Position sensorless high-performance control method for oil pump motor in low-speed domain

Abstract

The invention relates to a sensorless high-performance control method for an oil pump motor in a low-speed domain. The sensorless high-performance control method comprises the following steps: step 1, building a high-frequency signal injection model; based on a traditional permanent magnet synchronous motor model, a high-frequency mathematical model of the oil pump under a d-q axis system is derived, and the relation between q-axis high-frequency response current and the angle error is obtained; step 2, random amplitude and frequency signal injection; obtaining q-axis high-frequency response current based on a high-frequency response current equation and the injected voltage signal with random amplitude and frequency; 3, performing signal demodulation based on an adaptive filtering algorithm to obtain an angle error signal; and step 4, after the rotor position error signal is extracted in the step 3, realizing rotor position observation through a PI observer. The noise caused by fixed frequency injection can be greatly reduced in the low-speed domain state of the oil pump motor; and secondly, a target signal can be accurately extracted to participate in closed-loop control or rotor position observation, so that the control performance is improved.

Description

Technical Field

[0001] The invention relates to the technical field of oil pump motors, and in particular to a high-performance control method for low-speed oil pump motors without a position sensor. Background Art

[0002] In order to achieve high-performance control of the oil pump motor, it is necessary to obtain accurate rotor position angle. The traditional solution is to obtain it by installing mechanical position sensors, such as resolvers, encoders, etc. This solution will increase costs and have low adaptability. Therefore, position sensorless control solutions have become one of the development trends. In the low-speed domain, high-frequency signal injection is usually used to achieve oil pump position sensorless control, but the injection of high-frequency signals will cause torque fluctuations and noise. Summary of the invention

[0003] In view of the above problems, the present invention provides a high-performance control method for a low-speed oil pump motor without a position sensor, which solves the problems of high noise and multiple signal demodulation filters in the traditional low-speed sensorless control of the oil pump.

[0004] The technical solution adopted by the present invention to solve the above technical problems is: a high-performance control method for a low-speed oil pump motor without a position sensor, comprising the following steps:

[0005] Step 1: Building a high-frequency signal injection model;

[0006] Based on the traditional permanent magnet synchronous motor model, the high-frequency mathematical model of the oil pump in the dq axis system is derived, and the relationship between the q-axis high-frequency response current and the angle error is obtained;

[0007] Step 2, random amplitude and frequency signal injection;

[0008] Based on the high-frequency response current equation and the injected voltage signal of random amplitude and frequency, the q-axis high-frequency response current is obtained;

[0009] Step 3, signal demodulation based on an adaptive filtering algorithm to obtain an angle error signal;

[0010] From step 2, we know that after the high-frequency voltage signal is injected, the d-axis and q-axis currents of the system response contain three components, namely, the DC component, the high-frequency component and the PWM harmonic current component. Taking the q-axis as an example, It is expressed as:

[0011]

[0012] Where: is the DC component, is the high frequency injected signal amplitude, ω s , are the amplitude, frequency and phase of PWM harmonic current respectively,

[0013] After frequency doubling equation (3.1), we get

[0014]

[0015] Simplified:

[0016]

[0017] The formula (3.3) is Filtered by adaptive algorithm Said Contains the rotor position information, which is linearized to obtain:

[0018]

[0019] The adaptive filtering algorithm takes the center frequency of the target signal as input, performs orthogonal processing on it, performs linear combination by assigning different weight coefficients, and makes a difference between it and the reference input signal. The adaptive algorithm continuously adjusts the weight coefficient to finally separate the target signal from the reference input signal. The target signal is the target angular velocity ω. The specific implementation process of the algorithm is:

[0020]

[0021] Y k =ω k1 ·X s +ω k2 X c (3.8)

[0022] E k =U r -Y k (3.9)

[0023] ω k1 (n+1)=ω k1 (n)+μ2E k U r (3.10)

[0024] ω k2 (n+1)=ω k2 (n)+μ2E k U r (3.11)

[0025] Where: U r is the reference input signal, that is, X s, X c is the orthogonalized signal, ω 0 is the center frequency, ω k1 ,ω k2 is the weight coefficient, E k is the notch output, Y k is the filter output, that is,

[0026] Step 4: After the rotor position error signal is extracted in step 3, the rotor position observation is realized through the PI observer.

