Motor parameter online identification method based on improved golden section method

Abstract

The invention provides a motor parameter online identification method based on an improved golden section method, and belongs to the technical field of motor parameter identification, and the method comprises the steps: S1, obtaining a residual function of a recursive least square method with a forgetting factor when parameter identification is not carried out on a motor model for the first time; s2, detecting whether the residual function exceeds a preset threshold value or not, and executing a step S3 when the residual function exceeds the preset threshold value; s3, taking the residual function as a target function, and adjusting the forgetting factor by using a golden section method to obtain an optimized forgetting factor; and S4, carrying out motor parameter online identification by using a recursive least square method with the optimized forgetting factor. The method has the advantages that the forgetting factor is dynamically adjusted through the golden section method, the residual function serves as the target function, the optimal solution of the forgetting factor under the corresponding working condition can be found when the system changes rapidly, the tracking speed is increased, the steady-state precision is further improved, and therefore the overall performance of parameter identification is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motor parameter identification, and particularly relates to a motor parameter online identification method based on an improved golden section method. BACKGROUND

[0002] Permanent magnet synchronous motor has the advantages of high efficiency, high power density and high dynamic performance, and thus shows significant superiority in various application scenarios requiring the use of motors. In recent years, the application demand for permanent magnet synchronous motors in the electric drive field has shown an increasing trend year by year.

[0003] In the permanent magnet synchronous motor control system, the permanent magnet synchronous motor is easily disturbed by external factors during operation, thereby causing the parameters to change. For example, load disturbance will cause the working load borne by the motor to fluctuate, thereby affecting the electromagnetic and mechanical characteristics of the motor and significantly affecting the motor parameters; motor temperature rise will cause the physical properties of the motor internal materials to change, such as increased winding resistance, which will also have a non-negligible impact on the motor parameters. The implementation of various advanced control strategies and fault diagnosis functions of the permanent magnet synchronous motor cannot be achieved without the motor parameters as an important reference. Therefore, once the parameter update lags, the dynamic performance of the motor will decline. In view of this, obtaining accurate motor parameters is of great significance for achieving high-performance control and online fault diagnosis of the motor.

[0004] In order to solve the problem of online parameter identification of the motor, many scholars at home and abroad have proposed a plurality of parameter identification algorithms, including model reference adaptive method, extended Kalman filter algorithm, recursive least squares method and artificial intelligence algorithm. However, these methods still face many challenges in practical application, specifically poor convergence, poor robustness, large linear error, complex calculation, sensitivity to initial value, too long training time and the like. Moreover, in many application scenarios of the permanent magnet synchronous motor, the computing capacity of the main control chip is often very limited. Therefore, many complex artificial intelligence algorithms may not be suitable for these scenarios. This requires that the algorithm used must take into account simplicity and efficiency.

[0005] Recursive least squares method has been widely used in motor parameter identification due to its simple principle, small amount of calculation and accurate results. However, the algorithm has a slow convergence speed in complex motor operating environment, which makes it unable to track the changes of motor parameters in time. In order to solve this problem, scholars have proposed a least squares method with forgetting factor. However, the selection of forgetting factor often affects the accuracy of the identification result. If the forgetting factor is not properly selected, it may cause the algorithm to forget the historical data or respond to new data in time, thereby affecting the accuracy of the identification result. Therefore, how to select the appropriate forgetting factor for different working conditions has become one of the key links of algorithm improvement. SUMMARY

[0006] In order to solve the above technical problems, the application provides an improved golden section method-based motor parameter online identification method, which is based on a recursive least squares method with a forgetting factor, and the improved golden section method is used to dynamically change the size of the forgetting factor, improve the convergence speed of the algorithm, and make it have strong tracking performance.

[0007] The technical problems solved by the application can be implemented by the following technical solutions:

[0008] An improved golden section method-based motor parameter online identification method comprises the following steps:

[0009] Step S1, when performing parameter identification on a motor model for the first time, obtaining a residual function of a recursive least squares method with a forgetting factor;

[0010] Step S2, detecting whether the residual function exceeds a preset threshold, and executing step S3 when the residual function exceeds the preset threshold;

[0011] Step S3, using the golden section method to adjust the forgetting factor by taking the residual function as a target function, to obtain an optimized forgetting factor;

[0012] Step S4, performing motor parameter online identification by using the recursive least squares method with the optimized forgetting factor.

