CN104836501B - A kind of method of permasyn morot on-line parameter identification - Google Patents

Abstract

The present invention relates to PMSM parameter of electric machine on-line identification methods, based on the speed-less sensor vector control system of virtual dc qc coordinate systems, including：1), for non-salient pole PMSM, the qc axle steady-state equations for considering the influence of parameter of electric machine error are set upWherein, R_sFor stator winding resistance, L_dFor d-axis inductance,It is rotor velocity, u for magnetic linkage, ω_qcFor qc shaft voltages, i_qcFor qc shaft currents, i_dcFor dc shaft currents；2), repeatedly motor operating state is suitably adjusted, and records the steady-state value of each variable under lower state, party's formula is substituted into and obtains multigroup steady state relation, and obtains by the calculating of state variable increment the identifier of the parameter of electric machine.This method only in time adds i in original control strategy_dcAnd the microvariations in short-term of ω instructions, original control flow and system are normally run without influence, algorithm is simple and direct, can be achieved at low cost the on-line identification of non-salient pole or the salient pole PMSM parameters of electric machine.

Description

Permanent magnet synchronous motor parameter online identification method

Technical Field

The invention belongs to the technical field of electric transmission and power electronics, and particularly relates to a method for online identification of parameters of a permanent magnet synchronous motor.

Background

Permanent Magnet Synchronous Motor (PMSM) has the advantages of high specific power, energy conservation, high efficiency, accurate control and the like, and is more and more widely applied to various fields^[]]. The high-performance control method of the PMSM mainly includes vector control, direct torque control, and the like. Among them, a speed sensorless control system with low system cost and strong environmental adaptability becomes a research hotspot.

The performance of the PMSM sensorless control system depends greatly on the estimation accuracy of the rotor position and the rotation speed, and the detection methods can be divided into two categories. The first type is to estimate the position and the speed by injecting a specific signal, which has the advantages of insensitivity to motor parameters, wide applicable speed range, etc., but makes ideal assumption on motor characteristics, such as high-frequency injection method^[2]Salient pole effect of the motor, INFORM method, needs to be utilized^[3]The back electromotive force is assumed to be completely sinusoidal, and the calculation is complex, so that the requirement on a control chip is high. The second type is based on fundamental excitation and back-emf estimation methods, e.g. virtual coordinate system methods^[4]Sliding mode observer method^[5]. The method is simple in calculation and easy to implement, but the method is used for the motor parametersThe dependence is strong.

PMSM parameter has stator winding resistance R_sStraight axis inductor L_dQuadrature axis inductor L_qAnd magnetic linkageThe difference between the direct axis inductance and the quadrature axis inductance is determined by different installation forms of the permanent magnets. The permanent magnets are embedded and mounted in salient pole machines, L_d≠L_qAnd surface-mounted is a non-salient pole machine, L_d＝L_q. The experimental result shows that the motor temperature can cause the motor parameters to change, but the amplitude is limited; during variable frequency speed regulation operation, R_sIt is mainly affected by skin effect and has larger range of variation; l is_dRelatively stable, L_qChanges that occur significantly with load;mainly influenced by the stability and demagnetization characteristics of the magnetic steel. In order to ensure the application effect of the second method, the motor parameters need to be identified online. Conventional parameter identification methods, e.g. model reference adaptation^[6]Extended kalman filter method^[7]Least square method^[8]And the like, the rotor position and the rotating speed signal are assumed to be free from errors, and the influence of the errors of the position and the rotating speed signal on the identification result under the condition without a speed sensor is not considered.

The requirements and sensitivities of different control methods on motor parameter information are different, the method is also the basis of online identification, and the structure of the speed-sensorless vector control system based on the virtual dc-qc coordinate system is shown in figure 1. The state observation module estimates the rotating speed and the position of the rotor according to the voltage output and the current feedback, and a motor model and parameters are required to be used; the phase-locking control module adjusts the rotating speed of the dc-qc coordinate system according to the position angle difference of the dc-qc coordinate system and the d-q coordinate system (rotor synchronous coordinate system); the rotating speed control module adjusts a voltage vector according to the deviation of the rotating speed and applies the voltage vector to a motor stator winding through space voltage vector modulation (SVPWM); the method is simple and feasible, and easy to realize at low cost, so that the method is widely applied. In fig. 1, a dotted line part is a parameter identification module of the present invention, which identifies motor parameters on line based on a motor model and state variable feedback.

