Pmsm model predictive control system position sensor fault online diagnosis and compensation method

Abstract

The present application relates to the field of permanent magnet synchronous motor, specifically a kind of PMSM model predictive control system position sensor fault online diagnosis and compensation method;The basic principle of PMSM model predictive current control system is analyzed, the position deviation caused by the unbalance of resolver amplitude and the imperfect fault of quadrature is analyzed, and the fault characteristics presented in d, q axis stator current, and an effective extraction method based on the fault characteristics of stator q axis current is proposed, and then fault-tolerant control is realized based on the obtained position deviation;The present application is through the position sensor fault diagnosis method of PMSM model predictive current control system, and through the adaptive signal extraction method of the double-frequency fault characteristic current adopted;Through fault mode positioning, identification and fault degree evaluation, the online diagnosis of position sensor fault is realized, and the influence of the q-axis current ripple caused by the inverter dead-time effect on the position sensor fault diagnosis accuracy is avoided.

Description

Technical Field

[0001] This invention relates to the field of permanent magnet synchronous motors (PMSMs), specifically to an online diagnosis and compensation method for position sensor faults in a PMSM model predictive control system. Background Technology

[0002] Finite-state model predictive control (FCS-MPC) algorithms are widely used in the drive and control of permanent magnet synchronous motors (PMSMs) due to their simple structure and superior dynamic performance. FCS-MPC is a control algorithm that predicts the future state of a system based on its current state and a mathematical model, and then performs online optimization based on the desired value. Depending on the control variables, FCS-MPC is divided into model predictive current control (MPCC) and model predictive torque control (MPTC). MPTC requires an observer to obtain torque and stator flux, and it also requires determining appropriate weighting coefficients to construct the objective function. However, the determination of these weighting coefficients lacks effective theoretical support and relies heavily on continuous adjustments based on extensive simulation and experimental data, making the debugging process complex. In contrast, MPCC is simple to implement, the control variable current can be directly measured, and the objective function only includes current variables with consistent dimensions when constructed, avoiding the design problem of weighting coefficients. It can achieve high-performance current control of PMSM drive systems, which is conducive to meeting the requirements of PMSM drive systems with wide speed range, good dynamic characteristics, fast current response and high power density.

[0003] Compared to position sensors such as photoelectric encoders, resolvers have significant advantages in terms of shock and vibration resistance, environmental adaptability, and output absolute position. However, due to nonlinear effects such as resolver processing and installation errors, and temperature drift of excitation and output conditioning circuit components, amplitude imbalance and imperfect orthogonality faults occur in the output of the resolver's sine and cosine windings. This leads to deviations in the motor rotor position obtained based on the resolver, causing continuous oscillations in the motor torque and speed.

[0004] To ensure the safe and reliable operation of permanent magnet synchronous motor (PMSM) drive systems, it is urgent to research and propose a position sensor fault solution that integrates resolver position sensor fault location and identification, fault severity assessment, and fault-tolerant control. Summary of the Invention

[0005] The purpose of this invention is to overcome the defects described in the background art, thereby providing an online diagnosis and compensation method for position sensor faults in a PMSM model predictive control system.

[0006] To achieve the above-mentioned objectives, the technical solution of this invention is: an online diagnostic method for position sensor faults in a PMSM model predictive control system, comprising the following steps:

[0007] Step 1: Establish a prediction model for a permanent magnet synchronous motor (PMSM) in a three-phase coordinate system;

[0008] Step 2: Define the cost function, and use the voltage vector with the minimum cost function as the optimal voltage vector to apply to the PMSM drive system in the next sampling period;

[0009] Step 3: Analyze the resolver fault and establish the stator winding of the sinusoidal resolver and the rotor winding of the cosine resolver;

[0010] Step 4: Based on Step 3, establish a model for rotor position deviation and stator current oscillation caused by resolver fault;

[0011] Step 5: Based on Step 4, analyze the characteristics of the position deviation caused by the imperfect orthogonality of the resolver and the position sensor malfunction due to amplitude imbalance.

[0012] Step 6: Based on step 5, establish an effective method for extracting fault characteristics based on stator q-axis current.

[0013] Furthermore, the prediction model for the PMSM in step 1 includes the current state equation of the PMSM, which is:

[0014]

[0015] If the sampling period T of the control system s If the time interval is sufficiently short, the discrete-time model of the PMSM can be represented by a first-order Taylor series, approximately as follows:

[0016]

[0017] Discretizing equation (1) using equation (2) yields the discrete current prediction model for PMSM.