[0013] Preferably, the step S1 further comprises the following steps:

[0014] Step S01, determining whether it is the first time to perform parameter identification on a motor model, and executing the step S1 when it is determined that it is not the first time to perform parameter identification on a motor model, and executing step S02 when it is determined that it is the first time to perform parameter identification on a motor model;

[0015] Step S02, performing motor parameter online identification by using the recursive least squares method with a forgetting factor, and executing the step S1.

[0016] Preferably, the recursive least square method with forgetting factor is:

[0017]

[0018] wherein, represents a parameter matrix to be identified; k represents a time instant; K represents a gain matrix of a parameter identification model; y represents an output matrix of the parameter identification model; φ T represents a transpose matrix of a data collection matrix φ; P represents a covariance matrix; λ represents a forgetting factor.

[0019] Preferably, the residual function is:

[0020]

[0021]

[0022] wherein, ε(k) represents the residual function, k represents a time instant; y represents an output matrix of a parameter identification model; φ T represents a transpose matrix of a data collection matrix φ; represents a parameter matrix to be identified; represents a transpose matrix of a data collection matrix φ T and a parameter matrix to be identified .

[0023] Preferably, the step S2 further comprises: setting the forgetting factor as 1 and performing the step S4 when the residual function does not exceed the preset threshold.

[0024] Preferably, the step S3 comprises

[0025] Step S31, determining an initial first golden section point λ1 and a second golden section point λ2 using a golden section method according to an optimization interval of the forgetting factor;

[0026] Step S32, obtaining a first residual function ε(λ1) corresponding to the first golden section point and a second residual function ε(λ2) corresponding to the second golden section point based on the recursive least square method with forgetting factor;

[0027] Step S33, comparing the first residual function ε(λ1) and the second residual function ε(λ2), and updating the optimization interval of the forgetting factor according to a comparison result;

[0028] Step S34, repeating the steps S31-S33 according to the updated optimization interval of the forgetting factor until a length of the optimization interval of the forgetting factor is less than a preset length threshold or a preset iteration number is reached, to obtain the optimized forgetting factor.

[0029] Preferably, the first golden section point λ1 is:

[0030] λ1=λ min +(1-φ)(λ max -λ min )

[0031] Among them, λ min represents the lower limit of the forgetting factor; λ max represents the upper limit of the forgetting factor; φ represents the golden ratio;

[0032] The second golden section point λ2 is:

[0033] λ2=λ min +φ(λ max -λ min ).

[0034] Preferably, in step S33, when the comparison result indicates that the first residual function ε(λ1) is smaller than the second residual function ε(λ2), the upper limit value of the forgetting factor in the optimization interval of the forgetting factor is updated to the second golden section point λ2; and

[0035] When the comparison result indicates that the first residual function ε(λ1) is greater than the second residual function ε(λ2), the lower limit value of the forgetting factor in the optimization interval of the forgetting factor is updated to the first golden section point λ1.

[0036] Preferably, the optimized forgetting factor is an average value of a lower limit value of the forgetting factor and an upper limit value of the forgetting factor in the optimization interval of the updated forgetting factor.

[0037] Preferably, in the motor model, the d-axis current is controlled with reference to a given value i d =0, a square wave signal is injected into the d-axis current reference given value.

[0038] The advantages or beneficial effects of the technical solution of the present invention are:

[0039] The present invention dynamically adjusts the forgetting factor through the golden section method and takes the residual function as the objective function. It can find the optimal solution of the forgetting factor under the corresponding working conditions when the system changes rapidly, accelerate the tracking speed, and further improve the steady-state accuracy, thereby improving the overall performance of parameter identification. It provides reliable theoretical support and practical guarantee for the high-performance control of the motor system and significantly improves the dynamic performance of the motor control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A system topology diagram of online motor parameter identification based on the improved golden section method in a preferred embodiment of the present invention;

[0041] Figure 2 For the preferred embodiment of the present application, the flowchart of the motor parameter online identification method based on the improved golden section method is shown in the figure.