And establishing a synchronous d-q coordinate system by taking the actual position of the rotor magnetic pole as a reference, wherein the model of the PMSM is as follows:

in the formula, p is a differential operator, u_dAnd u_qThe direct-axis and quadrature-axis voltages are respectively, ω is a rotor angular velocity (p θ), and θ is a rotor position angle.

As shown in FIG. 2, the vector control is performed at the estimated rotor position angle θ_cThe position angle of the expansion is different from that of the d-q coordinate system by delta theta_cTheta, α - β is a stationary coordinate system, the transformation relationship of the state variables is

The motor model becomes:

u_dc＝R_si_dc+pL_di_dc-ω_cL_qi_qc-(L_d-L_q)i_qcpΔθ+E_0xsinΔθ (3)

u_qc＝R_si_qc+pL_di_qc+ω_cL_qi_dc+(L_d-L_q)i_dcpΔθ+E_0xcosΔθ (4)

in the formula, ω_cDc-qc coordinate system angle for phase lock controlSpeed, theta_c＝∫ω_cdt。

In the steady state, the differential term can be ignored, and the voltage equation and the position angle deviation are respectively

u_dc＝R_si_dc-ω_cL_qi_qc+E_0xsinΔθ (6)

u_qc＝R_si_qc+ω_cL_qi_dc+E_0xcosΔθ (7)

System controlled phase locked loop section by adjusting omega_cThe Δ θ asymptotically converges to 0. If the motor parameters used in the formula (8) are accurate, the dc-qc coordinate system is coincided with the d-q coordinate system in a steady state, namely the estimation results of the rotating speed and the position are consistent with the true value; however, if the motor parameters have errors, the two coordinate systems will always have deviations.

Assuming the steady state deviation angle of the two coordinate systems as delta theta', defining the motor parameter error as follows:

ΔR_s＝R′_s-R_s,ΔL_d＝L′_d-L_d,ΔL_q＝L′_q-L_q

wherein R is_s、L_dAnd L_qIs a motor parameter true value, R'_s、L′_dAnd L'_qIs the set value of the motor parameter in the control program. Due to R_s、i_dcAnd Delta theta' are both small, and the following approximation is made to the formula (8),

the effect of the motor parameter error on the rotor position estimate is as follows:

it follows that though as the frequency increases, Δ R_sIncrease significantly due to skin effect, but Δ R_sInstead,/ω decreases, and i_dcUsually very small, so Δ R_sThe influence on the estimation result is limited; Δ L_qInfluence on the estimation result and the load current i_qcThe influence is small when the load is light and large when the load is heavy; Δ L_dHas no influence on the estimation result.

The errors of the motor parameters not only affect the dynamic characteristics of the control system, but also cause the deviation of vector control in a steady state, and the running efficiency of the motor is reduced. Therefore, online parameter identification is of great significance to guarantee and improve the performance of the speed sensorless control system.

The related documents are:

[1] wanxiu and permanent magnet motor [ M ] second edition, beijing, china electric power press, 2010;

[2]Guo Qingding,Luo Ruifu,Wang Limei.Neural Network Adaptive ObserverBased Position and Velocity Sensorless Control of PMSM.AMC'96,1996：41-46；

[3]Schroedl M.Sensorless Control of AC Machines at Low Speed andStandstill Based on the"INFORM″Method.IEEE Conference on IndustrialApplication,1996:270-277；

[4]Kiyoshi Sakamoto,Yoshitaka Iwaji,Tsunehiro Endo.Position and SpeedSensorless Control for PMSM Drive Using Direct Position ErrorEstimation.IECON’01:The 27^thAnnual Conference of the IEEE IndustrialElectronics Society,2001:1680-1685；

[5]Song Chi,Longya Xu.Position Sensorless Control of PMSM Based onNovel Sliding Mode Observer over Wide Speed Range.IEEE IPEMC 2006(3):1-7；

[6] adaptive online parameter identification of an embedded permanent magnet synchronous motor, a motor and control journal, 2010,14(4): 9-13;

[7]X.Jiang,Z.Zhang,P.Sun and Z.Zhu.Estimation of Temperature Rise inStator Winding and Rotor Magnet of PMSM Based on EKF.IEEE ICCEE 2010vol.8:24-27；

[8] liping, modeling and parameter identification of permanent magnet synchronous motors computer simulation, 2011,28(8):401 and 404.

