Hybrid wide-range high-efficiency control method for MTPA

Abstract

The invention discloses a hybrid wide-range high-efficiency control method for MTPA. The hybrid wide-range high-efficiency control method comprises the following steps: step 1, obtaining introduced associated parameters through a rotating speed PI ring; step 2, introducing the associated parameters into a simplified maximum torque current ratio formula for searching to obtain adaptive current; 3, three-phase stator current of the I PMSM is collected, feedback current is obtained after coordinate change, the self-adaptive current, the feedback current, the expected voltage and the rotating speed which are obtained through searching are introduced into a parameter identification module, online parameter identification is conducted on the motor, and an identification parameter matrix is obtained; step 4, obtaining identification parameters according to the identification parameter matrix; and step 5, introducing the identification parameters into deadbeat control to obtain dq-axis voltage, and applying the dq-axis voltage to realize motor control. According to the method, the maximum torque current ratio control efficiency of the built-in permanent magnet synchronous motor can be effectively improved, and meanwhile, the self-adaptive current obtained through searching is introduced into the parameter identification module to improve the identification precision.

Description

Technical Field

[0001] The present invention relates to the technical field of permanent magnet synchronous motors, and in particular to a hybrid wide-range high-efficiency control method for MTPA. Background Art

[0002] Interior Permanent Magnet Synchronous Motor (IPMSM) has the advantages of high efficiency, high power density, and compact structure. It is widely used in electric aviation, new energy vehicles, motorcycles and other fields. However, changes in motor temperature and operating conditions can cause nonlinear changes in motor parameters, resulting in a decrease in motor control efficiency.

[0003] Because IPMSM torque is composed of both reluctance torque and electromagnetic torque, traditional control ignores reluctance torque and relies solely on electromagnetic torque, increasing energy loss. The Maximum Torque Per Ampere (MTPA) control strategy is a highly efficient vector control strategy that effectively allocates d / q axis currents to fully utilize both electromagnetic and reluctance torques, reducing motor energy loss. The table lookup method, the most widely used MTPA current distribution method in engineering, acquires prior data for MTPA current distribution through offline training. During actual operation, the d / q axis currents are distributed using a table lookup method, eliminating the need for complex calculations. However, in fields such as electric aviation and new energy vehicles, the operating conditions of motors are relatively complex. MTPA control only through table lookup method requires not only a large amount of implementation to obtain prior data, but also query data overflow under some complex working conditions, increasing motor energy loss; the MTPA formula method has the advantages of wide application range and fast response speed, but its control efficiency depends on motor parameters and the calculation is relatively complex; the MTPA search method obtains the optimal current distribution angle to achieve MTPA control by searching the current vector angle. It is not affected by motor parameters, but its MTPA control efficiency is often inversely proportional to the search step size. The smaller the step size and the longer the search time, the higher the control efficiency. A large number of experiments are required to obtain the optimal search step size. At the same time, the search method is not suitable for complex working conditions, so it needs further improvement.

[0004] The deadbeat control algorithm calculates the desired voltage by predicting the current at the next moment and the current at the current moment. This algorithm not only avoids the difficulty of parameter tuning in the PI loop, but also effectively reduces current pulsation. However, deadbeat control is affected by the accuracy of motor parameters, so parameter identification is required.

[0005] Traditional parameter identification modules usually suffer from a decrease in parameter identification accuracy due to lack of rank. Identification methods such as high-frequency signal injection increase the rank of the motor by injecting high-frequency signals into the dq axes to improve motor identification accuracy, but this also causes additional torque pulsation and current loss, which requires further improvement. Summary of the Invention

[0006] The purpose of the present invention is to overcome the shortcomings of the existing technology and propose a hybrid wide-range high-efficiency control method for MTPA, which can effectively improve the maximum torque-to-current ratio control efficiency of the built-in permanent magnet synchronous motor, and at the same time introduce the adaptive current obtained by the search into the parameter identification module to improve the identification accuracy.