[0018]

[0019] In the formula, T s Sampling time, and For the (k+1)th predicted current along the d-axis and the predicted current along the q-axis.

[0020] Furthermore, the cost function in step 2 is defined as:

[0021]

[0022] In the formula, i = 0, 1, ..., 7; and These are the direct-axis reference current and the quadrature-axis reference current, respectively. The last term is a nonlinear equation, specifically expressed as follows:

[0023]

[0024] In the formula, i dmax and i qmax These are the current limiting values ​​for the d-axis and q-axis, respectively.

[0025] Further, analyzing the resolver fault in step 3, firstly, the resolver amplitude is established. This includes establishing the stator windings for the sinusoidal resolver and the rotor windings for the cosine resolver. The stator windings are labeled D1D2 and D3D4, where D1D2 and D3D4 are the stator excitation winding and compensation winding, respectively, which are spatially 90 degrees apart. The stator windings are also labeled Z1Z2 and Z3Z4, where Z1Z2 and Z3Z4 are the sinusoidal output windings, respectively, which are spatially 90 degrees apart. The rotor consists of a group of cosine output windings. Then, when an alternating excitation voltage Us is applied to the excitation winding D1D2, a pulsating magnetic field is generated in the air gap. Its direction is the same as that of the axis of the D1D2 winding. This pulsating magnetic field induces pulsating electromotive forces with a phase difference of 90 degrees in the two windings of the rotor. As the angle θ between the axis of Z1Z2 and the axis of D1D2 changes, the phase of the electromotive forces induced in Z1Z2 and Z3Z4 also changes accordingly.

[0026] Furthermore, the model for determining the rotor position deviation and stator current oscillation caused by the resolver fault in step 4 includes:

[0027] Based on the transformer principle, the electromotive force induced in Z1Z2 and Z3Z4 can be expressed as the air gap magnetic flux Φ. D The functional relationship between θ and θ is shown below.

[0028]

[0029]

[0030] Where: N1 and f are the number of turns of the excitation winding and the excitation voltage frequency, respectively, and E m This is the amplitude of the induced electromotive force when the axis of the excitation winding and the axis of the rotor winding coincide;

[0031] Furthermore, step 5, determining the position deviation caused by the resolver orthogonality imperfection and amplitude imbalance position sensor malfunction, includes the following steps:

[0032] Let the electromotive force ED air gap flux Φ D It is obtained by induction in the stator excitation windings D1D2, where E and ω s These are the amplitude and angular frequency of the excitation voltage, respectively.

[0033] E D =E sin(ω) s t) (8)

[0034] According to the principle of transformers, we know that...

[0035]

[0036] Where: N1 and N2 are the number of turns of the excitation winding and the rotor sine and cosine output windings, respectively;

[0037] Therefore, the output voltages of the resolver's sine and cosine output windings can be summarized as follows:

[0038]

[0039]

[0040] The PMSM drive system position sensor uses a single-pole sine and cosine rotary transformer. The rotary transformer's sine and cosine output signals, which may contain orthogonal imperfections and amplitude imbalances, can be expressed as follows:

[0041]

[0042] In the formula: U sin U cos These represent the output voltages of the sine and cosine windings of the resolver, respectively, ω s The excitation voltage angular frequency of the resolver stator excitation winding;

[0043] The ADS145 position demodulation chip performs position demodulation based on the following formula.

[0044] U err =U sin-cos =kE×sinω s t×(sin(θ+β)cosφ-(1+α)cosθsinφ) (13)

[0045] Where: φ is the motor rotor position obtained based on the decoding chip;

[0046] The ADS145 demodulation chip's internal closed-loop feedback algorithm enables φ to track the actual motor position θ, thereby allowing U to... err As it approaches 0, equation 13 becomes

[0047] 0≈sin(θ+β)cosφ-(1+α)cosθsinφ (14)

[0048] Let φ - θ = θ err Simplifying Equation 14, we can obtain the position deviation caused by the incomplete orthogonality of the resolver and the amplitude imbalance of the position sensor as follows:

[0049]

[0050] Furthermore, the effective extraction method for fault features based on stator q-axis current in step 6 includes coordinate transformation based on the rotor position demodulated by ADS145 in step 5, to obtain the expression for the motor dq-axis stator current in the synchronous rotating coordinate system as follows:

[0051]

[0052] In the formula: I n This represents the magnitude of the stator current vector;

[0053] Given that the value of β is relatively small, sinβ≈β and cosβ≈1 are approximately true. Equation 15 can be simplified to...