[0027] Preferably, the step 1 is specifically as follows:

[0028] The dq mathematical model of PMSM in the coordinate system is expressed as follows:

[0029]

[0030] The frequency of the high-frequency signal usually injected is ω h is 0.5~2kHz, when the working condition is zero low speed, ω h Much larger than the motor fundamental frequency ω e , so the items related to the rotor electrical angular velocity can be ignored, so the mathematical model is simplified as follows under high-frequency signal excitation:

[0031]

[0032] Formula (1.2) can be expressed as a matrix model:

[0033]

[0034] Constructing Observations The rotor angle observation error is obtained through its relationship with the actual dq axis system:

[0035]

[0036] In observation After the high-frequency voltage signal is injected into the shaft system, the response current is expressed as follows:

[0037]

[0038] Inject the following high-frequency sinusoidal voltage signal:

[0039]

[0040] Substituting into formula (1.4), the high-frequency current after response is:

[0041]

[0042] Where: is the amplitude of the high frequency response current, L is the average inductance, ΔL is the semi-differential inductance,

[0043] For formula (1.6) Simplified:

[0044]

[0045] Where: k = I h ΔL.

[0046] Preferably, the step 2 is specifically as follows:

[0047] The injected voltage is expressed as:

[0048]

[0049] In the formula, x = 1, 2,

[0050]

[0051] Indicates the injection frequency of the signal with lower frequency, U h1 ,ω h1 is the amplitude and frequency of the lower frequency signal, Indicates the injection frequency of a higher signal, U h2 ,ω h2 is the amplitude and frequency of the higher frequency signal, where ω h1 <ω h2 ,

[0052] In order to reduce noise, it is necessary to select appropriate amplitude and frequency, and set the amplitude and frequency ratio of the two frequency signals to be the same, that is:

[0053]

[0054] After injecting a voltage signal with random amplitude and frequency, the high-frequency current after response is:

[0055]

[0056] Simplified q-axis current It is expressed as:

[0057]

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention can realize closed-loop control of the oil pump motor in the low-speed state, and by injecting signals of random frequency and amplitude, it can greatly reduce the noise caused by fixed frequency injection; secondly, the fixed frequency signal can be accurately extracted through an adaptive filtering algorithm to reduce the amount of harmonics in the target signal, thereby accurately extracting the target signal to participate in closed-loop control or rotor position observation, thereby improving control performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of each coordinate system of the IPMSM of the present invention;

[0061] Figure 2 It is a schematic diagram of the ANF principle block diagram of the present invention;

[0062] Figure 3 It is a schematic diagram of the PI observer block diagram of the present invention;

[0063] Figure 4 It is the principle diagram of position sensorless control of the present invention. DETAILED DESCRIPTION

[0064] The following will be combined Figure 1-4 The present invention is described in detail. The exemplary embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0065] A low-speed oil pump motor position sensorless high-performance control method comprises the following steps:

[0066] Step 1: Building a high-frequency signal injection model;

[0067] Based on the traditional permanent magnet synchronous motor model, the high-frequency mathematical model of the oil pump in the dq axis system is derived, and the relationship between the q-axis high-frequency response current and the angle error is obtained;

[0068] The mathematical model parameters of the permanent magnet synchronous motor in the ABC three-phase stationary coordinate system are coupled, which makes it difficult to control. In order to simplify the motor control, the control method of the DC motor is referred to, and the flux linkage and torque are decoupled and controlled separately, that is, the mathematical model is constructed in the dq axis system. The dq mathematical model of the PMSM in the coordinate system can be expressed as follows:

[0069]

[0070] The frequency of the high-frequency signal usually injected is ω h is 0.5~2kHz, when the working condition is zero low speed, ω h Much larger than the motor fundamental frequency ω e , so the items related to the rotor electrical angular velocity can be ignored, so the mathematical model is simplified as follows under high-frequency signal excitation:

[0071]

[0072] The formula (1.2) can be expressed as a matrix model:

[0073]

[0074] In order to realize the rotor position observation without position sensor, it is necessary to build an observation The relationship between the axis system and the actual dq axis system is as follows Figure 1 As shown in Figure 2, the rotor angle observation error is:

[0075] In observation After the high-frequency voltage signal is injected into the shaft system, the response current can be expressed as follows:

[0076]

[0077] Inject the following high-frequency sinusoidal voltage signal:

[0078]

[0079] Substituting into formula (1.4), the high-frequency current after response is obtained as:

[0080]

[0081] Where: is the amplitude of the high frequency response current, L is the average inductance, ΔL is the semi-differential inductance,

[0082] For formula (1.6) Simplified:

[0083]

[0084] Where: k = I h ΔL.