[0007] In order to achieve the above object, the technical solutions specifically adopted by the present invention are as follows:

[0008] A hybrid wide-range high-efficiency control method for MTPA comprises the following steps:

[0009] Step 1: Obtain the introduced correlation parameter η through the speed PI loop;

[0010] Step 2: Introduce the associated parameter η into the simplified maximum torque current ratio formula to search and obtain the adaptive current λ;

[0011] Step 3: Collect the three-phase stator current of IPMSM and obtain the feedback current i after coordinate change. d 、i q , introducing the adaptive current, feedback current, expected voltage and speed obtained by the search into the parameter identification module, performing online parameter identification on the motor to obtain an identification parameter matrix, and the expected voltage is obtained by deadbeat;

[0012] Step 4: Obtain identification parameters according to the identification parameter matrix, wherein the identification parameters include dq axis inductance L d , L q and motor flux ψ f ;

[0013] Step 5: Introduce the identification parameters into the deadbeat control to obtain the dq axis voltage u d 、u q , apply dq axis voltage u d 、u q Realize motor control.

[0014] Preferably, in step 1, a correlation parameter η is introduced, and its relationship with the current of the dq axis is as follows:

[0015] η=i q -i d .

[0016] Preferably, in step 2, the simplified method of the maximum torque current ratio is:

[0017] First, the traditional maximum torque-to-current ratio expression of the interior permanent magnet synchronous motor is constructed;

[0018] Introduce the associated parameter η into the maximum torque current ratio expression, and let the adaptive current λ=ψ f / [2(L q -L d )], and after simplification we get:

[0019]

[0020] i q =η+λ

[0021] Replace the dq axis currents in the torque equation with η and λ, and the torque equation after replacement is:

[0022]

[0023] Among them L d , L q is the inductance of the motor dq axis, λ is the introduced adaptive current, η is the introduced associated parameter, P n represents the number of motor pole pairs, ψ f Indicates the motor flux parameters.

[0024] Preferably, in step 2, an adaptive moment estimation method is used to search for the adaptive current:

[0025]

[0026] m t =β1m t-1 +(1-β1)g t ;

[0027]

[0028]

[0029] Among them, g k represents the objective function gradient; m t represents the partial first moment estimate; v t is the partial second moment estimate; t-1 is the adaptive current to be updated, λ t The adaptive current of the current update; α is the learning rate; β1 and β2 are the exponential decay rates of the moment estimate; ε is a small positive number; f t (λ t-1 ) is the objective function.

[0030] Preferably, in step 3, the recursive least squares method is used for parameter identification, and the expression is as follows:

[0031] y(k)=θ T (k)x;

[0032] Among them, y(k) is the system output matrix, θ T (k) is the system input matrix, and x is the identification parameter matrix.

[0033] Preferably, in step 3, when the recursive least squares method is used for parameter identification, the optimal identification parameter matrix is obtained by iteration, as follows:

[0034] x m =x m-1 +K m [y m -θ T x m-1 ];

[0035]

[0036] P m =P m-1 -K m θ m P m-1 ;

[0037] P(0)=αI;

[0038]

[0039] Among them, x m is the identification parameter matrix after the mth iteration; K m is the m-th gain matrix; P(k) is the m-th covariance matrix; P(0) is the first covariance matrix, where α is a maximum value, usually between 1e4 and 1e6, I is the unit matrix; x(0) represents the initial parameter matrix to be identified, and θ represents the zero vector.

[0040] Preferably, after the mth iteration, y m -θ T x m-1 = 0, that is, when the system output matrix of the mth time is the same as the system input matrix of the m-1th time, the optimal identification parameter matrix is obtained.