[0042] Figure 3 For the preferred embodiment of the present application, the flowchart of the online identification method is shown in the figure.

[0043] Figure 4 For the preferred embodiment of the present application, the flowchart of step S3 is shown in the figure.

[0044] 1, speed PI module; 2, current PI module; 3, Park inverse transformation module; 4, SVPWM module; 5, three-phase inverter module; 6, permanent magnet synchronous motor; 7, Clark coordinate transformation module; 8, Park coordinate transformation module; 9, position detection module; 10, motor parameter identification module based on golden section method. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0046] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0047] The present application will be further described below with reference to the drawings and specific embodiments, but not as a limitation of the present application.

[0048] In the preferred embodiment of the present application, in view of the above problems existing in the prior art, for the permanent magnet synchronous motor required by advanced power equipment such as new energy vehicles and robots, a motor parameter online identification method based on the improved golden section method dynamic forgetting factor recursive least square method is provided. The forgetting factor recursive least square method is used to sample the current and voltage values i a , i b , i c , u a , u b , u c as input, and the estimated parameters R s , L d , L q , ψ f, the improved golden section method is used to take the residual function of the forgetting factor recursive least square method as a target function, a residual threshold is set, the system parameter state is changed, the algorithm is started, the adjustable range of the forgetting factor is taken as an interval boundary, the optimal solution of the forgetting factor under the corresponding working condition is found, the dynamic response characteristic of the least square parameter identification algorithm is enhanced, and the least square parameter identification result under the optimal forgetting factor is obtained. The application aims at dynamically adjusting the parameters required by the control algorithm in complex external disturbance, so as to improve the dynamic performance of motor driving.

[0049] As Figure 1 The system topology diagram of the motor parameter online identification based on the improved golden section method is shown. The system comprises a speed PI (proportion-integral) module 1, a current PI module 2, a Park inverse transformation module 3, an SVPWM (Space Vector Pulse Width Modulation) module 4, a three-phase inverter module 5, a PMSM (Permanent Magnet Synchronous Motor) 6, a Clark coordinate transformation module 7, a Park coordinate transformation module 8, a position detection module 9 and a motor parameter identification module 10 based on the golden section method.

[0050] The speed PI module 1 is used for outputting the reference value i ref of the q-axis current of the PMSM according to the difference between the input target speed ω e and the actual speed ω qref .

[0051] The current PI module 2 comprises two PI sub-modules. Wherein,

[0052] The first PI sub-module is used for outputting the q-axis voltage reference given value u qref according to the reference value i q of the q-axis current and the q-axis current reference given value i qref .

[0053] The second PI sub-module is used for outputting the d-axis voltage reference given value u dref according to the reference value i dref of the d-axis current and the d-axis current reference given value i d .

[0054] The Park inverse transformation module 3 is used for performing Park inverse transformation processing to obtain the alpha-axis voltage reference given value u e according to the input q-axis voltage reference given value u qref , the d-axis voltage reference given value u dref and the motor electric angle θqref , β-axis voltage reference given value u dref .

[0055] Among them, the mathematical formula of Park inverse transform is:

[0056]

[0057] SVPWM module 4, connected to the above Park inverse transformation module 3, is used to set the α-axis voltage reference value u qref , β-axis voltage reference given value u dref Perform space vector pulse width modulation and output the control signal of the inverter switch.

[0058] The three-phase inverter module 5 is connected to the above-mentioned SVPWM module 4 and is used to control the switching state of the three-phase inverter according to the control signal of the inverter switch and output the PWM voltage waveform required by the permanent magnet synchronous motor 6.

[0059] Clark coordinate transformation module 7 is used to convert the a-axis current reference value i in the natural coordinate system into a , b-axis current reference given value i b , c-axis current reference given value i c , a-axis voltage reference given value u a , b-axis voltage reference given value u b , c-axis voltage reference given value u c , perform Clark coordinate transformation and output the reference given value i of the α-axis current in the stationary coordinate system α ,β-axis current reference given value i β 、α-axis voltage reference given value u α , β-axis voltage reference given value u β .

