High-precision motor parameter identification method under debugging-free architecture

Abstract

A high-precision motor parameter identification method under a debugging-free architecture belongs to the field of motor system control technology. The present invention designs a safe and fast identification process. That is: offline open-loop identification of apparent resistance + preliminary setting of the controller + closed-loop identification of the maximum saturated incremental inductance. In the open-loop identification, the present invention proposes a variable-frequency injection open-loop identification strategy, which can effectively reduce the influence of delay, sampling error, and inverter nonlinearity. An algorithm for automatically searching the amplitude and frequency of the injection signal is designed. And for subsequent closed-loop control and identification, a high-bandwidth current controller is designed. In the closed-loop identification, by solving the inverter model parameters during step wave injection, resistance identification and voltage error compensation are realized simultaneously without additional injection, reducing the time and process complexity of identification. The influence of duty cycle on digital delay and skin effect in the high-frequency injection process are analyzed respectively, and they are compensated and corrected in the frequency model respectively, thereby improving the accuracy of identification.

Description

Technical Field

[0001] The present invention belongs to the technical field of motor system control, and in particular relates to a high-precision motor parameter identification method under a debugging-free architecture. Background Art

[0002] Permanent magnet synchronous motors are increasingly used due to their high efficiency, high power density and good dynamic performance. To meet the needs of different scenarios, advanced control schemes such as controller self-tuning, position sensorless design, and ampere maximum torque control are used. These algorithms are basically inseparable from the corresponding motor parameters, so the parameters should be obtained before the motor is started.

[0003] However, in actual use, users can only obtain very limited initial motor data, such as the voltage, current, rated speed and number of poles on the standard nameplate. Since the resistance, inductance and other parameters of the motor change nonlinearly with the motor operating conditions, they are usually not accurately given, which brings challenges to motor control and servo system debugging. In order to improve the versatility of the identification algorithm, it should take into account safety, rapidity, short displacement and accuracy.

[0004] Specific signal injection is a necessary means of parameter identification strategy. For stator resistance identification, the quotient of voltage and current must be calculated when the motor is in a stable state. In order to avoid the influence of dead zone effect, parasitic capacitance effect, pulsation error, etc. on resistance identification, the literature "Self-commissioning of permanent magnet synchronous machine drives atstandstill considering inverter nonlinearities,"Feb2014." identifies and compensates for inverter nonlinearities, and obtains stator resistance by calculating the gradient through linear regression. However, this further increases the time required for identification, which does not meet the requirements of fast debugging. The identification of the inductance parameters of permanent magnet synchronous motors can be roughly divided into using time domain data or in the frequency domain according to the calculation method. In the literature "Offline Parameter Self-Learning Method for General-Purpose PMSM Drives With Estimation Error Compensation,"Nov.2019.", the cross-coupling effect and magnetic saturation effect caused by inductance change are considered, and the dq axis inductance under various working conditions is obtained by injecting square wave voltage and differentiating the flux. However, this method greatly increases the amount of data calculation and storage in the parameter identification process, resulting in resource consumption and increased time costs. And the impact of parameter changes under different working conditions on the characteristics of the motor drive control system is not considered. In addition, the inductance identification strategy based on dq axis injection needs to be performed under the current closed loop, but the setting of the current closed loop controller and the rotor position depend on accurate motor parameters, and there is a contradiction between the premises of each other. In contrast, the identification based on open-loop signal injection is much simpler, but it is easy to trigger overcurrent protection and other problems when the motor has low impedance. However, its solution lacks consideration of rapidity and safety. In order to improve the safety and versatility of identification, a method of estimating the spatial inductance diagram by controllable sinusoidal voltage open-loop injection is proposed in the literature "Spatial Inductance Estimation for Current Loop Auto-Tuningin IPMSM Self-Commissioning," May 2020. However, this self-adjusting algorithm only uses a fixed step size to adjust the amplitude and frequency, which makes it difficult to quickly and accurately achieve the required working conditions.