[0054]

[0055] Furthermore, a PMSM model predictive control system position sensor fault compensation method is provided for fault diagnosis and fault-tolerant compensation of the rotary transformer based on the motor rotor position deviation information generated by the rotary transformer. Specifically, the method includes the following steps:

[0056] S1: When the rotor of the rotary transformer rotates synchronously with the shaft of the permanent magnet synchronous motor, the sine output signal and cosine output signal containing amplitude imbalance fault and / or quadrature imperfection fault are respectively obtained through the sine output winding and cosine output winding of the rotary transformer.

[0057] S2: The above sine and cosine output signals are demodulated by the demodulation chip to obtain the motor rotor position θ under amplitude imbalance fault and / or orthogonality imperfection fault.

[0058] S3: Based on the adaptive signal extraction algorithm, the second harmonic pulsating current of the q-axis stator current caused by amplitude imbalance faults and / or orthogonality imperfection faults is adaptively extracted to serve as the basis for fault diagnosis of the rotary transformer.

[0059] S4: The position deviation of the rotary transformer is obtained by processing the second-harmonic pulsating current. Specifically...

[0060] S4.1): Obtain the amplitude Δi of the second harmonic pulsating current of the q-axis stator current. q ;

[0061] S4.2): Calculate Δi qAmplitude imbalance fault component Δi q ×sign(sin2θ), and perform integral calculation and PI adjustment in sequence to obtain the amplitude imbalance α;

[0062] Calculate Δi q orthogonal imperfect fault component Δi q ×sign(cos2θ), and perform integral operations and PI adjustment in sequence to obtain the orthogonal imperfection β;

[0063] S4.3): Calculate and obtain the position deviation θ of the rotary transformer. err ,for

[0064]

[0065] S5: Compare the motor rotor position θ obtained in step S2 with the position deviation θ obtained in step S4.3). err The two components are added together to achieve fault-tolerant compensation for rotary transformer faults.

[0066] Furthermore, in step S3, the method for adaptively extracting the second harmonic pulsating current of the q-axis stator current is as follows:

[0067] Define the PMSM stator current i(t), which includes the target extraction signal i. o i(t) and other signals i1(t), i.e., i(t) = i o (t)+i1(t);

[0068] Define the actual extracted signal of the stator current as i ext (t), the cost function is defined as

[0069]

[0070] Where γ is a parameter vector representing the instantaneous values ​​of the actual extracted signal amplitude I(t), frequency ω(t), and phase δ(t);

[0071] The unknown parameter vector γ is adjusted using gradient descent to make the cost function J(t,γ) converge to the minimum point. The adjustment method is as follows:

[0072]

[0073] The convergence process of the cost function generates a set of nonlinear differential equations characterizing the instantaneous values ​​of the target extracted signal amplitude, frequency, and phase. Specifically:

[0074] dI(t) / dt=μ1e(t)sinφ(t) (21)

[0075] dω(t) / dt=μ2I(t)e(t)cosφ(t) (22)

[0076] dφ(t) / dt=μ2μ3e(t)cosφ(t)+ω(t) (23)

[0077] i ext (t)=I(t)sinφ (24)

[0078] Where I(t), ω(t), and φ(t) represent the actual extracted signal i, respectively. ext The amplitude, frequency, and instantaneous phase of (t) are given by e(t), which represents the extraction error. μ1, μ2, and μ3 are positive constants that determine the signal extraction accuracy and extraction speed.

[0079] Solve the above nonlinear differential equations to ultimately achieve adaptive extraction of the target extraction signal.

[0080] Furthermore, when both amplitude imbalance fault and orthogonality imperfection fault exist, the amplitude imbalance fault component Δi q ×sign(sin2θ) and orthogonal imperfect fault component Δi q ×sign(cos2θ) are mutually orthogonal and do not affect each other.

[0081] The beneficial effects of the online diagnosis and compensation method for position sensor faults in the PMSM model predictive control system of the present invention are as follows:

[0082] (1) The present invention is based on an effective method for extracting fault characteristics of stator q-axis current, namely, adaptive extraction of the second harmonic pulsating current of q-axis stator current, and online diagnosis of position sensor faults through fault mode location, identification and fault degree assessment, which cleverly avoids the influence of q-axis current pulsation caused by inverter dead zone effect on the accuracy of position sensor fault diagnosis.