[0060] Park coordinate transformation module 8 is used to convert the α-axis current reference given value i in the stationary coordinate system into α ,β-axis current reference given value i β 、α-axis voltage reference given value u α , β-axis voltage reference given value u β Perform Park coordinate transformation and output the d-axis current reference given value i in the rotating coordinate system d , q-axis current reference given value i q , d-axis voltage reference given value u d , q-axis voltage reference given value u q .

[0061] The position detection module 9 includes a position sensor and a speed sensor provided on the permanent magnet synchronous motor 6, and is used to collect the speed signal, angle signal and motor electrical angle θ of the permanent magnet synchronous motor 6.e The actual rotation speed ω e may be obtained by derivation of the motor electric angle θ e , i.e. ω e = dθ e / dt.

[0062] The motor parameter identification module 10 based on the golden section method applies a recursive least square motor parameter identification algorithm based on the improved golden section method to dynamically adjust the forgetting factor, uses the recursive least square method with the forgetting factor to identify the input d-axis current reference given value i d , q-axis current reference given value i q , d-axis voltage reference given value u d , q-axis voltage reference given value u q , and actual rotation speed ω e , and identifies the motor parameter identification value, including the stator resistance R s , d-axis inductance L d , q-axis inductance L q , and the magnetic flux coupled by the rotor permanent magnet on the stator Ψ f .

[0063] The motor parameter online identification method based on the improved golden section method applied by the motor parameter identification module 10 based on the golden section method of the embodiment of the application is described below, as shown in Figure 2 and Figure 3 , the method comprises the following steps:

[0064] Step S01, determining whether it is the first time to identify the motor model parameters, and executing step S1 when it is determined that it is not the first time to identify the motor model parameters, and executing step S02 when it is determined that it is the first time to identify the motor model parameters;

[0065] Step S02, using the recursive least square method with the forgetting factor to identify the motor parameters online, and executing step S1.

[0066] Step S1, when it is not the first time to identify the motor model parameters, obtaining the residual function of the recursive least square method with the forgetting factor;

[0067] Step S2, detecting whether the residual function exceeds the preset threshold, and executing step S3 when the residual function exceeds the preset threshold;

[0068] Step S3, using the golden section method to adjust the forgetting factor by taking the residual function as the objective function, to obtain the optimized forgetting factor;

[0069] Step S4, using the recursive least square method with the optimized forgetting factor to identify the motor parameters online.

[0070] Specifically, the traditional dynamic forgetting factor recursive least square parameter identification algorithm has the advantages of small calculation amount and accurate result, but cannot quickly converge in the case of parameter change, and cannot balance accuracy and good tracking performance.

[0071] In the embodiment, the golden section method is improved to dynamically adjust the forgetting factor, the residual function is taken as the target function, and the residual threshold is set, so that the optimal solution of the forgetting factor under the corresponding working condition can be found when the system changes quickly, the tracking speed is accelerated, and the overall performance of parameter identification is improved, thereby providing reliable theoretical support and practical guarantee for high-performance control of the motor system, and significantly improving the dynamic performance of the motor control system.

[0072] Specifically, when the method of the embodiment of the application is applied, firstly, it is judged whether it is the first time parameter identification; if it is the first time parameter identification, a parameter identification is first performed, otherwise, residual function threshold comparison is performed.

[0073] In the first time parameter identification, the estimated value of the parameter to be identified and the covariance matrix P(0) are initialized, and the voltage and current reference given values i d , i q , u d , u q and the actual speed ω e in the dq rotating coordinate system are input.

[0074] The recursive least square parameter method with forgetting factor is used, the forgetting factor λ(k) initially set, the parameter matrix to be identified at k-1 time and the covariance matrix P(k-1) at k-1 time are used, and the parameter matrix to be identified is updated according to the current collected data φ T (k). The recursive least square method with forgetting factor is expressed as:

[0075]

[0076] wherein, represents the parameter matrix to be identified; k represents the time; K represents the gain matrix of the parameter identification model; y represents the output matrix of the parameter identification model; φ T represents the transpose matrix of the data collection matrix φ; P represents the covariance matrix; λ represents the forgetting factor.

[0077] After the initial calculation is completed, the residual function ε(k) is calculated:

[0078]

[0079] wherein, ε(k) represents a residual function, k represents a time; y represents an output matrix of a parameter identification model; φ T represents a transpose matrix of a data collection matrix φ; represents a parameter matrix to be identified; represents a transpose matrix of a data collection matrix φ T and a parameter matrix to be identified .