[0085] Step 2, random amplitude and frequency signal injection;

[0086] Based on the high-frequency response current equation and the injected voltage signal of random amplitude and frequency, the q-axis high-frequency response current is obtained;

[0087] Traditional fixed frequency and amplitude signal injection will generate large high-frequency noise. In order to overcome the above problems, this solution reduces noise by injecting signals of different frequencies and amplitudes.

[0088] The injected voltage can be expressed as:

[0089]

[0090] In the formula, x = 1, 2,

[0091]

[0092] Indicates the injection frequency of the signal with lower frequency, U h1 ,ω h1 is the amplitude and frequency of the lower frequency signal, Indicates the injection frequency of a higher signal, U h2 ,ω h2 is the amplitude and frequency of the higher frequency signal, where ω h1 <ω h2 ,

[0093] In order to prevent the injection of random signals from causing greater noise, it is necessary to select appropriate amplitude and frequency. The present invention sets the amplitude and frequency ratio of the two frequency signals to be the same, that is:

[0094]

[0095] After injecting a voltage signal with random amplitude and frequency, the high-frequency current after response is obtained as:

[0096]

[0097] Simplified q-axis current It is expressed as:

[0098]

[0099] Step 3, signal demodulation based on an adaptive filtering algorithm to obtain an angle error signal;

[0100] From step 2, we can see that after the high-frequency voltage signal is injected, the d-axis and q-axis currents of the system response contain three components, namely, the DC component, the high-frequency component and the PWM harmonic current component. Taking the q-axis as an example, It can be expressed as:

[0101]

[0102] Where: is the DC component, is the high frequency injected signal amplitude, ω s , are the amplitude, frequency and phase of PWM harmonic current respectively.

[0103] Among them, the DC component It is necessary to participate in the closed-loop control of the current loop. In order to improve the control effect, the traditional solution needs to use a low-pass filter to filter out other frequency components.

[0104] In order to achieve position observation, it is necessary to extract the rotor angle error in the high-frequency response current signal After frequency doubling equation (3.1), we can get

[0105]

[0106] Simplified:

[0107]

[0108] Where:

[0109]

[0110] After frequency doubling contains the rotor position information. When the rotor angle error is very small, linear processing can be obtained:

[0111]

[0112] In order to improve the performance of the sensorless control system, it is necessary to extract the precise DC component as well as Traditional systems use low-pass filters and band-pass filters to filter and process the target signal. Low-pass filters will limit the system bandwidth and affect system performance, while band-pass filters will cause phase lag and reduce Accuracy, and because the injection signal frequency is a random signal, when the frequency and amplitude signal are injected, the filter parameters need to be changed, which cannot meet the algorithm requirements. In view of the above limitations, an adaptive filtering algorithm is proposed, that is, Filtered by adaptive algorithm Its principle is as follows Figure 2 As shown,

[0113] The algorithm takes the center frequency of the target signal as input, performs orthogonal processing on it, performs linear combination by assigning different weight coefficients, and makes a difference between it and the reference input signal. The weight coefficient is continuously adjusted by the adaptive algorithm, and finally the target signal is separated from the reference input signal. The target signal is the target angular velocity ω. The specific implementation process of the algorithm is:

[0114]

[0115] Y k =ω k1 ·X s +ω k2 X c (3.8)

[0116] E k =Ur -Y k (3.9)

[0117] ω k1 (n+1)=ω k1 (n)+μ2E k U r (3.10)

[0118] ω k2 (n+1)=ω k2 (n)+μ2E k U r (3.11)

[0119] Where: U r is the reference input signal, that is, X s , X c is the orthogonalized signal, ω 0 is the center frequency, ω k1 ,ω k2 is the weight coefficient, E k is the notch output, Y k is the filter output, that is,

[0120] Through the algorithm, the reference input signal U r Implement filtering and notch processing to obtain accurate Y k Moreover, the algorithm has a narrow bandwidth during filtering and notching, a large amplitude gain, a good filtering effect, and few other frequency signal components in the extracted signal.

[0121] Step 4: After the accurate rotor position error signal is extracted, the rotor position observation is realized through the PI observer. The transfer function is shown in formula (4.1):

[0122]

[0123] , the observation principle diagram is as follows Figure 3 As shown, when the rotor position error is obtained After that, after the PI observer, we can get After integration, we can get the observed angle

[0124] During the implementation process, Figure 4 As shown, when the desired speed and motor output speed The difference after comparison is used as the input of the speed loop PI controller to obtain the desired reference q-axis current. And the expected d-axis current Respectively The corresponding motor output dq axis current i d0 、i q0 The difference is input to two PI controllers, and the controllers output the corresponding voltage signal u d 、u q At this time, a signal with random amplitude and frequency is superimposed on the d-axis voltage, and u is obtained after the coordinate change. α 、u β , through the space vector pulse width modulation (SVPWM), by switching different switching modes, the voltage corresponding to the inverter output is controlled, and the three-phase current output by the motor is collected and obtained after coordinate transformation. d 、i q , after demodulating the q-axis current, the angle error signal is output Finally, the angle is observed through the PI regulator to achieve closed-loop control of the motor without position sensor.