[0083] (2) The present invention can compensate for the position deviation of the motor rotor caused by the rotary transformer, thereby effectively ensuring the safe and reliable operation of the permanent magnet synchronous motor drive system. Attached Figure Description

[0084] Figure 1 This is a schematic diagram of the stator and rotor winding distribution structure of a rotary transformer;

[0085] Figure 2 This is a schematic diagram of amplitude imbalance and orthogonality imperfection faults in a rotary transformer;

[0086] Figure 3 This is a schematic diagram of the workflow of the online diagnostic and compensation system. Detailed Implementation

[0087] This invention discloses an online diagnosis and compensation system for position sensor faults in a PMSM model predictive control system, used for fault diagnosis and fault-tolerant compensation of a rotary transformer, to form a position sensor fault solution that integrates fault location and identification, fault severity assessment and fault-tolerant control of the rotary transformer.

[0088] To fully disclose the technical solution of this invention, this invention first explains the basic principle of the PMSM model predictive current control system, and then explains the position deviation caused by the amplitude imbalance and orthogonality imperfection faults of the rotary transformer and the fault characteristics presented in the d and q axis stator currents; based on this, this invention proposes an effective method for extracting fault characteristics based on the stator q axis current, and realizes online diagnosis of position sensor faults through fault mode location, identification and fault degree assessment, and then performs fault-tolerant control based on the acquired position deviation.

[0089] Example 1

[0090] 1. Establish a predictive model for a permanent magnet synchronous motor (PMSM) in a three-phase coordinate system, and explain the predictive control principle of the PMSM model.

[0091] 1) Prediction Model

[0092] The current state equation of PMSM is:

[0093]

[0094] If the sampling period Ts of the control system is short enough, the discrete-time model of the PMSM can be represented by a first-order Taylor series, approximately as follows:

[0095]

[0096] Discretizing equation (1.1) using equation (1.2) yields the discrete current prediction model for PMSM.

[0097]

[0098] In the formula, Ts is the sampling time. and For the (k+1)th predicted current along the d-axis and the predicted current along the q-axis.

[0099] 2) Cost function

[0100] A two-level three-phase PMSM drive system has eight basic voltage vectors, including six non-zero voltage vectors and two zero voltage vectors. In model predictive current control, to enable the stator current to track the reference current with high performance, a reasonable cost function needs to be defined, and the voltage vector with the minimum cost function is used as the optimal voltage vector for the next sampling period of the PMSM drive system.

[0101] The cost function is defined here as follows:

[0102]

[0103] In the formula, i = 0, 1, ..., 7; and These are the direct-axis reference current and the quadrature-axis reference current, respectively. The last term is a nonlinear equation, specifically expressed as follows:

[0104]

[0105] In the formula, i dmax and i qmax These are the current limit values ​​for the d-axis and q-axis, respectively. When the predicted current amplitude generated by a voltage vector exceeds the maximum allowable current amplitude, the cost function becomes infinite, and the controller cannot select that voltage vector; when the predicted current amplitude is within the allowable range, the cost function (5) only has the first two terms remaining, and the one that makes g... i The minimum optimal voltage vector is applied to the PMSM drive system in the next cycle.

[0106] 2. Causes of rotor position deviation and stator current oscillation due to rotary transformer faults

[0107] The basic principle of a resolver is similar to that of a transformer. Its primary and secondary sides are embedded in the resolver stator and rotor respectively. The resolver rotor (secondary side) rotates synchronously with the motor shaft. The magnetic field coupling between the primary and secondary sides is related to the different positions of the resolver rotor, so that the induced voltage output by the rotor winding is related to the absolute position of the rotor.

[0108] The stator and rotor winding distributions of sine and cosine rotary transformers are as follows: Figure 1 As shown, D1D2 and D3D4 are stators that are 90 degrees apart in space, and Z1Z2 and Z3Z4 are sine output windings and cosine output windings that are 90 degrees apart in space, respectively.

[0109] When an alternating excitation voltage Us is applied to the excitation winding D1D2, a pulsating magnetic field is generated in the air gap. Its direction is the same as that of the D1D2 winding axis. This pulsating magnetic field induces pulsating electromotive forces with a phase difference of 90 degrees in the two windings of the rotor. As the angle θ between the Z1Z2 axis and the D1D2 axis changes, the phase of the electromotive forces induced in Z1Z2 and Z3Z4 also changes accordingly.