[0041] As an advantage, the identification parameters in the optimal identification parameter matrix are the dq axis inductance L d , L q , the adaptive current λ obtained by searching is compared with the identification parameter to obtain the required third motor flux parameter, which is expressed as follows:

[0042]

[0043] Preferably, in step 4, the adaptive current λ obtained by the search is introduced into the parameter module, and the rank in the parameter identification module is increased according to the expression of the adaptive current λ obtained by the search and the voltage expression of the built-in permanent magnet synchronous motor, so as to obtain the objective function of the parameter identification:

[0044] u d =Ri d +pL d i d -ω e L q i q ;

[0045] u q =Ri q +pL q i q +ω e (L d i d +ψ f );

[0046] Where R represents the stator resistance of the motor; p represents the differential operator; ω e Expressed as the electrical angular velocity of the motor.

[0047] Preferably, the desired voltage The method for obtaining is to equate the actual current at time k+1 to the expected voltage at time k, thereby achieving deadbeat current tracking. The expression is as follows:

[0048]

[0049] in, is the expected voltage of the dq axis at the kth moment, is the expected current of dq axis at the kth moment.

[0050] The present invention has the following characteristics and beneficial effects:

[0051] The present invention replaces the traditional vector angle search method by searching for adaptive current, thereby avoiding the defect of vector angle change caused by torque change and improving the application scope of the search method. Because the adaptive current search method in this article applies the current distribution formula in the MTPA formula method, but the traditional MTPA formula method is relatively complex to calculate, which greatly increases the calculation amount of the motor control algorithm. Therefore, a new associated parameter η is introduced into the MTPA current distribution formula to simplify the MTPA formula and reduce the calculation amount of the search method; and in order to solve the problem that the identification accuracy of parameter identification decreases due to lack of rank, the adaptive current λ obtained by the search is introduced into the parameter identification to obtain a new rank, thereby improving the identification accuracy; the identified parameters are introduced into the zero-beat control link to avoid the influence of motor parameter changes on it, reduce current fluctuations, and improve the control efficiency of the motor. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a schematic diagram of obtaining related parameters based on the speed PI loop.

[0053] Figure 2 This is a control flow chart of the maximum torque current ratio control method of the present invention.

[0054] Figure 3 MTPA curve on the axis.

[0055] Figure 4 It is the associated parameter value obtained after passing through the PI loop.

[0056] Figure 5 This is the flow chart of adaptive moment estimation.

[0057] Figure 6 Searches for the adaptive current value for the MTPA search method.

[0058] Figure 7 This is the identification result diagram of the axis inductance in the parameter identification module.

[0059] Figure 8 This is the identification result diagram of the axis inductance in the parameter identification module.

[0060] Figure 9 This is the identification result diagram of the stator flux in the parameter identification module.

[0061] Figure 10 The current diagrams are the adaptive moment estimation, MTPA algorithm and traditional search method, where the torque is 5Nm from 0 to 3 seconds and 8Nm from 3 to 6 seconds. DETAILED DESCRIPTION

[0062] The present invention is described in detail below in conjunction with specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0063] This embodiment discloses a hybrid wide-range high-efficiency control method for MTPA, which mainly includes an MTPA hybrid search method, a parameter identification module, and a deadbeat control module. Figure 1 As shown, the following steps are included:

[0064] Step 1: Obtain the introduced correlation parameter η through the speed PI loop;

[0065] like Figure 4 As shown, the desired correlation parameter η is obtained from the speed loop through the following expression:

[0066]

[0067] Where: T e is the electromagnetic torque of IPMSM, P n is the number of pole pairs of IPMSM, ψ f is the flux linkage of IPMSM, λ is the adaptive current introduced in the maximum torque current ratio formula, L d , L q is the inductance in the dq coordinate system.

[0068] Then the expression of the associated parameter η is obtained as follows:

[0069] η=i q -i d

[0070] Among them, i d 、i q Indicates the current of the dq axis.

[0071] Step 2: Introduce the associated parameter η into the simplified maximum torque current ratio formula to search for the optimal current distribution scheme and the adaptive current λ.