[0005] In addition to the nonlinearity of the inverter, there are many other nonlinear factors that are important factors that cause identification errors. Digital control systems inevitably have execution delays and discrete sampling errors. At the same time, as the frequency of the injected signal increases, the impact of the skin effect is also increasing. In addition, as a nonlinear strongly coupled system, the motor is sensitive to changes in resistance and inductance parameters, which directly affects the robustness of the system. At present, most studies use strategies such as online parameter identification and controller correction. However, due to the lag in the results of identification and tuning, it will be too late to switch after online calculation.

[0006] In summary, in order to achieve high-quality intelligent control of servo drive control systems, academia and industry have carried out a lot of research on motor characteristic parameter identification and servo system debugging-free strategy. At present, the following problems still exist:

[0007] 1) Currently, there are two strategies for motor parameter identification: open-loop injection and closed-loop injection. When performed in the current closed-loop dq coordinate system, the identification accuracy is affected by the controller parameters, and the self-tuning of the controller parameters depends on the precise parameters of the motor. There is a contradictory relationship between the two. In the case of open-loop injection, for electromechanical systems with unknown parameters, unreasonable signal injection may cause overcurrent problems. The current identification algorithm is too simple and lacks consideration of the impact of parameter uncertainty on the identification strategy. It is difficult to meet the growing demand, and how to design an excellent identification process is still blank.

[0008] 2) As a nonlinear strongly coupled system, the motor has various environmental factors such as temperature, magnetic saturation, cross-coupling effect, skin effect, etc. that affect the main parameters and system characteristics. In addition, various nonlinear factors such as inverter, digital delay, discrete sampling, skin effect, etc. affect the system performance and identification quality. How to achieve fast, safe and accurate parameter identification while taking the above factors into consideration has become another difficult problem. Summary of the invention

[0009] In order to solve the above problems, the present invention further provides a high-precision motor parameter identification method under a debugging-free architecture.

[0010] The technical solution adopted by the present invention is:

[0011] The high-precision motor parameter identification method under the debugging-free architecture includes:

[0012] S1. Offline open loop identification of apparent resistance;

[0013] S2. Preliminary tuning of controller;

[0014] S3. Closed loop identification of maximum saturation incremental inductance.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] 1. In response to the above problem 1), the changing rules of motor parameters under different working conditions were analyzed, combined with the analysis of their impact on system characteristics. Considering the identification of unfamiliar motors and subsequent adjustment and control, the safety and stability of the system is the premise of design. Starting from the stability of the system, the unstable working condition of the system was established, and a safe and fast identification process was designed based on the working condition, namely: offline open-loop identification of apparent resistance + preliminary adjustment of the controller + closed-loop identification of the maximum saturated incremental inductance.

[0017] 2. In response to the above problem 2), in the open-loop identification, the errors of traditional identification calculations are analyzed, and a variable-amplitude frequency injection open-loop identification strategy is proposed, which can effectively reduce the influence of delay, sampling error, and inverter nonlinearity. At the same time, in order to achieve fast, safe and effective injection, an algorithm for automatically searching the amplitude and frequency of the injection signal is designed. In order to improve the subsequent identification accuracy and control performance, a high-bandwidth current controller design method is provided. In the closed-loop identification, in order to improve the accuracy of resistance identification, by solving the inverter model parameters during step wave injection, resistance identification and voltage error compensation are realized simultaneously without additional injection, reducing the identification time and process complexity. The influence of duty cycle on digital delay and the skin effect during high-frequency injection are analyzed respectively, and they are compensated and corrected in the frequency model respectively, which improves the accuracy of identification.

[0018] 3. The present invention provides a high-precision parameter identification technology under a debugging-free framework. This strategy takes into account and well meets the requirements of safety, rapidity and universality in practical engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] One or more embodiments are exemplarily described by corresponding drawings, which do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements, and the drawings do not constitute a scale limitation, and wherein:

[0020] Figure 1 It is a schematic diagram of an intelligent debugging-free architecture for a servo control system provided by an embodiment of the present disclosure;

[0021] Figure 2 It is a schematic diagram of a high-precision motor parameter identification method under a debugging-free architecture provided by an embodiment of the present disclosure;

[0022] Figure 3 is a flow chart for constructing a model of an electromechanical control system provided by an embodiment of the present disclosure;

[0023] Figure 4 is a schematic diagram of a time-varying characteristic model of a motor system parameter provided by an embodiment of the present disclosure;

