new coordinate system, the high-frequency coordinate system of winding 2 needs to be adjusted to coincide with the high-frequency coordinate system of winding 1. In order to make the stator coordinate system of winding 2 advance by 30°, the following transformation can be performed:

[0114]

[0115] Among them, i d2h and i q2h are the high-frequency response currents of the dq axes in winding 2, and They are the high frequency response currents of the αβ axis after phase shift in winding 2. Then the high frequency components in winding 1 and winding 2 are synthesized

[0116]

[0117] Among them, i αh and i βh Respectively represent the synthesized αβ axis high frequency response current. Therefore, the synthetic position estimation coordinate system can be obtained, and its high frequency current can be expressed as

[0118]

[0119] The synthetic position coordinate system can be transformed into the demodulated coordinate system through the following matrix:

[0120]

[0121] in, and They represent the high-frequency response current of the dq axis in the demodulation coordinate system, Δθ e is the position estimation error of the synthetic coordinate system, L dif is the differential inductance, and L dif =(L dh -L qh ) / 2. By processing the rotational component current of the demodulated coordinate system, we can obtain

[0122]

[0123] Then, through inverse transformation, we can get the stator component current of the demodulated coordinate system:

[0124]

[0125] in, and They represent the high-frequency response current of the α and β axes after demodulation, and

[0126]

[0127] After filtering out the high-frequency signal with a low-pass filter (LPF), normalization is performed to remove the influence of parameter changes, and we can get

[0128]

[0129] in, and They represent the high-frequency response current of the α and β axes output after demodulation and the low-pass filter, respectively. e is the true position angle of the synthetic coordinate system. By multiplying the current component output by the demodulated coordinate system with the sine and cosine values ​​of the position angle, the position error can be obtained as

[0130]

[0131] Where ε represents the position error in the synthetic coordinate system. The proportional integral controller (PI) is then used to track the position error, and the estimated position angle can be obtained through the integrator. Figure 4 Shown is the block diagram of high-frequency signal demodulation.

[0132] Figure 5 The high frequency response current in the two sets of windings, Figure 6 These are the power spectrum density waveforms of different injection methods; Figure 7 and Figure 8 The following are comparisons of the motor's actual and estimated position angles, and actual and estimated speeds, respectively. It can be seen that the proposed method can effectively diffuse the power spectrum density spikes of the high-frequency injection signal, thereby reducing noise. It also achieves good position estimation results, ensuring stable operation of sensorless control.

[0133] The above embodiments are intended only to illustrate the design concepts and features of the present invention, and are intended to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. Therefore, any equivalent changes or modifications based on the principles and design concepts disclosed in the present invention are intended to fall within the scope of protection of the present invention.

Claims

1. A noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor, characterized in that: The following steps are involved: Step 1: Randomly select the injection voltage signal: Determine two sinusoidal voltage signals with different frequencies based on the motor parameters, and use the linear congruential method to generate the first random number between 0 and 1 to select one of them as the injection signal; Step 2: Randomly select a winding for signal injection: Use the linear congruential method to generate a second random number between 0 and 1 to determine a set of motor windings for injection. When one cycle of signal injection is completed, return to step 1 and execute the loop. Step 3: Perform power spectral density analysis on the high-frequency response current caused by the injected signal: derive the power spectral density expression of the response current, separate the discrete spectral density expression of the current, and determine whether there is a power spectral density peak based on whether there is a discrete spectrum; Step 4: Demodulate the position information from the high-frequency current response: First, establish a high-frequency inductance model of the motor at low speed, solve the high-frequency current in the stator coordinate system, synthesize the high-frequency current responses in the two sets of windings, establish a synthetic position estimation coordinate system, and transform it to the demodulation coordinate system. After low-pass filtering and normalization, the position angle and speed are finally estimated using an orthogonal phase-locked loop. The specific steps for solving the high-frequency current in the stator coordinate system in step 4 include: Since the two sets of winding parameters of the dual three-phase permanent magnet motor are the same, they only differ by 30° in space, that is, winding 2 lags 30° relative to winding 1. Taking the first set of windings as an example, according to the high-frequency inductance model of the motor, when the injected voltage is as follows in, and are the estimated high-frequency injection voltages of the d and q axes in winding 1, V h Indicates the injected voltage u inj The amplitude, ω h represents the angular frequency of the injected voltage, "^" represents the estimated coordinate axis or estimated component; The Park transformation matrix from the real rotation coordinate system to the estimated coordinate system is defined as the Park inverse transformation matrix (Anti-Park) Among them, the position estimation error θ e1 is the true position angle of winding 1, is the estimated position angle of winding 1, Δθ e1 is the position estimation error; Therefore, the high-frequency current in the rotating coordinate system is calculated to be Among them, s is the differential operator, i d1h and i q1h are the high-frequency response currents of the d and q axes in winding 1, L dh and L qh is the high-frequency inductance of the motor dq axis; At the same time, the high-frequency current of the stator coordinate system is further calculated and expressed as Similarly, the high-frequency current of winding 2 is expressed as Among them, θ e2 is the position angle of winding 2.

2. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1 is characterized in that: The specific steps of step 1 include: Define two sinusoidal voltage signals as u inj1 and u inj2 , whose amplitude and angular frequency are V1, V2 and w1, w2 respectively, where w1>w2. At the same time, in order to ensure the ideal signal-to-noise ratio, the amplitude-frequency ratio of the two signals remains equal, that is, V1 / w1=V2 / w2; a random number P between 0 and 1 is generated by the linear congruential method. When P is less than 0.5, the injection voltage signal is selected as u inj1 , otherwise the injection signal is selected as u inj2 .

3. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1, characterized in that: The specific steps of step 2 include: Define the ABC winding and DEF winding of the dual three-phase permanent magnet motor as winding 1 and winding 2, respectively. According to the linear congruential method, generate the first random number P and the second random number Q between 0 and 1. Use the second random number to select one of the two windings of the dual three-phase motor for signal injection. When Q is greater than 0.5, inject the voltage signal determined in step 1 into winding 2. Otherwise, inject the selected voltage signal into winding 1. When one cycle of injection is completed, update the two random numbers P and Q, and return to step 1 to start the loop.

4. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1, characterized in that: The specific steps of step 3 include: According to the form of the injected signal, the current response of the random signal is expressed as in, represents a random frequency signal, Represents a single-cycle signal, t is the time, t k is the kth moment; when E[e j2πfT ]=1, the power spectrum density of the random frequency signal is expressed as Where E[] represents the mathematical expectation factor, I(f) is the Fourier transform of the current signal within one cycle, and T represents one cycle. When the following equation holds true, the discrete spectrum in the current is eliminated, thereby reducing the noise; E[|I(f)|]=0.

5. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1, characterized in that: The specific steps for establishing the high-frequency inductance model of the motor at low speed in step 4 include: Since the motor angular velocity is low at zero speed or low speed, the back EMF term and cross-coupling term in the voltage equation are ignored, and the resistance R is ignored at high frequency. s Therefore, the model of the permanent magnet synchronous motor under high-frequency signal excitation is equivalent to a pure inductance model, and the voltage equation is simplified to: Among them, u dh 、u qh 、i dh 、i qh They are the motor dq axis voltage and current, L dh and L qh is the high-frequency inductance of the motor dq axis.

6. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1, characterized in that: The specific steps for establishing the synthetic position estimation coordinate system in step 4 include: In order to adjust the high-frequency currents in windings 1 and 2 to a new coordinate system, the high-frequency coordinate system of winding 2 needs to be adjusted to coincide with the high-frequency coordinate system of winding 1. In order to make the stator coordinate system of winding 2 advance by 30°, the following transformation is performed: Among them, i d2h and i q2h are the high-frequency response currents of the dq axes in winding 2, T represents one cycle, θ e2 is the position angle of winding 2, and are the high-frequency response currents of the α and β axes after phase shift in winding 2, respectively, and the high-frequency components in winding 1 and winding 2 are synthesized; Among them, i αh and i βh They represent the synthesized αβ axis high frequency response current, i α1h and i β1h are the high-frequency response currents of the dq axes in winding 1, respectively. Therefore, the synthetic position estimation coordinate system is obtained, and its high-frequency current is expressed as 7. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1, characterized in that: The specific steps of transforming to the demodulation coordinate system in step 4 include: The synthetic position coordinate system is transformed into the demodulation coordinate system through the following matrix: Among them, s is the differential operator, u inj is the injection voltage, and They represent the high-frequency response current of the dq axis in the demodulation coordinate system, Δθ e is the position estimation error of the synthetic coordinate system, L dh and L qh is the high-frequency inductance of the motor dq axis, L dif is the differential inductance, and L dif =(L dh -L qh ) / 2, and process the rotation component current of the demodulated coordinate system to obtain Then, through inverse transformation, we can get the stator component current of the demodulated coordinate system: in, and They represent the high-frequency response current of the α and β axes after demodulation, and 8. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 1, characterized in that: The specific steps of low-pass filtering and normalization processing in step 4 include: After filtering out the high-frequency signal with a low-pass filter LPF, normalization is performed to remove the influence of parameter changes, and we can get in, and They represent the high-frequency response current of the α and β axes output after demodulation and the low-pass filter, respectively. e is the true position angle of the synthetic coordinate system.

9. The noise reduction method for low-speed position sensorless control of a dual three-phase permanent magnet motor according to claim 8, characterized in that: The specific steps of the orthogonal phase-locked loop in step 4 to estimate the position angle and speed include: Multiply the current component output by the demodulated coordinate system with the sine and cosine values ​​of the position angle to obtain the position error: Where ε represents the position error in the synthetic coordinate system, To estimate the position angle of the synthetic coordinate system, the proportional integral controller (PI) is used to track the position error to obtain the motor angular velocity, and the estimated position angle is obtained through the integrator.

Images