[0110] Based on the transformer principle, the electromotive force induced in Z1Z2 and Z3Z4 can be expressed as the air gap magnetic flux Φ. D Functional relationship with θ:

[0111]

[0112]

[0113] Based on transformer principles, a model is established to determine the rotor position deviation and stator current oscillation caused by resolver faults. Here, N1 and f represent the number of turns in the excitation winding and the excitation voltage frequency, respectively, and E... m It represents the amplitude of the induced electromotive force when the axis of the excitation winding and the axis of the rotor winding coincide.

[0114] Let the electromotive force E D air gap flux Φ D It is obtained by induction in the stator excitation windings D1D2, where E and ω s These are the amplitude and angular frequency of the excitation voltage, respectively.

[0115] E D =E sin(ω) s t) (8)

[0116] According to the principle of transformers, we know that...

[0117]

[0118] Wherein, N1 and N2 are the number of turns of the excitation winding and the rotor sine and cosine output windings, respectively.

[0119] Therefore, the output voltages of the resolver's sine and cosine output windings can be summarized as follows:

[0120]

[0121]

[0122] As can be seen from the above analysis, the resolver output signal is a sine and cosine voltage signal containing motor position information. Under healthy conditions, the amplitude of its sine and cosine output voltage signals is kE and the phase difference is 90 degrees. The two sine and cosine voltage signals are demodulated by the ADS145 demodulation chip and output as the actual motor rotor position signal θ.

[0123] like Figure 2 As shown, due to errors in the processing and installation of the resolver, and the nonlinearity of the resolver's excitation and conditioning circuits, the sine and cosine output signals of the resolver may exhibit amplitude imbalance faults (significantly equal amplitudes) and / or quadrature imperfection faults (significantly 90-degree phase differences). Here, α is defined as the amplitude imbalance degree, and β as the quadrature imperfection degree.

[0124] The PMSM drive system position sensor uses a single-pole sine and cosine rotary transformer. The rotary transformer's sine and cosine output signals, which may contain orthogonal imperfections and amplitude imbalances, can be expressed as follows:

[0125]

[0126] In the formula, Usin U cos These represent the output voltages of the sine and cosine windings of the resolver, respectively, ω s It is the angular frequency of the excitation voltage of the resolver stator excitation winding.

[0127] The ADS145 position demodulation chip performs position demodulation based on the following formula.

[0128] U err =U sin-cos =kE×sinω s t×(sin(θ+β)cosφ-(1+α)cosθsinφ) (13)

[0129] Where φ is the motor rotor position obtained based on the decoding chip.

[0130] The ADS145 demodulation chip's internal closed-loop feedback algorithm enables φ to track the actual motor position θ, thereby allowing U to... err As it approaches 0, equation (13) becomes

[0131] 0≈sin(θ+β)cosφ-(1+α)cosθsinφ (14)

[0132] Let φ - θ = θ err Simplifying equation (2.2), we can obtain the position deviation caused by the orthogonal imperfection and amplitude imbalance faults of the rotary transformer as follows:

[0133]

[0134] Based on the rotor position demodulated by ADS145, coordinate transformation is performed to obtain the expressions for the stator currents of the motor along the d and q axes in the synchronous rotating coordinate system.

[0135]

[0136] In the formula, I n This represents the magnitude of the stator current vector.

[0137] Given that the value of β is relatively small, sinβ≈β and cosβ≈1 are approximately true, so equation (2.4) can be simplified to...

[0138]

[0139] Thus, equations (16) and (17) clearly reveal that once orthogonality imperfection and amplitude imbalance faults occur, the rotor position obtained based on the ADS145 position demodulation chip will have a position error, which will cause the rotor position obtained based on the rotary transformer to exhibit periodic continuous oscillation. In addition, this position deviation will also cause the stator current to have a second harmonic pulsating component, causing the PMSM drive system to have continuous oscillation of torque and speed, which directly affects the dynamic and static performance of the PMSM drive system and its safe and reliable operation.

[0140] Example 2

[0141] Based on the content described in Example 1, such as Figure 3 As shown, the specific steps of the present invention for fault diagnosis and fault-tolerant compensation of a rotary transformer are as follows:

[0142] S1: The rotary transformer adopts a single-pole sine and cosine rotary transformer. When the rotor of the rotary transformer rotates synchronously with the shaft of the permanent magnet synchronous motor, the sine output winding and cosine output winding of the rotary transformer respectively obtain the sine output signal and cosine output signal containing amplitude imbalance fault and / or orthogonality imperfection fault.

[0143] S2: The ADS145 demodulation chip is used to demodulate the above sine and cosine output signals to obtain the motor rotor position θ under amplitude imbalance fault and / or orthogonality imperfection fault.