[0024] Figure 5It is a schematic diagram of time-frequency domain fluctuations of a motor system characteristic provided by an embodiment of the present disclosure;

[0025] Figure 6 It is a framework diagram of a high-precision motor parameter identification method provided by an embodiment of the present disclosure;

[0026] Figure 7 It is a flow chart of a voltage adaptive variable amplitude frequency injection strategy provided by an embodiment of the present disclosure;

[0027] Figure 8 is a preliminary control block diagram of a controller provided by an embodiment of the present disclosure;

[0028] Fig. 9 It is an overall algorithm framework for system parameter identification provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] In order to be able to understand the features and technical contents of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0030] The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0031] Unless otherwise stated, the term "plurality" means two or more.

[0032] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.

[0033] The term "and / or" is a description of the association relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.

[0034] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.

[0035] In the present technical solution, an empirical formula refers to the need to seek an approximate expression of a function based on a set of experimental data in production practice. This approximate expression is called an empirical formula.

[0036] Combination Figure 1 As shown, the embodiment of the present disclosure provides an intelligent debugging-free architecture for a servo control system, including:

[0037] S1. Motor parameter identification strategy;

[0038] S2. Current loop setting;

[0039] S3. Arbitrary phase sequence detection and safety correction or position sensorless design;

[0040] S4. Servo system characteristics modeling;

[0041] S5. Speed ​​or position loop adjustment;

[0042] S6. Controller architecture optimization.

[0043] exist Figure 1 Under the debug-free framework constructed, combined with Figure 2 As shown, the embodiment of the present disclosure provides a high-precision motor parameter identification method under a debugging-free architecture, comprising:

[0044] S1. Offline open loop identification of apparent resistance;

[0045] S2. Preliminary tuning of controller;

[0046] S3. Closed loop identification of maximum saturation incremental inductance.

[0047] The high-precision motor parameter identification method provided by the embodiment of the present disclosure in the debugging-free framework is adopted to determine the most unstable situation when the motor is the controlled object by analyzing the influence of different environments and working conditions on the parameters. Based on the safety and rapidity of debugging-free, a general identification process is designed in a targeted manner. In the identification, an intelligent amplitude-frequency search injection algorithm that does not rely on the setting of initial parameters is designed, combined with compensation measures for inverter nonlinearity, digital delay, discrete sampling, skin effect, etc., while improving the identification accuracy, the compatibility and intelligence of the algorithm are improved, and the setting time is effectively shortened, thereby providing better support for the debugging-free implementation of the servo system.

[0048] Combination Figure 3 As shown, the process of constructing the model of the electromechanical control system includes:

[0049] S1. Establishing the permanent magnet synchronous motor transfer function of the control object;

[0050] S2. Establishing a parameter time-varying characteristic model based on the established transfer function;

[0051] S3. Establish an impact model of system characteristics based on the established time-varying characteristic model;

[0052] S4. Based on the established impact model, establish a parameter identification framework with safety and speed as the top priorities.

[0053] In this embodiment, the system control object includes a mechatronic system having a drive motor and an inverter, and the parameter formula is expressed as follows:

[0054]

[0055] In the formula, u dq is the dq axis voltage, i dq is the dq axis voltage, R s is the stator resistance, ω e is the electrical speed, ψ dq is the dq axis flux, which can be expressed as ψ d =L d_app i d+ ψ f , ψ q =L q_app i q , substituting the magnetic flux into the above formula, we get:

[0056]

[0057] Among them, L dq_inc is the dq-axis incremental inductance of the permanent magnet synchronous motor; L dq_app is the apparent inductance of the dq axis, ψ f is the permanent magnet flux of the motor.

[0058] In summary, if accurate motor control is to be performed, various motor parameters including apparent inductance and incremental inductance should be obtained at the same time. Although the incremental inductance curve under all working conditions can be identified, it greatly increases resource consumption and time cost, which is contrary to actual application requirements.