[0072] Specifically, we first establish the traditional maximum torque current ratio (MTPA) expression of the interior permanent magnet synchronous motor:

[0073]

[0074] Simplify the MTPA formula and let λ=ψ f / [2(L q -L d )], and introduce the associated parameter η into the traditional MTPA formula, the expression is as follows:

[0075]

[0076] i q =η+λ

[0077] The simplified MTPA curve is as follows: Figure 3 As shown in the figure, not only is the square root term eliminated, but the square term is also simplified. The adaptive current λ is calculated using the motor's d-axis inductance, q-axis inductance, and flux linkage. Using this simplified MTPA formula for MTPA control reduces the MCU's computational workload. However, during motor operation, motor parameters often change due to operating conditions, motor operating temperature, and the external environment. This can cause the adaptive current λ to deviate from the standard value, reducing MTPA control efficiency and increasing energy loss. The simplified maximum torque-to-current ratio formula is then used for search.

[0078] It should be noted that, in this embodiment, before searching the simplified MTPA formula, it is necessary to prove the adaptive current λ. There is a certain λ corresponding to the optimal MTPA operating point. The optimized MTPAd-axis formula is transformed into the following expression:

[0079]

[0080] Among them, i s represents the motor’s vector current, θ represents the motor’s vector current angle, and θ∈(90°,145°). According to the above expression, λ>0, and the IPMSM torque expression is as follows:

[0081] T e =1.5p n (ψ f i q +(L d -L q )i d i q )

[0082] dq axis current and vector current i s The relationship is as follows:

[0083]

[0084] Furthermore, the vector current is introduced into the torque equation, and its specific expression is as follows:

[0085]

[0086] Furthermore, the adaptive current λ is introduced into the vector current formula. To simplify the equation after introduction, its specific expression is as follows:

[0087]

[0088] Furthermore, for λ 2 Find the partial derivative of θ, whose λ 2 The expression of the partial derivative with respect to θ is as follows:

[0089]

[0090] Among them, in the built-in permanent magnet synchronous motor, L q >L d , the denominator 3Pn[(L q -L d )tanθtan 2 2θ] 2 Greater than zero, for accurate λ 2 Regarding the range of θ, let k = sec 2 θtan 2 θ+4tanθtan2θsec 2 θ, further, its numerator is changed, and the specific expression is as follows:

[0091]

[0092] Furthermore, from the above situation, we know that θ∈(90°,135°), that is, 2θ∈(180°,270°). At this time, according to the angle range, we can know that cos 4 θ>0, cos2θ<0, sinθ>0 and (sin2θ+cos2θ)<0. According to the above judgment, k>0, that is,

[0093] Furthermore, according to the above judgment, λ>0, It can be seen that Taking the reciprocal of it, we can know From the above, we can see that there is only one θ value corresponding to the MTPA operating point, namely There is only one extreme point. It can be decomposed into the following expression:

[0094]

[0095] Furthermore, according to the above There is also a unique extreme point, namely the MTPA operating point, and MTPA control can be achieved by searching the value of λ.

[0096] After completing the proof of the adaptive current λ, the simplified maximum torque-to-current ratio formula is applied for searching.

[0097] It should be noted that in order to avoid the influence of parameter changes on MTPA control efficiency, the adaptive current λ is searched to obtain accurate adaptive current λ to avoid the problem of MTPA operation efficiency being reduced due to adaptive current changes caused by parameter changes. The traditional search method usually searches for the vector angle β and distributes the dq axis current according to the vector angle and vector current. However, this type of search method is not suitable for complex working conditions because the adaptive current λ is not accurate according to the vector angle. Figure 3 The MTPA curve diagram shows that different torques correspond to different vector angles, which means that the vector angle must be re-searched whenever the torque changes. However, the search for the adaptive current is not affected by other variables such as torque, but only by parameter changes. Therefore, the search method based on the adaptive current λ has a wider range of adaptability. Introducing the searched adaptive current λ into the parameter identification module can also increase the rank of the parameter identification module and improve the identification accuracy.