Disclosure of Invention

The invention provides a method for identifying parameters of a Permanent Magnet Synchronous Motor (PMSM) on line, which is realized by a speed-sensor-free vector control system of a virtual dc-qc coordinate system and can effectively improve the performance of the control system.

The method is based on the on-line identification technology of the system steady-state characteristic parameters under small disturbance input and a plurality of running states.

R 'is used in the above formula (8)'_sAnd L'_qIn the phase-locked control, the steady-state voltage equation (3) of Δ θ → 0, dc-axis is forcibly locked

u_dc＝R′_si_dc-ωL′_qi_qc(10)

Therefore, the method cannot be used for parameter identification and belongs to the typical under-rank problem.

The steady-state equation (4) of the qc axis is affected by the motor parameter error and becomes

For non-salient PMSM, L_d＝L_qThe above formula (11) is changed to the formula (12)

Obviously, only in a single operation state, the online identification of multiple motor parameters cannot be realized.

The method of the invention comprises the following steps:

1) for non-salient PMSM (L) as described above with reference to_d＝L_q) Establishing a qc-axis steady-state equation (12) after considering the influence of motor parameter errors:

wherein R is_sIs stator winding resistance, L_dIs a straight-axis inductor and is characterized in that,is flux linkage, ω is rotor angular velocity, u_qcIs the qc axis voltage, i_qcIs the qc axis current, i_dcIs the dc axis current;

2) properly adjusting the running state of the motor for multiple times, recording steady-state values of all variables in a steady-state, substituting into the qc-axis steady-state equation (12) to obtain multiple groups of steady-state relations, and calculating the increment of the state variables to obtain R_s、L_d、Is identified value of

The specific implementation process is shown in an online identification flow chart 4:

21) keeping the rotation speed unchanged, and setting i_dc ^*Control is effected and the steady state values of the variables are recorded at 0ω₀、u_qc0And i_qc0Substituting formula (12) to obtain formula (13):

22) small change i_dc ^*Setting i_dc ^*＝Δi_dcNot equal to 0, control is carried out and the corresponding steady state value u is recorded_qc1And i_dc1Substituting formula (12) to obtain formula (14):

since the rotation speed and the load are not changed, i_qc1≈i_qc0L is obtained by subtracting the formula (13) from the formula (14)_dIs identified value of

23) Small amplitude regulation of omega^*Setting ω^*＝ω₀+ Δ ω, and hold i_dc ^*At 0, control is effected and the corresponding steady-state value ω is recorded₁、u_qc2And i_qc2Formula (12) may be substituted with formula (16):

in view of the short-term small-amplitude regulation of ω^*While, the load torque is not changed, i_qc2≈i_qc0Formula (13) × ω₁And formula (16) × omega₀Subtracting to obtain the Rs identification value

The formula (16) is subtracted from the formula (13) to obtainIs identified value of

Therefore, the online identification of the 3 motor parameters is realized.

For salient pole PMSM (L)_d≠L_q) In principle, the method cannot directly identify the quadrature axis inductance L_q. But may utilize quadrature inductance L provided by the motor manufacturer or obtained off-line identification_qAnd a direct axis inductor L_dThe ratio K ═ L is calculated in advance_q/L_dThen, the quadrature axis inductance L is estimated by the following formula_qIs identified value of

Wherein,the direct axis inductance L obtained in step 22)_dThe identification value of (1).

In the above step 22), the Δ i_dcOf rated current of the motor3-6%, preferably 5%.

In the step 23), the value of Δ ω is 1 to 3%, preferably 2%, of the operating angular velocity.