[0059] Therefore, this example establishes a parameter time-varying characteristic model based on the established transfer function. The main factors affecting the motor resistance value are temperature and frequency. For metal conductors, the higher the temperature, the greater the resistance; the higher the frequency, the greater the skin effect and the greater the resistance. The inductance of a permanent magnet synchronous motor is determined by the quotient of the electromagnetic field flux and the current in the equivalent circuit. During motor operation, due to the characteristics of electromagnetic materials, magnetic saturation effects and cross-coupling phenomena will occur. With the increase of magnetic field intensity, the degree of magnetic saturation increases, the magnetic permeability μ decreases, and then the equivalent inductance decreases. In addition, for built-in permanent magnet synchronous motors, since the direct and alternating axis magnetic circuits are not equal, the direct axis magnetic circuit has a large magnetic resistance and the cross axis magnetic circuit has a small magnetic resistance, resulting in a direct axis inductance L d <Q-axis inductance Lq . Figure 4 It shows the schematic diagram of the time-varying characteristic model of the motor system parameters, where: Figure 4 (a) shows the change of resistivity of copper with temperature. Figure 4 (b) is a schematic diagram of the influence of skin effect on motor characteristics. Figure 4 (c) is the effect of magnetic field intensity on magnetic flux density, Figure 4 (d) is the influence of cross-coupling effect.

[0060] Optionally, in order to establish the impact model of system characteristics, the resistance-inductance characteristics are changed based on the current closed-loop control characteristics. According to engineering experience, the closed-loop controller is designed according to the system phase margin of 50° and the amplitude margin of 8dB. The schematic diagram of the time-frequency domain fluctuation is as follows Figure 5 shown.

[0061] Figure 5 The left part of the diagram shows the system characteristics after the resistance changes. R12 , γ R12 They represent the gain margin and phase margin after the system changes. As shown in the figure, as the motor resistance increases, the system amplitude margin and phase margin both increase, and the temperature and frequency values ​​are positively correlated with the system stability. Taking all factors into consideration, resistance identification should be achieved through DC injection at the beginning of the system free debugging. Similarly, Figure 5 The blue and green curves in the left part are the system characteristics after the incremental inductance changes. Reducing the inductance will cause the system gain margin h L2 The apparent inductance determines the amount of feedforward compensation for the back EMF. A large amount of feedforward will cause a large overshoot and stabilization time. The time domain response is as follows: Figure 5 As shown in the right part of the figure. In actual engineering, a smaller feedforward amount is often used to ensure system stability. In summary, both incremental inductance and apparent inductance need to be identified under the minimum working condition to ensure the stability of the system during subsequent operation.

[0062] Combined with the analysis in the previous section, the present embodiment provides a method for obtaining the minimum Ld at the maximum magnetic saturation, and designing the controller with the identification value of this working condition. This value can also be used as a reference boundary for online controller correction to prevent system instability caused by unreasonable design. Considering that the incremental inductance under the maximum magnetic saturation working condition needs to be obtained under the current controllable state, the motor driver characteristics and the actual debugging-free requirements are comprehensively considered, combined with Figure 6 As shown, the embodiment of the present disclosure is Figure 3 Based on this, it provides a fast and safe intelligent debugging-free strategy, including:

[0063] S11. High-precision identification strategy for variable amplitude frequency;

[0064] S12. Open-loop adaptive security injection strategy;

[0065] S21. Controller parameter model method setting;

[0066] S22. Advanced feedforward link setting;

[0067] S31. Inverter compensation without additional injection;

[0068] S32.ZOH compensation strategy;

[0069] S33. Skin effect compensation strategy;

[0070] Optionally, the offline open-loop identification of apparent resistance inductance requires the establishment of a high-precision identification strategy and an adaptive safety open-loop injection strategy. In this embodiment, optionally, the sinusoidal voltage signal injected on the α-axis is U m e j(ωt+θu) , the response of the motor equivalent impedance model is expressed as:

[0071]

[0072] According to the above formula, L d and R s It can be expressed as:

[0073]

[0074] Among them, the amplitude I in the current response m , phase angle θ i It can be obtained through sampling to realize the calculation of motor parameters. However, in the actual signal injection process, it will be limited by many non-ideal factors such as inverter nonlinearity and digital control discretization. And as the sampling frequency decreases and the electrical frequency increases, the system error will be aggravated. In comparison, AD sampling has higher accuracy and the error caused by it is smaller. Therefore, this embodiment abandons data such as time and phase that are easily affected by discretization, and instead uses the amplitude difference of the signal to solve the parameters.