[0144] S3: Based on the adaptive signal extraction algorithm, the second harmonic pulsating current of the q-axis stator current generated by amplitude imbalance fault and / or orthogonality imperfection fault is adaptively extracted;

[0145] Specifically, the PMSM stator current i(t) is defined, which includes the target extracted signal i o i(t) and other signals i1(t), i.e., i(t) = i o (t)+i1(t);

[0146] Define the actual extracted signal of the stator current as i ext (t), the cost function is defined as

[0147]

[0148] Where γ is a parameter vector representing the instantaneous values ​​of the actual extracted signal amplitude I(t), frequency ω(t), and phase δ(t);

[0149] Gradient descent provides a method for adjusting the unknown parameter vector γ to make the cost function J(t,γ) converge to the minimum point, specifically:

[0150]

[0151] The convergence process of the cost function above can generate a set of nonlinear differential equations characterizing the extraction process of the instantaneous values ​​of the target extracted signal amplitude, frequency, and phase, specifically:

[0152] dI(t) / dt=μ1e(t)sinφ(t) (20)

[0153] dω(t) / dt=μ2I(t)e(t)cosφ(t) (21)

[0154] dφ(t) / dt=μ2μ3e(t)cosφ(t)+ω(t) (22)

[0155] i ext (t)=I(t)sinφ (23)

[0156] Where I(t), ω(t), and φ(t) represent the actual extracted signal i, respectively. ext The amplitude, frequency, and instantaneous phase of (t) are given by e(t), which represents the extraction error. μ1, μ2, and μ3 are positive constants that determine the signal extraction accuracy and extraction speed.

[0157] By solving the above nonlinear differential equations, adaptive extraction of the target extraction signal can be achieved. This extracted signal can then be used as a diagnostic basis for position sensor faults in a PMSM model predictive current control system.

[0158] S4: Based on the above-mentioned second-harmonic pulsating current, the position deviation generated by the rotary transformer is obtained, specifically:

[0159] S4.1): Obtain the amplitude Δi of the second harmonic pulsating current of the q-axis stator current. q ;

[0160] S4.2): Calculate Δi q Amplitude imbalance fault component Δi q ×sign(sin2θ), and perform integral calculation and PI adjustment in sequence to obtain the amplitude imbalance α;

[0161] Calculate Δi q orthogonal imperfect fault component Δi q ×sign(cos2θ), and perform integral operations and PI adjustment in sequence to obtain the orthogonal imperfection β;

[0162] The amplitude imbalance α and orthogonality imperfection β are further explained in detail below: When only amplitude imbalance faults exist, the stator q-axis current amplitude imbalance fault component Δi qThe integral of ×sign(sin2θ) is not zero, and the output after passing through the PI controller is not zero, while the orthogonal imperfect fault component Δi q The integral of ×sign(cos2θ) is equal to 0, and the output after passing through the PI regulator is also 0. That is, the amplitude imbalance fault has no effect on the acquisition of the degree of orthogonal imperfection fault.

[0163] When only orthogonal imperfect faults exist, the stator q-axis current orthogonal imperfect fault component Δi q The integral of ×sign(cos2θ) is not zero, and the output after passing through the PI controller is also not zero, while the amplitude imbalance fault component Δi q The integral of ×sign(sin2θ) is zero, and the output remains 0 after passing through the PI regulator. That is, the orthogonal imperfection fault has no effect on the acquisition of the magnitude imbalance fault degree.

[0164] When both amplitude imbalance fault and orthogonality imperfection fault exist, the amplitude imbalance fault component Δi q ×sign(sin2θ) and orthogonal imperfect fault component Δi q ×sign(cos2θ) are mutually orthogonal and do not affect each other.

[0165] S4.3): Calculate and obtain the position deviation θ of the rotary transformer. err ,for

[0166]

[0167] S5: Compare the motor rotor position θ obtained in step S2 with the position deviation θ obtained in step S4.3). err The two components are added together to achieve fault-tolerant compensation for rotary transformer faults.

[0168] The preferred embodiments and examples of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments and examples. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the concept of the present invention.