And the stability and accuracy of online identification are ensured based on the steady-state value calculation and feedback of low-pass filtering. The actual system is influenced by the control characteristics, the load characteristics and the PWM voltage action, and cannot be in a completely ideal steady state, and both the voltage and the current have certain pulsation. The change in the steady state value of each state variable caused by the small perturbation input is small and often lower than the pulsation amplitude. Therefore, if instantaneous value feedback of voltage and current under different operation states is adopted in online identification, the change of the steady-state value may be submerged by pulsation, and an accurate identification result cannot be obtained. A detailed motor and controller model is established for a 1.1kW/380V/3000rpm PMSM on a Matlab software platform, the switching frequency is 8kHz, the voltage vector control period is 250 mus, and the rotating speed control period is 5 ms. The simulated waveforms of the voltage and the current under the action of the small disturbance are shown in fig. 3, wherein (a) is an original waveform, and (b) is a waveform obtained by root filtering 1000 continuous sampling data. It can be seen that the ripple amplitude of the voltage is about 10V and the ripple amplitude of the current is greater than 1A. After low-pass filtering, effective information of steady-state value change under the action of small disturbance can be obtained. When i is_dc ^*When changing from 0 to 0.4A, u_qcAn increase of about 1.5V; when N is present_r ^*When changing from 1500 to 1530rpm, u_qcAn increase of about 2.9V.

Therefore, the voltage and current signals are subjected to low-pass filtering processing in steps 21), 22) and 23) of the embodiment, so that the stability and accuracy of online identification are further improved.

The permanent magnet synchronous motor online identification method only adds i in the original control strategy at proper time_dcAnd the short-time small disturbance of the omega instruction has no influence on the normal operation of the original control flow and system, does not increase any state variable, has simple and convenient algorithm, and can realize the PMSM motor parameter with insignificant salient pole effect or salient pole effect at low costAnd (4) identifying online.

Drawings

FIG. 1 is a block diagram of a velocity sensorless vector control system based on a virtual dc-qc coordinate system;

FIG. 2 is a relationship diagram of a synchronous coordinate system, wherein α - β is a stationary coordinate system, d-q is a rotor synchronous coordinate system, and dc-qc is a virtual rotor synchronous coordinate system;

FIG. 3 is a simulated waveform of voltage and current under small disturbance, (a) an original waveform, and (b) a low-pass filtered waveform;

FIG. 4 is a flow chart of the online identification of the present invention;

fig. 5 shows measured parameter variation characteristics of PMSM, (a) temperature characteristics, and (b) frequency characteristics.

Detailed Description

The method is suitable for the PMSM control system with insignificant salient pole or salient pole effect, and can be easily realized by adding a corresponding software program in the variable frequency control.

Example (b): the system of the embodiment is the same as that of fig. 1, and the system structure is as described above, and the system adopts a variable frequency controller and an eddy current brake load which are independently developed by the applicant. The tested motor is not an ideal non-salient PMSM, and the specification parameters of the tested motor are shown in a table 1:

TABLE 1 tested Motor parameters

Using LCR bridge and off-line parameter identification method (direct current voltammetry to measure R)_sAttenuation of direct currentMeasuring L by the method_dAnd L_qMeasurement of idling end voltage) And detecting the motor parameters and confirming the change characteristics of the motor parameters.

Fig. 5(a) shows measured temperature characteristics. With increasing motor temperature T, R_s、L_dAnd L_qIs enlarged, andslightly reduced. When the temperature is increased from 25 ℃ to 100 ℃, R_s、L_dAnd L_qRespectively increased by 7.5 percent, 7.5 percent and 15 percent,the reduction was 8.1%.

Fig. 5(b) shows the measured frequency characteristic (using an LCR bridge). With the increase of the frequency f of the injected current, the skin effect is obvious because the winding wire is thicker, and R_sIs significantly increased. When injecting 100Hz current, R_sIncreased by 129% compared with DC injection_dAnd L_qThe reduction was 7.7% and 7.3%, respectively.

Firstly, the method of the invention is simulated and confirmed. R_s、L_dAndthe true values of (c) are set to 2 Ω, 12mH, and 0.46Wb, respectively, and the initial setting values in the control program are 0.5 Ω, 13.3mH, and 0Wb, respectively. The simulation waveform of the parameter identification process is shown in FIG. 3, wherein i_dc ^*：0→0.4A；N_r ^*: 1500 → 1530 rpm; (a) showing the original waveform, and (b) showing the waveform after the low-pass filtering process. R_s、L_dAndthe identification results of (1.98 omega, 12.1mH and 0.459 Wb) respectively, and the errors of (0.8%, -4% and-0.2%) respectively.