[0075] The inductor L d and stator resistance R s The expression of is summed up by square:

[0076]

[0077] Subtract the two sets of signal data with different frequencies and calculate the inductance as:

[0078]

[0079] Among them: U1, I1, ω1; U2, I2, ω2 are the voltage given amplitude, current response amplitude and frequency of two groups of signals injected with different frequencies.

[0080] In order to reduce the identification error caused by the introduction of equivalent impedance in the dead zone, the resistance value is calculated by subtracting two sets of signal data with the same frequency but different amplitudes, and the calculated inductance is substituted into the inductance to derive the identification formula of the stator resistance:

[0081]

[0082] Among them: U3, I3, ω3 are the voltage given amplitude, current response amplitude and frequency injected by signals with different amplitudes.

[0083] Optionally, to further reduce the error caused by factors such as dead zone, a signal extraction module such as a bandpass filter, PR, Goertzel algorithm or DFT may be added during fixed frequency injection, which essentially processes only the response signal with the same frequency as the injection.

[0084] The variable amplitude frequency injection open-loop identification strategy proposed above can effectively reduce the influence of delay, sampling error, and inverter nonlinearity. However, the identification strategy has certain requirements on the amplitude and frequency of the injected voltage. If the injected signal is too low, it will be difficult to calculate the truth due to the low signal-to-noise ratio, or it will be too high and cause overcurrent faults. In addition to the most basic requirement of ensuring the safety and effectiveness of identification, it is also necessary to identify the minimum inductance under the maximum current open-loop injection as described above. Therefore, on the basis of the open-loop identification strategy, according to actual application requirements, in order to quickly realize the maximum sinusoidal current response of the motor, this embodiment designs an automatic search injection strategy. The strategy can be summarized into four parts: fast search, step-by-step search, triggering overvoltage and variable frequency injection. The specific implementation process is as follows. Figure 7 shown.

[0085] In the face of unknown electromechanical systems, the number of unknown parameters that users input should be reduced, which improves the intelligence of the strategy while avoiding the harm to the system caused by setting errors as much as possible. Taking all factors into consideration, this embodiment only requires the user to set the current limit value I according to the limit value of the driver and motor and the working needs. lim As for other parameters, bus voltage U N Can be obtained by sampling; optionally, the initial injection amplitude U m0 Set to 2 times the tube voltage drop (IGBT tube voltage drop V CE Relatively fixed; MOSFET voltage drop V DS Simply press R dson *I load Calculation); The injection signal frequency f should be selected to be close to the reciprocal of the electrical time constant and much higher than the switching frequency. Optionally, this embodiment adopts the smaller value of 1 / 10 switching frequency and 100Hz. To reduce the error caused by transient response fluctuations, the period T m Set to 3 / f.

[0086] After the initial safety injection is completed, the injection signal amplitude and frequency need to be gradually increased to achieve the target operating condition. In order to complete the search as quickly as possible, the embodiment of the present application optionally abandons the traditional fixed amplitude growth strategy and chooses to increase the given value at an exponential rate of 2 until the current response amplitude I m Greater than 0.4*I max , at this time I m It will be at 0.4*I max and 0.8*I max The target operating condition is about to be reached. Exit the index search phase and set the next injection amplitude to:

[0087] U m =0.8*U m *I m / I Max

[0088] Where: I max In order to respond to the maximum value of the signal current, it can usually be set to the set current limit value I lim。

[0089] In order to obtain the data required for resistance identification, the amplitude of the injected signal is increased by a fixed amplitude after exponential search. m 1 / 20 of the response current amplitude I m Greater than 0.94*I max To obtain the data required for inductance identification, the command frequency f is further multiplied by 2 and injected again. At this time, the impedance of the controlled object is approximately doubled. In order to maintain a similar magnetic saturation level, the command amplitude U m Also need to multiply by 2.

[0090] In the actual signal injection process, it will be limited by the DC bus voltage, especially when the bus voltage is low and the resistance and inductance parameters are large, the output voltage per unit value is large. lim The maximum amplitude of the fundamental wave of the phase voltage output by the linear modulation inverter And provide margin for the final frequency conversion injection, set the limit voltage of open-loop iterative injection to U lim / 2.