[0098] Specifically, such as Figure 5 As shown, adaptive moment estimation is used to search for the corresponding λ value. The adaptive moment estimation algorithm combines the optimization of the root mean square transfer algorithm and the momentum gradient descent method. It realizes automatic adjustment of the learning rate by calculating the first-order moment estimation and the second-order moment estimation of the gradient. When the gradient is small, the learning rate is reduced to avoid the system oscillation caused by too fast parameter update. When the gradient is large, the learning rate is increased to speed up the parameter update speed and improve the convergence speed of the system. At the same time, the introduction of momentum and second-order moment estimation will increase the stability of the system. Therefore, the adaptive moment estimation has the advantages of fast convergence speed and the learning rate adjustment is not easy to fall into the local optimum, which avoids the problem of difficult step size adjustment of the traditional search method. Its search process is as follows Figure 6 As shown in Figure 2, the specific algorithm of the adaptive moment estimation algorithm is as follows:

[0099]

[0100] m t =β1m t-1 +(1-β1)g t ;

[0101]

[0102] Among them, g k represents the objective function gradient; m t represents the partial first moment estimate; v t is the partial second moment estimate; t-1 is the adaptive current to be updated, λ t The adaptive current of the current update; α is the learning rate; β1 and β2 are the exponential decay rates of the moment estimate; ε is a small positive number; f t (λ t-1 ) is the objective function.

[0103] In this embodiment, before the adaptive moment estimation search begins, the initial learning rate α is set to 0.01, the partial first-order exponential decay rate β1 is set to 0.9, the partial second-order exponential decay rate β2 is set to 0.99, and ε is set to 1e-8.

[0104] It should be noted that, in the actual search process, there is often a problem that the adaptive current is not accurate enough due to insufficient search progress. Therefore, further improvements are made to the search interval to complete the search.

[0105] In order to improve the accuracy of the search value, in this embodiment, the learning rate is improved. The specific improved expression is as follows:

[0106]

[0107] Among them, C is a fixed constant, G k-1 is the adaptive current λ, θ found last time t is the current search value, and α is the learning rate. This allows the human intervention step size to be shortened when the search function is within a certain range from the previous target function, thereby increasing the search accuracy.

[0108] Furthermore, to shorten the setting time of α and improve the search speed, the initial value of α is set to 0.1. During the search process, three search values with the same interval are recorded, which are recorded as simple_1, simple_2, and simple_3. To prevent the search results from oscillating frequently around the exact value due to the step size being set too short, the three recorded search results are calculated as follows:

[0109] k=max(simple_1-simple_3,simple_3-simple_2,simple_2-simple_1);

[0110] Furthermore, if the k value is less than a fixed constant, the search is stopped, and the average of the three recorded results is taken as the search result, which is input into the simulation module for simulation calculation. Note that the time interval between the three records cannot be too short. The purpose of this step is to reduce the search time and prevent the gradient from being less than the set threshold due to the search step being too small.

[0111] Step 3: Collect the three-phase stator current of IPMSM and obtain the feedback current i after coordinate change. d 、i q The adaptive current, feedback current, expected voltage and speed obtained by the search are introduced into the parameter identification module, and the motor parameters are identified online to obtain the identification parameter matrix. The expected voltage is obtained by zero beat.

[0112] Specifically, the parameter identification module adopts the recursive least squares method, and its specific expression is as follows:

[0113] y(k)=θ T (k)x;

[0114] Among them, y(k) is the system input matrix, θ T (k) is the input matrix of the system, and x is the system identification parameter matrix.