[0080] Subsequently, the calculated residual function ε(k) is compared with a preset threshold value:

[0081] If the residual function ε(k) is less than the preset threshold value, it is proved that the system is stable, and the forgetting factor is set to 1, so that the algorithm has good stability.

[0082] If the residual function ε(k) is greater than the preset threshold value, it is proved that the system has not yet stabilized, and at this time the golden section method is used to dynamically adjust the genetic factor, so that the parameters converge quickly.

[0083] Using the golden section method, first, the optimal interval [λ min ,λ max ] of the forgetting factor is determined, wherein λ min represents the lower limit value of the forgetting factor; and λ max represents the upper limit value of the forgetting factor.

[0084] Then, the improved golden section method is used to calculate the initial first golden section point λ1 and the second golden section point λ2:

[0085] λ 1＝ λ min +(1-φ)(λ max -λ min )

[0086] λ 2＝ λ min +φ(λ max -λ min );

[0087] wherein, λ min represents the lower limit value of the forgetting factor; λ max represents the upper limit value of the forgetting factor; φ represents the golden section ratio, φ = 0.618; λ1 represents the first golden section point; and λ2 represents the second golden section point.

[0088] The residual function ε(k) of the recursive least squares method with the forgetting factor is taken as the objective function, and the convergence precision ε and the maximum iteration number N max , wherein N is the data length:

[0089] For each λ i, i = 1, 2, run the recursive least square method with the forgetting factor, and calculate the residual function ε(λ i ) by using the residual function calculation formula to obtain the first residual function ε(λ1) and the second residual function ε(λ2).

[0090] Further, the ε(λ1) and the ε(λ2) are compared.

[0091] If the ε(λ1) < the ε(λ2), the upper limit value λ max of the forgetting factor in the optimization interval is updated to the second golden section point λ2, and the updated optimization interval of the forgetting factor is [λ min , λ2];

[0092] On the contrary, if the ε(λ1) ≥ the ε(λ2), the lower limit value λ min of the forgetting factor in the optimization interval is updated to the first golden section point λ1, and the updated optimization interval of the forgetting factor is [λ1, λ max ].

[0093] Subsequently, the repeated calculation of the updated optimization interval of the forgetting factor is searched until the interval length is less than the convergence precision ε or the maximum iteration number N max is reached, and the optimal forgetting factor is obtained. The optimal forgetting factor is the average value of the updated lower limit value and the upper limit value of the forgetting factor.

[0094]

[0095] In the formula, λ* represents the optimal forgetting factor, that is, the optimized forgetting factor.

[0096] As a preferred embodiment, the step S2 further comprises: setting the forgetting factor to 1 when the residual function does not exceed the preset threshold, and executing the step S4.

[0097] Specifically, in the embodiment, the value of the residual function ε(k) is continuously monitored during the operation of the motor. When it is monitored that the value of the residual function ε(k) is less than the preset threshold, it indicates that the system state is stable.

[0098] When the system state is stable, the forgetting factor is set to 1, which can further improve the steady-state accuracy and make the system have good stability in the steady state, so that the system can have good tracking performance while ensuring accuracy.

[0099] As a preferred embodiment, the step S3 comprises Figure 4

[0100] ​Step S31, according to the optimization interval of the forgetting factor, using the golden section method, determine the initial first golden section point λ1 and the second golden section point λ2;

[0101] Step S32, based on the recursive least squares method with forgetting factor, obtain the first residual function ε(λ1) corresponding to the first golden section point and the second residual function ε(λ2) corresponding to the second golden section point;

[0102] Step S33, compare the first residual function ε(λ1) and the second residual function ε(λ2), and update the optimization interval of the forgetting factor according to the comparison result;

[0103] In step S33, when the comparison result indicates that the first residual function ε(λ1) is less than the second residual function ε(λ2), the upper limit value of the forgetting factor in the optimization interval of the forgetting factor is updated to the second golden section point λ2; and

[0104] When the comparison result indicates that the first residual function ε(λ1) is greater than the second residual function ε(λ2), the lower limit value of the forgetting factor in the optimization interval of the forgetting factor is updated to the first golden section point λ1.