[0091] In order to realize the subsequent closed-loop injection and identification, after completing the identification of the apparent parameters of the system, this embodiment uses the model method to realize the preliminary tuning of the controller.

[0092] In this embodiment, when the motor is stationary, the transfer function G of the controller of the current control system is established. OL (s), including:

[0093] The simplified model transfer function of the motor system is G M (s), the transfer function G of the proportional-integral controllerPI (s) and the transfer function G of the discrete holding equivalent model zoh (s). The transfer function G is established according to the following formula OL (s):

[0094]

[0095] Where: k p The parameter is the controller proportionality factor, k i The parameter of the controller is the integral coefficient, τ i is the time constant of the controller, equal to 1 / k i ; τ i is the time constant of the motor, equal to L d / R s ; T s is the discrete time parameter, ω is the signal frequency, and s is the Laplace transform factor.

[0096] Optionally, in an embodiment, some parameters of the controller may be preset as:

[0097] k p =0.4*L*f s k i =0.7*R s / L d

[0098] Considering the need to identify the inductance in a wide frequency range, this embodiment further reduces the inertia of the system through feedforward compensation. The controlled system can be simply approximated by a first-order low-pass system, so G M The inverse model of (s) is a first-order compound differential link. Considering that the differential term has an amplifying effect on noise, this embodiment chooses to use an advance link to replace the differential term. lead The equation for (s) is as follows:

[0099]

[0100] Where: G lead (s) is the phase-frequency function, s is the Laplace transform factor, α is the bandwidth factor, ω l is the center frequency. α determines the degree of the phase advance angle. Optionally, in the embodiment of the present application, α=10; ω l Set to 3kHz to increase the phase advance angle. Introduce the feedforward proportional factor k f To dynamically adjust the system characteristics, this value can be dynamically adjusted according to the response error, k f ∈[0.3,0.7]. Figure 8 A control block diagram for preliminary tuning of a controller is provided.

[0101] In summary, this embodiment uses open-loop adaptive high-frequency voltage injection and calculates the apparent motor parameters through current response. However, this method is not suitable for obtaining the incremental inductance under magnetic saturation conditions, and resistance identification will also be affected by the skin effect. Therefore, after completing the pre-design of the current controller, it is necessary to complete the required identification in closed-loop control mode. First, estimate the stator resistance. Then use the current feedback and the amplitude phase difference of the controller output voltage to calculate the incremental inductance under magnetic saturation.

[0102] Stator resistance identification is achieved by injecting a step wave with gradually increasing amplitude into the d-axis. Compared with the open-loop identification of stator resistance, this method can avoid the influence of skin effect while also observing the nonlinearity of the inverter. In order to reduce the complexity of the injection process and the identification time, this embodiment designs a method that does not require additional injection and calculates the inverter nonlinear model and voltage compensation while injecting resistance identification.

[0103] In order to achieve the maximum magnetic saturation identification condition, it is first necessary to inject a ramp / step wave into the current closed loop, and the motor resistance can be identified in this process. When the system response reaches a steady state, only the resistive part exists in the voltage equation, which can be expressed as:

[0104] u d =R s i d +Δu derr =R s i d +u inv +u mos

[0105] Where: u d is the motor d-axis voltage, i d is the d-axis current, u derr is the d-axis voltage error.

[0106] The non-ideal characteristics of the inverter will introduce errors in the motor parameter identification and affect the system performance. According to the error characteristics, it can be roughly divided into the tube conduction voltage drop u mos and other nonlinear characteristics such as dead zone, parasitic capacitance, etc. inv Optionally, in this embodiment, the voltage error Δu of the basic component xerr It can be expressed as:

[0107]

[0108] Among them, Δu xerr The x in the index represents the three-phase abc, that is, Δu aerr ,Δu berr ,Δu cerr ; k is a model parameter determined by the device characteristics and dead time, and ΔU is determined by the bus voltage and dead time.

[0109] After coordinate transformation, the voltage error expression in the dq axis system can be obtained as:

[0110]

[0111] Where: Δu derr , Δu qerr is the dq axis voltage error calculated by coordinate transformation.