[0115] Furthermore, when the recursive least squares method is used for parameter identification, the optimal identification parameter matrix is obtained through iteration as follows:

[0116] x m =x m-1 +K m [y m -θ T x m-1 ]

[0117]

[0118] P m =P m-1 -K m θ m P m-1 ;

[0119] Among them, x m is the parameter identification vector for the mth time; K m is the m-th gain matrix; P(k) is the m-th covariance matrix; the parameter identification module needs to be initialized before starting, that is, let P(0) = αI, where α takes a larger value, generally between 1e4 and 1e6, I is the unit matrix, let in Take a smaller value or a zero vector. When the identification module is stable, it is used as an accurate value. After the mth iteration, y m -θ T x m-1 = 0, that is, when the system output matrix for the mth time is the same as the system input matrix for the m-1th time, the optimal identification parameter matrix is obtained. When the parameters are stable, the adaptive current λ remains unchanged. When the identification parameters change and the difference between them and the accurate value exceeds a certain value, the MTPA starts to search again for the accurate adaptive current λ.

[0120] Further, specific, such as Figure 2As shown in the figure, the identified parameters are introduced into the deadbeat control to reduce the impact of parameter changes on the deadbeat control. Among them, the error beat control mainly uses the discretization of the voltage equation to equate the actual current at time k+1 to the expected voltage at time k, thereby achieving deadbeat current tracking. Compared with traditional PI control, deadbeat control not only avoids the problem of difficult PI parameter tuning, but also has the advantages of small current fluctuation and fast dynamic response. Its specific expression is as follows:

[0121]

[0122] in, is the expected voltage of the dq axis at the kth moment, is the expected current of the dq axis at the kth moment. Through online parameter identification, the influence of parameter changes on the deadbeat control can be reduced.

[0123] Step 4: Obtain identification parameters according to the identification parameter matrix, wherein the identification parameters include dq axis inductance L d , L q and motor flux ψ f The identification parameters in the optimal identification parameter matrix are the dq axis inductance L d , L q , the adaptive current λ obtained by searching is compared with the identification parameter to obtain the required third motor flux parameter, which is expressed as follows:

[0124]

[0125] Thus, the three optimal parameters are obtained.

[0126] Step 5: Introduce the identification parameters into the objective function of the deadbeat control to obtain the dq axis voltage u d 、u q , apply dq axis voltage u d 、u q Realize motor control.

[0127] Specifically, such as Figure 7-10 As shown in the figure, the adaptive current λ obtained by the search is introduced into the parameter module. The rank in the parameter identification module is increased according to the expression of the adaptive current λ and the voltage expression of the internal permanent magnet synchronous motor. This method is used to increase the identification accuracy of the parameter identification module. The specific expression is as follows:

[0128] u d =Ri d +pL d i d -ω e ψ q ;

[0129] uq =Ri q +pL q i q +ω e (L d i d +ψ f ).

[0130]

[0131] In order to verify the feasibility and effectiveness of the present invention, an IPMSM simulation model was built for verification. The main parameters of the simulated motor are shown in Table 1:

[0132] Table 1 Simulation motor parameters

[0133]

[0134] When the motor speed is 700 rpm and the load torque is 5 N·m, the associated parameter η obtained by PI link feedback is as follows: Figure 4 As shown in the figure, given two different loads of 8N·m and 5N·m, at a speed of 700rpm, the adaptive current λ is observed and searched. The specific results are as follows: Figure 6 As shown; the parameter identification module uses the recursive least squares method to simultaneously identify the d-axis inductance, q-axis inductance and permanent magnet flux linkage ψ f , let the observer without the adaptive current λ be used as identification module 1, let the observer with the adaptive current λ be used as identification module 2, and the observation results are as follows Figure 7 As shown; when the motor speed is 700rpm and the speed is 10N·m, the current and speed fluctuations are recorded. The specific results are as follows Figure 8 shown.