Claims

1. An online diagnostic method for position sensor faults in a PMSM model predictive control system, characterized in that: The steps are as follows: Step 1: Establish a prediction model for a permanent magnet synchronous motor (PMSM) in a three-phase coordinate system; Step 2: Define the cost function, and use the voltage vector with the minimum cost function as the optimal voltage vector to apply to the PMSM drive system in the next sampling period; Step 3: Analyze the resolver fault and establish the stator windings of the sinusoidal resolver and the rotor windings of the cosine resolver. Analyze the resolver fault in Step 3 by first establishing the resolver amplitude. The resolver amplitude includes establishing the stator windings of the sinusoidal resolver and the rotor windings of the cosine resolver. The stator windings are labeled D1D2 and D3D4, where D1D2 and D3D4 are the stator excitation winding and compensation winding, respectively, which are spatially separated by 90 degrees. The stator windings are labeled Z1Z2 and Z3Z4, where Z1Z2 and Z3Z4 are respectively... These are sinusoidal and cosine output windings that are 90 degrees apart in space. Then, when an alternating excitation voltage Us is applied to the excitation winding D1D2, a pulsating magnetic field is generated in the air gap. Its direction is the same as that of the axis of the D1D2 winding. This pulsating magnetic field induces pulsating electromotive forces with a phase difference of 90 degrees in the two windings of the rotor. As the angle θ between the Z1Z2 axis and the D1D2 axis changes, the phase of the electromotive forces induced in Z1Z2 and Z3Z4 also changes accordingly. Step 4: Based on Step 3, establish a model for rotor position deviation and stator current oscillation caused by resolver fault; Step 5: Based on Step 4, analyze the characteristics of the position deviation caused by the imperfect orthogonality of the resolver and the position sensor malfunction due to amplitude imbalance. Step 6: Based on step 5, establish an effective method for extracting fault characteristics based on stator q-axis current.

2. The online diagnostic method for position sensor faults in a PMSM model predictive control system according to claim 1, characterized in that: The prediction model for PMSM in step 1 includes the current state equation for PMSM, which is: If the sampling period Ts of the control system is short enough, the discrete-time model of the PMSM can be represented by a first-order Taylor series, approximately as follows: Discretizing equation (1) using equation (2) yields the discrete current prediction model for PMSM. In the formula, Ts is the sampling time. and For the (k+1)th predicted current along the d-axis and the predicted current along the q-axis.

3. The online diagnostic method for position sensor faults in a PMSM model predictive control system according to claim 1, characterized in that: The cost function in step 2 is defined as follows: In the formula, i = 0, 1, ..., 7; and These are the direct-axis reference current and the quadrature-axis reference current, respectively. The last term is a nonlinear equation, specifically expressed as follows: In the formula, i dmax and i qmax These are the current limiting values ​​for the d-axis and q-axis, respectively.

4. The online diagnostic method for position sensor faults in a PMSM model predictive control system according to claim 1, characterized in that: The model for determining rotor position deviation and stator current oscillation caused by resolver fault in step 4 includes: Based on the transformer principle, the electromotive force induced in Z1Z2 and Z3Z4 can be expressed as the air gap magnetic flux Φ. D The functional relationship between θ and θ is shown below. E Z12 =4.44fN1Φ D sinθ (6) =And m sinθ Where: N1 and f are the number of turns of the excitation winding and the excitation voltage frequency, respectively, and E m It represents the amplitude of the induced electromotive force when the axis of the excitation winding and the axis of the rotor winding coincide.