[0112] In the resistance identification working condition of the embodiment, the electrical angle θ is 0°, I d ≈I a ≈-0.5*I b ≈-0.5*I c , so the above formula can be simplified to:

[0113]

[0114] Assume x = e kid / 2 , because i d >0, so x>1. The initial value of ΔU can be obtained from the bus voltage U dc and dead time T deadtime The product of is estimated. Rearranging the above formula gives:

[0115]

[0116] In order to facilitate calculation, ΔU x is to set the intermediate variable. Δ is the total discriminant of Shengjin formula.

[0117] Therefore, the dead zone parameter solution can be transformed into the following univariate cubic solution problem:

[0118] (ΔU x -1)x 3 +ΔU x x 2 +ΔU x x+(ΔU x +1)=0

[0119] Optionally, in the disclosed embodiment, the Shengjin formula may be used to solve the equation, and the discriminant thereof may be simplified to be expressed as:

[0120]

[0121] ΔU x =3ΔU derr / 4ΔU∈[0,1], the discriminant Δ is positive, and there is a unique positive real root:

[0122]

[0123] in:

[0124]

[0125] In the current range with large parasitic capacitance and dead zone effect, the influence on the resistance term is small, so the resistance is first calculated in the maximum current range, and then the inverter characteristic parameter k is calculated by the strategy described in this embodiment. A more accurate inverter nonlinear model can be obtained by multiple iterations.

[0126] Furthermore, after completing the closed-loop current identification and inverter nonlinear compensation, this embodiment identifies the incremental inductance under maximum magnetic saturation. In the digital control system of the permanent magnet synchronous motor drive, it appears as ZOH at the Nyquist frequency, and ZOH inevitably introduces delay and amplitude distortion. In the previous open-loop preliminary identification algorithm, this effect can be minimized by variable amplitude frequency injection identification. However, under fixed magnetic saturation conditions, the amplitude difference cannot be used for calculation. To solve this problem, this patent proposes a ZOH characteristic compensation strategy based on high-frequency signals.

[0127] In the embodiment of the present application, under a wide frequency sweep range, the transfer function H of the ZOH characteristic is established according to the following formula: ZOH (ω) and the compensation equation v r (t):

[0128]

[0129] In practical applications, T is the only unknown quantity in the above equation. s , its value is not fixed, about 1.5 sampling cycles, and is only an approximate value. As the duty cycle increases, the actual system delay will decrease. This is closely related to the motor parameters, voltage and current, so this embodiment completes the identification and compensation of ZOH under the same working conditions as the identification.

[0130] In this reference embodiment, the system characteristics of the controlled object are represented by the motor transfer function G M (s) and the transfer function of the ZOH characteristic, namely:

[0131]

[0132] The impedance-inductance characteristics of the injected data are compensated. Considering the change of the impedance-inductance characteristics under high-frequency saturation conditions, the system characteristics after compensation of the impedance-inductance are:

[0133]

[0134] in and represents the assumed system resistance value. In this embodiment, the motor parameter value identified by the open loop above is optionally used. can be ignored and simplified to:

[0135]

[0136] ∠G p ' l =-ωT s / 2-π / 2

[0137] Where: G′ pl is the characteristic transfer function of ZOH identified, ω is the frequency, T s is the discrete time parameter; is the inductance value identified in S11.

[0138] The above formula only exists and T s The two unknown quantities can be easily fitted by the phase angle curve to the influence of discrete control. It should be noted that the higher the frequency, the smaller the influence of the resistance-inductance error, and the more accurate the value. To avoid signal distortion at high frequencies, this embodiment increases the system bandwidth as much as possible in the initial design of the controller.

[0139] Furthermore, during the high-frequency current injection process, the skin effect causes the high-frequency current to tend to the surface of the conductor, and the equivalent resistance increases accordingly. Especially under the condition of maximum current, the higher current density further exacerbates this phenomenon. In addition, the skin effect is also reflected on the external connecting wires of the motor.