The control chip of the frequency conversion controller adopts SH71253 of the Thysasa company, the switching frequency is 8kHz, the voltage vector control period is 250 mus, and the rotating speed control period is 5 ms. Adding the parameter on-line identification subprogram shown in FIG. 4 to the original control program, R_sAnd L_qThe initial settings of (1.7 omega) and (16.7 mH) are respectively, with a large deviation from the true value. Setting different running states and carrying out multiple identification. During the identification process, the current i_dcHas a disturbance input of 0.125A and a rotation speed N_rThe perturbation input of (2) is 24 rpm. Table 2 shows the results of some of the experiments.

TABLE 2 Online identification of experimental results

As can be seen from the experimental results shown in Table 2, R is increased with the increase in the number of revolutions and the increase in the load_sGradually increase, L_dIs slightly reduced, andthe identification result of (2) has a weak reduction trend due to salient pole effect approximation error, and is consistent with the offline detection result.

Claims (8)

1. A method for identifying parameters of a Permanent Magnet Synchronous Motor (PMSM) on line is a speed-sensorless vector control system based on a virtual dc-qc coordinate system, and comprises the following steps:

1) for non-salient PMSM, L_d＝L_qEstablishing a qc-axis steady-state equation (12) after considering the influence of motor parameter errors:

wherein R is_sIs stator winding resistance, L_dIs a direct-axis inductor, L_qIs a quadrature axis inductor,Is flux linkage, omega is rotor angular velocity, u_qcIs the qc axis voltage, i_qcIs the qc axis current, i_dcIs the dc axis current;

2) the running state of the motor is properly adjusted for many times, steady-state values of all variables in a steady-state are recorded, the steady-state values are substituted into a qc-axis steady-state equation (12) to obtain multiple groups of steady-state relations, and R is obtained through calculation of state variable increment_s、L_d、Is identified value of

2. The method of claim 1, further comprising: step 3) estimating quadrature axis inductance L_qFor salient pole PMSM, L_d≠L_qQuadrature axis inductance L_qIs identified value ofBy passingAnd estimating, wherein,for L obtained in step 2)_dK is L_q/L_d，L_q、L_dAnd providing or identifying the obtained quadrature axis inductance and direct axis inductance off line for a motor manufacturer.

3. A method as claimed in claim 1 or 2, characterized by: the specific implementation process of the step 2) is as follows:

21) the dc-axis current i is set while keeping the rotation speed constant_dcInstruction i of_dc ^*Control is effected and the steady state value ω of each variable is recorded at 0₀、u_qc0And i_qc0Substituting formula (12) to obtain formula (13):

22) change by a small margin i_dc ^*Setting i_dc ^*＝Δi_dc≠0，Δi_dcFor small disturbances of the dc-axis current command for parameter identification, control is carried out and the corresponding steady-state value u is recorded_qc1And i_dc1Substituting formula (12) to obtain formula (14):

since the rotation speed and the load are not changed, i_qc1≈i_qc0L is obtained by subtracting the formula (13) from the formula (14)_dIs identified value of

23) Instruction omega for slightly adjusting rotor angular speed omega^*Setting ω^*＝ω₀+ Δ ω, Δ ω is a minute disturbance amount of the rotor angular velocity command for parameter identification; and hold i_dc ^*At 0, control is effected and the corresponding steady-state value ω is recorded₁、u_qc2And i_qc2Substituting formula (12) to obtain formula (16):

in view of the short-term small-amplitude regulation of ω^*While, the load torque is not changed, i_qc2≈i_qc0Formula (13) × ω₁And formula (16) × omega₀Subtracting to obtain the Rs identification value

The formula (16) is subtracted from the formula (13) to obtainIs identified value of

。

4. The method of claim 3, wherein: and in the steps 21), 22) and 23), low-pass filtering is carried out on the voltage and current signals in the control process.

5. The method of claim 3, wherein: in step 22), the Δ i_dcThe value of (A) is 3-6% of the rated current of the motor.

6. The method of claim 3, wherein: in step 23), the value of Δ ω is 1-3% of the operating angular velocity.

7. The method of claim 3, wherein: Δ i in step 22)_dcThe value of (A) is 3-6% of the rated current of the motor; in the step 23), the value of delta omega is 1-3% of the operation angular speed.

8. The method of claim 7, further comprising: Δ i in step 22)_dcThe value of (a) is 5% of the rated current of the motor; in the step 23), the value of Δ ω is 2% of the operating angular velocity.