[0105] Step S34, repeat steps S31-S33 according to the updated optimization interval of the forgetting factor, until the length of the optimization interval of the forgetting factor is less than a preset length threshold or reaches a preset iteration number, to obtain an optimized forgetting factor.

[0106] Wherein, the optimized forgetting factor is the average value of the lower limit value and the upper limit value of the forgetting factor in the updated optimization interval of the forgetting factor.

[0107] Specifically, in the present embodiment, first, according to the optimization interval of the forgetting factor, determine the upper limit value and the lower limit value of the forgetting factor; then, based on the determined upper limit value and the lower limit value of the forgetting factor, using the golden section method, determine the initial first golden section point λ1 and the second golden section point λ2.

[0108] Run the recursive least squares method with forgetting factor;

[0109] Set the forgetting factor to the first golden section point λ1 to obtain the first residual function ε(λ1).

[0110] Similarly, set the forgetting factor to the second golden section point λ2 to obtain the second residual function ε(λ2).

[0111] Then, compare the obtained first residual function ε(λ1) and the second residual function ε(λ2).

[0112] If ε(λ1) < ε(λ2), the upper limit value of the optimization interval of the forgetting factor is updated, that is, the upper limit value λ max of the forgetting factor in the optimization interval of the forgetting factor is updated to the second golden section point λ2, and the updated optimization interval of the forgetting factor is [λ min , λ2].

[0113] Conversely, if ε(λ1) ≥ ε(λ2), the lower limit value of the optimization interval of the forgetting factor is updated, that is, the lower limit value λ min of the forgetting factor in the optimization interval of the forgetting factor is updated to the first golden section point λ1, and the updated optimization interval of the forgetting factor is [λ1, λ max ].

[0114] Specifically, if the updated optimization interval of the forgetting factor is [λ min , λ2], the optimized forgetting factor λ * = (λ min + λ2) / 2.

[0115] If the updated optimization interval of the forgetting factor is [λ1, λ max ], the optimized forgetting factor λ * = (λ1+ λ max ) / 2.

[0116] As a preferred embodiment, in the motor model, the d-axis current reference given value i d = 0, a square wave signal is injected into the d-axis current reference given value.

[0117] Specifically, the parameter identification is performed in combination with the mathematical model of the permanent magnet synchronous motor 6. The cross-axis voltage equation set of the permanent magnet synchronous motor 6 is:

[0118]

[0119] In order to solve the underdetermined problem of the equation set, a high-frequency square wave signal is injected into the motor direct-axis current reference value for i d = 0 control.

[0120] When the motor is stably running at a constant signal, the current differential term in the formula can be ignored.

[0121] At two stable states (denoted as k1 and k2) of the high-frequency square wave signal, two sets of steady-state data can be obtained by collecting motor running data.

[0122] The above cross-axis voltage equation set of the permanent magnet synchronous motor 6 is discretized. Discretization is a method of converting a continuous mathematical model into a discrete form. After discretization, the discretized voltage equation set is obtained:

[0123]

[0124] The recursive least square parameter identification algorithm model with a forgetting factor is obtained by arranging the above voltage equation group:

[0125]

[0126] In the formula,

[0127] The system output matrix is represented by Y;

[0128] The parameter matrix to be identified is represented by X;

[0129] The data acquisition matrix is represented by R;

[0130] Finally, the parameter identification result of the motor can be output.

[0131] The purpose of the embodiment of the application is to take the residual function of the recursive least square method as a target function, set the residual threshold, take the adjustable range of the forgetting factor as the interval boundary when the system parameter state changes, calculate the optimal forgetting factor size of the current working condition through the improved golden section method, and set the forgetting factor to 1 when the system state is stable, so that the forgetting factor has good stability in the steady state. While ensuring the accuracy, the convergence speed of parameter identification is improved, and the tracking performance is enhanced.

[0132] The above is only the preferred embodiment of the application, and does not limit the implementation and protection scope of the application. For those skilled in the art, it should be realized that any equivalent replacement and obvious change made by applying the content of the specification and drawings should be included in the protection scope of the application.