[0135] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A hybrid wide-range high-efficiency control method for MTPA, characterized in that: The steps include: Step 1: Obtain the introduced correlation parameter η through the speed PI loop; Step 2: Introduce the associated parameter η into the simplified maximum torque current ratio formula to search and obtain the adaptive current λ; Step 3: Collect the three-phase stator current of IPMSM and obtain the feedback current i after coordinate change. d 、i q , introducing the adaptive current, feedback current, expected voltage and speed obtained by the search into the parameter identification module, performing online parameter identification on the motor to obtain an identification parameter matrix, and the expected voltage is obtained by deadbeat; Step 4: Obtain identification parameters according to the identification parameter matrix, wherein the identification parameters include dq axis inductance L d , L q and motor flux ψ f ; Step 5: Introduce the identification parameters into the objective function of the deadbeat control to obtain the dq axis voltage u d 、u q , apply dq axis voltage u d 、u q Realize motor control.

2. The method according to claim 1, characterized in that In step 1, an associated parameter η is introduced, and its relationship with the current of the dq axis is as follows: η=i q -i d 。 3. The method according to claim 2, characterized in that In step 2, the simplified method of the maximum torque current ratio is: First, the traditional maximum torque-to-current ratio expression of the interior permanent magnet synchronous motor is constructed; Introduce the associated parameter η into the maximum torque current ratio expression, and let the adaptive current λ=ψ f / [2(L q -L d )], and after simplification we get: I q =η+λ Replace the dq axis currents in the torque equation with η and λ, and the torque equation after replacement is: Among them L d , L q is the inductance of the motor dq axis, λ is the introduced adaptive current, η is the introduced associated parameter, P n represents the number of motor pole pairs, ψ f Indicates the motor flux parameters.

4. The method according to claim 3, characterized in that In step 2, the adaptive current is searched using the adaptive moment estimation method: g k =▽ θ f t (l t-1 ); m t =β1m t-1 +(1-β1)g t ; Among them, g k represents the objective function gradient; m t represents the partial first moment estimate; v t is the partial second moment estimate; t-1 is the adaptive current to be updated, λ t The adaptive current of the current update; α is the learning rate; β1 and β2 are the exponential decay rates of the moment estimate; ε is a small positive number; f t (λ t-1 ) is the objective function.

5. The method according to claim 4, characterized in that In step 3, the recursive least squares method is used to identify the parameters, and the expression is as follows: y(k)=θ T (k)x: Among them, y(k) is the system output matrix, θ T (k) is the system input matrix, and x is the identification parameter matrix.

6. The method according to claim 5, characterized in that In step 3, when the recursive least squares method is used for parameter identification, the optimal identification parameter matrix is obtained by iteration, as follows: x m =x m-1 +K m [y m -θ T x m-1 ]: P.S m JP m-1 -K m θ m P.S m-1 100. P(0)=αI; Among them, x m is the identification parameter matrix after the mth iteration; K m is the m-th gain matrix; P(k) is the m-th covariance matrix; P(0) is the first covariance matrix, where α is a maximum value, usually between 1e4 and 1e6, and I is the unit matrix; x(0) represents the initial parameter matrix to be identified, Represented as a zero vector.

7. The method according to claim 6, characterized in that After the mth iteration, y m -θ T x m-1 = 0, that is, when the system output matrix of the mth time is the same as the system input matrix of the m-1th time, the optimal identification parameter matrix is obtained.

8. The method according to claim 7, characterized in that The identification parameters in the optimal identification parameter matrix are dq axis inductance L d , L q , the adaptive current λ obtained by searching is compared with the identification parameter to obtain the required third motor flux parameter, which is expressed as follows:

9. The method according to claim 7, characterized in that In step 4, the adaptive current λ obtained by the search is introduced into the parameter module, and the rank in the parameter identification module is increased according to the expression of the adaptive current λ obtained by the search and the voltage expression of the interior permanent magnet synchronous motor, thereby obtaining the objective function of the parameter identification: u d =Ri d +pL d i d -ω e L q i q ; u q =Ri q +pL q i q +ω e (L d i d +ψ f ); Where R represents the stator resistance of the motor; p represents the differential operator; ω e Expressed as the electrical angular velocity of the motor.

Images