[0140] The increase in resistance causes the mismatch of the original identification model parameters. The voltage drop of the actual inductor is less than the theoretical voltage drop, causing the calculated inductance to be greater than the actual value. To improve this problem, this article introduces an intersection correction link based on the inductance calculation in the previous section. For a purely inductive link, the phase angle should be -90°. For a frequency band other than -90° (due to the resistance-inductance characteristic, it is theoretically greater than -90°), the following formula is used for correction:

[0141]

[0142] Where: L M For the initial identification of the inductance, L' M is the inductance corrected for the skin effect. FFT The phase angle calculated for the Fourier transform.

[0143] It should be noted that both the delay correction in the previous section and the skin effect correction in this section use the phase angle in the frequency domain. Therefore, it is necessary to fit the ZOH model at high frequencies first, and then perform skin effect correction to improve the identification accuracy.

[0144] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A high-precision motor parameter identification method under a debugging-free architecture, characterized in that: include: S1. Offline open loop identification of apparent resistance; S2. Preliminary controller tuning; S3. Closed loop identification of maximum magnetic saturation incremental inductance; The control object of the identification method includes a mechatronic system having a drive motor and an inverter; Establishing a transfer function of the control object includes: S1. Establishing the permanent magnet synchronous motor transfer function of the control object; S2. Establishing a parameter time-varying characteristic model based on the established transfer function; S3. Establish an impact model of system characteristics based on the established time-varying characteristic model; S4. Based on the established impact model, establish a parameter identification framework with safety and speed as the top priorities; The offline open-loop identification of apparent resistance in S1 includes the following steps: S11. High-precision identification strategy for variable amplitude frequency; In order to avoid the influence of delay, sampling error and inverter nonlinearity on the identification accuracy, the stator resistance and apparent inductance are identified by injecting open-loop voltage signals of different amplitudes and frequencies according to the following formula: Where: L d is the direct-axis inductance, R s is the stator resistance, U1, I1, ω1; U2, I2, ω2; U3, I3, ω3 are the voltage given amplitude, current response amplitude and frequency of the injection signal with different frequency amplitudes respectively; S12. Open-loop adaptive security injection strategy: In order to quickly and safely achieve the maximum sinusoidal current response to the motor, based on the open-loop identification strategy, self-adaptation is achieved through four parts: fast search, step-by-step search, overvoltage triggering and variable frequency injection.

2. The high-precision motor parameter identification method under the debugging-free architecture according to claim 1 is characterized in that: include: The model method is used in S2 to implement preliminary tuning of the controller, including the following steps: S21. Preliminary setting of controller parameters; The controller model method setting of the current control system is established according to the following formula: k p =0.4*L*f s k i =0.7*R s / L d Where: k p is the controller proportional coefficient, k i is the integral coefficient of the controller, f s is the sampling frequency; S22. Advanced feedforward link setting, The high bandwidth feedforward compensation link is established according to the following formula: Where: G lead (s) is the phase-frequency function, s is the Laplace transform factor, α is the bandwidth factor, ω l is the center frequency, and α determines the degree of the phase advance angle.

3. The high-precision motor parameter identification method under the debugging-free architecture according to claim 1 is characterized in that: include: The closed-loop identification of incremental parameters in S3 includes the following steps: S31. Steady-state stator resistance identification and inverter compensation without additional injection; S32.ZOH compensation strategy; S33. Skin effect compensation strategy.

4. The high-precision motor parameter identification method under the debugging-free architecture according to claim 3 is characterized in that: In S31, the inverter nonlinear compensation link is established according to the following formula: Where: ΔU x For the convenience of calculation, the intermediate variable is set to 3Δu derr / 4ΔU; Δ is the total discriminant of Shengjin formula.

5. The high-precision motor parameter identification method under the debugging-free architecture according to claim 3 is characterized in that: In S32, the influence of ZOH is identified and compensated according to the following formulas: Identification: compensate: Where: G′ pl is the characteristic transfer function of ZOH identified, ω is the frequency, T s is the discrete time parameter, is the inductance value identified in S11.

6. The high-precision motor parameter identification method under the debugging-free architecture according to claim 3 is characterized in that: In S33, the inductance identification deviation caused by the skin effect is compensated according to the following formula: Where: L M For the initial identification of the inductance, L' M is the inductance corrected for skin effect, θ FFT The phase angle calculated for the Fourier transform.

Images