5. The online diagnostic method for position sensor faults in a PMSM model predictive control system according to claim 1, characterized in that: Step 5, determining the position deviation caused by the incomplete orthogonality of the resolver and the amplitude imbalance of the position sensor, includes the following steps: Let the electromotive force E D air gap flux Φ D It is obtained by induction in the stator excitation windings D1D2, where E and ω s These are the amplitude and angular frequency of the excitation voltage, respectively. ITS D =Ex(ω s t) (8) According to the principle of transformers, we know that... Where: N1 and N2 are the number of turns of the excitation winding and the rotor sine and cosine output windings, respectively; Therefore, the output voltages of the resolver's sine and cosine output windings can be summarized as follows: The PMSM drive system position sensor uses a single-pole sine and cosine rotary transformer. The rotary transformer's sine and cosine output signals, which may contain orthogonal imperfections and amplitude imbalances, can be expressed as follows: In the formula: U sin U cos These represent the output voltages of the sine and cosine windings of the resolver, respectively, ω s The excitation voltage angular frequency of the resolver stator excitation winding; The ADS145 position demodulation chip performs position demodulation based on the following formula. U err =U sin-cos =kE×sinω s t×(sin(θ+β)cosφ-(1+α)cosθsinφ) (13) Where: φ is the motor rotor position obtained based on the decoding chip; The ADS145 demodulation chip's internal closed-loop feedback algorithm enables φ to track the actual motor position θ, thereby causing Uerr to approach 0, and Equation 13 becomes... 0≈sin(θ+β)cosφ-(1+a)cosθsinφ (14) Let φ - θ = θ err Simplifying Equation 14, we can obtain the position deviation caused by the incomplete orthogonality of the resolver and the amplitude imbalance of the position sensor as follows: 。 6. The online diagnostic method for position sensor faults in a PMSM model predictive control system according to claim 5, characterized in that: The effective method for extracting fault features based on stator q-axis current in step 6 includes coordinate transformation based on the rotor position demodulated by ADS145 in step 5, to obtain the expression for the motor dq-axis stator current in the synchronous rotating coordinate system as follows: In the formula: In represents the magnitude of the stator current vector; Given that the value of β is relatively small, sinβ≈β and cosβ≈1 are approximately true. Equation 15 can be simplified to... 。 7. A method for position sensor fault compensation in a PMSM model predictive control system, used to perform fault diagnosis and fault-tolerant compensation for the rotary transformer based on the rotor position deviation information of the motor generated by the rotary transformer, characterized in that... Specifically, the following steps are included: S1: When the rotor of the rotary transformer rotates synchronously with the shaft of the permanent magnet synchronous motor, the sine output signal and cosine output signal containing amplitude imbalance fault and / or quadrature imperfection fault are respectively obtained through the sine output winding and cosine output winding of the rotary transformer. S2: The above sine and cosine output signals are demodulated by the demodulation chip to obtain the motor rotor position θ under amplitude imbalance fault and / or orthogonality imperfection fault. S3: Based on the adaptive signal extraction algorithm, the second harmonic pulsating current of the q-axis stator current caused by amplitude imbalance faults and / or orthogonality imperfection faults is adaptively extracted to serve as the basis for fault diagnosis of the rotary transformer. S4: The position deviation of the rotary transformer is obtained by processing the second-harmonic pulsating current. Specifically... S4.1): Obtain the amplitude Δiq of the second harmonic pulsating current of the q-axis stator current; S4.2): Calculate the amplitude imbalance fault component Δiq×sign(sin2θ), and perform integral calculation and PI adjustment in sequence to obtain the amplitude imbalance degree α; Calculate the orthogonal imperfection fault component Δiq×sign(cos2θ) of Δiq, and perform integral calculation and PI adjustment in sequence to obtain the orthogonal imperfection degree β; S4.3): Calculate and obtain the position deviation θ of the rotary transformer. err ,for S5: Compare the motor rotor position θ obtained in step S2 with the position deviation θ obtained in step S4.3). err The two components are added together to achieve fault-tolerant compensation for rotary transformer faults.

8. The position sensor fault compensation method for the PMSM model predictive control system as described in claim 7, characterized in that: In step S3, the method for adaptively extracting the second harmonic pulsating current of the q-axis stator current is as follows: Define the PMSM stator current i(t), which includes the target extracted signal io(t) and other signals i1(t), i.e., i(t) = i o (t)+i1(t); The actual extracted signal of the stator current is defined as iext(t), and the cost function is defined as follows: Where Y is a parameter vector representing the instantaneous values ​​of the actual extracted signal amplitude I(t), frequency w(t), and phase δ(t); The unknown parameter vector Y is adjusted using gradient descent to make the cost function J(t,Y) converge to the minimum point. The adjustment method is as follows: The convergence process of the cost function generates a set of nonlinear differential equations characterizing the instantaneous values ​​of the target extracted signal amplitude, frequency, and phase. Specifically: dI(t) / dt=μ1e(t)sinφ(t) (21) dω(t) / dt=μ2I(t)e(t)cosφ(t) (22) dφ(t) / dt=μ2μ3e(t)cosφ(t)+ω(t) (23) i ext (t)=I(t)sinφ (24) Where I(t), w(t), and φ(t) represent the instantaneous values ​​of the amplitude, frequency, and phase of the actual extracted signal iext(t), respectively, e(t) represents the extraction error, and μ1, μ2, and μ3 are positive constants that determine the signal extraction accuracy and extraction speed. Solve the above nonlinear differential equations to ultimately achieve adaptive extraction of the target extraction signal.

9. The PMSM model predictive control system position sensor fault compensation method as described in any one of claims 7 to 8, characterized in that: When both amplitude imbalance fault and orthogonality imperfection fault exist, the amplitude imbalance fault component Δi q ×sign(sin2θ) and orthogonal imperfect fault component Δi q ×sign(cos2θ) are mutually orthogonal and do not affect each other.

Images