[0125] The technical solutions provided by the embodiments of the present invention are introduced in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the embodiments of the present invention. The description of the above embodiments is only applicable to help understand the principles of the embodiments of the present invention. At the same time, for those skilled in the art, according to the embodiments of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A high-performance control method for a low-speed oil pump motor without a position sensor, characterized in that: The following steps are involved: Step 1: Building a high-frequency signal injection model; Based on the traditional permanent magnet synchronous motor model, the high-frequency mathematical model of the oil pump in the dq axis system is derived, and the relationship between the q-axis high-frequency response current and the angle error is obtained; Step 2, random amplitude and frequency signal injection; Based on the high-frequency response current equation and the injected voltage signal of random amplitude and frequency, the q-axis high-frequency response current is obtained; Step 3, signal demodulation based on an adaptive filtering algorithm to obtain an angle error signal; From step 2, we know that after the high-frequency voltage signal is injected, the d-axis and q-axis currents of the system response contain three components, namely, the DC component, the high-frequency component and the PWM harmonic current component. Taking the q-axis as an example, It is expressed as: Where: is the DC component, is the high frequency injected signal amplitude, ω s , are the amplitude, frequency and phase of the PWM harmonic current respectively, After frequency doubling equation (3.1), we get Simplified: The formula (3.3) is Filtered by adaptive algorithm Said Contains the rotor position information, which is linearized to obtain: The adaptive filtering algorithm takes the center frequency of the target signal as input, performs orthogonal processing on it, performs linear combination by assigning different weight coefficients, and makes a difference between it and the reference input signal. The adaptive algorithm continuously adjusts the weight coefficient to finally separate the target signal from the reference input signal. The target signal is the target angular velocity ω. The specific implementation process of the algorithm is: Y k =ω k1 ·X s +oh k2 X c (3.8) Yes k =You r -Y k (3.9) oh k1 (n+1)=ω k1 (n)+μ2E k U r (3.10) oh k2 (n+1)=ω k2 (n)+μ2E k U r (3.11) Where: U r is the reference input signal, that is, X s , X c is the orthogonalized signal, ω0 is the center frequency, ω k1 ,ω k2 is the weight coefficient, E k is the notch output, Y k is the filter output, that is, Step 4: After the rotor position error signal is extracted in step 3, the rotor position observation is realized through the PI observer.

2. According to claim 1, the low-speed oil pump motor position sensorless high-performance control method is characterized by: The step 1 specifically comprises: The dq mathematical model of PMSM in the coordinate system is expressed as follows: The frequency of the high-frequency signal usually injected is ω h is 0.5~2kHz, when the working condition is zero low speed, ω h Much larger than the motor fundamental frequency ω e , so the items related to the rotor electrical angular velocity can be ignored, so the mathematical model is simplified as follows under high-frequency signal excitation: Formula (1.2) can be expressed as a matrix model: Constructing Observations The rotor angle observation error is obtained through its relationship with the actual dq axis system: In observation After the high-frequency voltage signal is injected into the shaft system, the response current is expressed as follows: Inject the following high-frequency sinusoidal voltage signal: Substituting into formula (1.4), the high-frequency current after response is: Where: is the amplitude of the high frequency response current, L is the average inductance, ΔL is the semi-differential inductance, For formula (1.6) Simplified: Where: k = I h ΔL.

3. According to claim 1, the low-speed oil pump motor position sensorless high-performance control method is characterized by: The step 2 is specifically as follows: The injected voltage is expressed as: In the formula, x = 1, 2, Indicates the injection frequency of the signal with lower frequency, U h1 ,ω h1 is the amplitude and frequency of the lower frequency signal, Indicates the injection frequency of a higher signal, U h2 ,ω h2 is the amplitude and frequency of the higher frequency signal, where ω h1 <ω h2 , In order to reduce noise, it is necessary to select appropriate amplitude and frequency, and set the amplitude and frequency ratio of the two frequency signals to be the same, that is: After injecting a voltage signal with random amplitude and frequency, the high-frequency current after response is: Simplified q-axis current It is expressed as:

Images