Claims

1. A method for online identification of motor parameters based on improved golden section method, characterized in that: include: Step S1, when parameter identification of the motor model is not performed for the first time, obtaining a residual function of a recursive least square method with a forgetting factor; Step S2, detecting whether the residual function exceeds a preset threshold, and executing step S3 when the residual function exceeds the preset threshold; Step S3, using the residual function as the objective function and using the golden section method to adjust the forgetting factor to obtain an optimized forgetting factor; Step S4: using the recursive least squares method with the optimized forgetting factor to perform online identification of motor parameters.

2. The online motor parameter identification method based on the improved golden section method according to claim 1 is characterized in that: Before step S1, the following steps are also included: Step S01, determining whether it is the first time to perform parameter identification on the motor model, and executing step S1 if it is determined that it is not the first time to perform parameter identification on the motor model, and executing step S02 if it is determined that it is the first time to perform parameter identification on the motor model; Step S02: Use the recursive least square method with forgetting factor to perform online identification of motor parameters, and execute the step S1.

3. The online motor parameter identification method based on the improved golden section method according to claim 1 is characterized in that: The recursive least squares method with forgetting factor is: in, represents the parameter matrix to be identified; k represents the time; K represents the gain matrix of the parameter identification model; y represents the output matrix of the parameter identification model; φ T represents the transposed matrix of the data acquisition matrix φ; P represents the covariance matrix; λ represents the forgetting factor.

4. The online motor parameter identification method based on the improved golden section method according to claim 1 is characterized in that: The residual function is: Wherein, ε(k) represents the residual function, k represents the time; y represents the output matrix of the parameter identification model; φ T represents the transposed matrix of the data acquisition matrix φ; represents the parameter matrix to be identified; Represents the transposed matrix φ of the data acquisition matrix T and the parameter matrix to be identified The product of .

5. The online motor parameter identification method based on the improved golden section method according to claim 1 is characterized in that: The step S2 further includes: when the residual function does not exceed the preset threshold, setting the forgetting factor to 1, and executing step S4.

6. The online motor parameter identification method based on the improved golden section method according to claim 1 is characterized in that: The step S3 includes Step S31, using the golden section method according to the optimization interval of the forgetting factor, determining an initial first golden section point λ1 and a second golden section point λ2; Step S32, obtaining a first residual function ε(λ1) corresponding to the first golden section point and a second residual function ε(λ2) corresponding to the second golden section point based on the recursive least squares method with forgetting factor; Step S33, comparing the first residual function ε(λ1) and the second residual function ε(λ2), and updating the optimization interval of the forgetting factor according to the comparison result; Step S34 , repeating steps S31 - S33 according to the updated optimization interval of the forgetting factor until the length of the optimization interval of the forgetting factor is less than a preset length threshold or reaches a preset number of iterations, thereby obtaining the optimized forgetting factor.

7. The method for online identification of motor parameters based on the improved golden section method according to claim 6, characterized in that: The first golden section point λ1 is: l 1＝ l min +(1-φ)(λ) max -l min ) Among them, λ min represents the lower limit of the forgetting factor; λ max represents the upper limit of the forgetting factor; φ represents the golden ratio; The second golden section point λ2 is: l 2＝ l min +φ(λ max -l min )。 8. The method for online identification of motor parameters based on the improved golden section method according to claim 6, characterized in that: In the step S33, when the comparison result indicates that the first residual function ε(λ1) is smaller than the second residual function ε(λ2), the upper limit value of the forgetting factor in the optimization interval of the forgetting factor is updated to the second golden section point λ2; as well as When the comparison result indicates that the first residual function ε(λ1) is greater than the second residual function ε(λ2), the lower limit value of the forgetting factor in the optimization interval of the forgetting factor is updated to the first golden section point λ1.

9. The method for online identification of motor parameters based on the improved golden section method according to claim 6, characterized in that: The optimized forgetting factor is an average value of a lower limit value of the forgetting factor and an upper limit value of the forgetting factor in the optimization interval of the updated forgetting factor.

10. The online motor parameter identification method based on the improved golden section method according to claim 1, characterized in that: In the motor model, the d-axis current is controlled by a reference given value i d =0, a square wave signal is injected into the d-axis current reference given value.

Images