Variable-reluctance permanent magnet motor uncertain working condition adaptive efficiency optimization control method

Abstract

The application discloses a variable leakage magnetic motor uncertain working condition self-adaptive efficiency optimization control method, firstly introduces DCEE technology, respectively divides the estimated dq axis inductance change range into N points, obtains N groups of inductance, constructs gradient cost function, and then calculates flux linkage estimation value and N groups of inductance values and corresponding N groups of flux linkage estimation values, calculates N groups of dq axis current prediction values based on the flux linkage estimation value and the inductance value, and then obtains motor copper loss prediction value and iron loss branch current prediction value, thereby calculating motor iron loss prediction value and total loss prediction value, defining nominal estimation value, obtaining optimal current angle target function of minimum total loss prediction value; N-dimensional estimator is used to quantify the influence of working condition uncertainty, to avoid higher estimation error caused by improper initial setting, to improve efficiency optimization precision, to independently optimize current angle, to avoid approximation error caused by solving high-order equation set, and to obtain more accurate efficiency optimization result.

Description

Technical Field

[0001] This invention belongs to the field of motor control, specifically relating to a variable leakage magnet permanent magnet motor, and more particularly to an adaptive efficiency optimal control method for a variable leakage magnet permanent magnet motor considering uncertain operating conditions. Background Technology

[0002] As the core of energy conversion in mechatronic systems, the electric motor and its drive control significantly determine the overall energy utilization efficiency of the system. Complex applications such as electric vehicles and electric agricultural equipment often exhibit variable operating conditions, requiring motors to possess high torque, wide speed range, and high efficiency. As a special type of motor designed for variable operating conditions (hereinafter referred to as a multi-condition motor), the Variable-Leakage-Flux Permanent Magnet (VLF-PM) motor, based on the principle of "controllable leakage flux," achieves higher leakage flux under light loads and lower leakage flux under heavy loads, thus widening the motor's speed range while maintaining high operating efficiency. However, under variable operating conditions, the electromagnetic parameters of the VLF-PM motor change significantly with variations in external load. This makes it impossible to obtain an accurate model for efficiency optimization control, ultimately leading to a significant reduction in motor control efficiency and hindering effective energy management within the mechatronic system.

[0003] Currently, significant progress has been made in research on optimal motor efficiency control. For example, the patent CN202310694451.9 discloses a method for maximizing the efficiency of a permanent magnet synchronous motor, which establishes a relatively accurate core loss model under different motor operating conditions, significantly improving the accuracy and reliability of maximum motor efficiency control. The electromagnetic parameter estimation method proposed in the paper "Full Parameter Estimation for Permanent Magnet Synchronous Motors" (Transactions on Industrial Electronics, 2022) can reduce the impact of parameter variations on the model to some extent. However, existing methods fail to consider the uncertainties of the motor's operating conditions, relying solely on the motor system's own model. This undoubtedly ignores the connection between external uncertainties and the motor system, limiting the improvement of motor efficiency under multiple operating conditions.

[0004] In recent years, the paper "Concurrent Active Learning in Autonomous Airborne SourceSearch: Dual Control for Exploration and Exploitation" (Transactions on Automatic Control, 2023) proposed exploring the use of Dual Control for Exploration and Exploitation (DCEE). DCEE quantifies the uncertainty of the actuator due to environmental influences through multiple estimators or particle swarm optimization and feeds this information back to the controller, thus reducing the impact of external uncertainties on the achievement of control objectives. Furthermore, DCEE avoids errors caused by improper settings of a single estimator and the large amount of data required to train a model in intelligent algorithms. However, because it does not consider the model characteristics of the motor system, DCEE still needs to further consider the uncertainties of the motor control system and the quantification methods. Currently, its research in the field of motor control is still lacking. Summary of the Invention

[0005] Purpose of the invention: To address the problems existing in the prior art, this invention proposes an adaptive efficiency optimization control method for variable leakage magnet permanent magnet motors under uncertain operating conditions. Based on field-oriented control, it introduces DCEE technology into the field of motor control for the first time, fully considering the uncertainty of motor operating conditions and integrating the electromagnetic characteristics of VLF-PM motors to realize the motor's adaptive capability to operate in uncertain environments and improve the efficiency of motor operation under multiple operating conditions.

[0006] Technical Solution: To achieve the above-mentioned objectives, this invention proposes an adaptive efficiency control method for a variable leakage magnet permanent magnet motor considering environmental uncertainties. The technical solution includes the following steps:

[0007] Step 1): The three-phase current and voltage of the motor are transformed to obtain the current i. d i q and voltage u d u q The rotor position θ is converted into mechanical angular velocity ω, based on the iron loss resistance R. c Calculate the torque current component i of the dq axis wd i wq With iron loss branch component current i cd i cq ;

[0008] Step 2): Calculate the estimated dq-axis inductance L d L q The range of variation is divided into N points on average to obtain N groups of inductance L. dq，i / N With current iwq Construct the gradient cost function with the condition that the differential of the derivative approaches zero. Then, the flux linkage estimate is calculated. and N sets of inductance values ​​L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N and the corresponding N sets of flux linkage estimates , 1 < i ≤ N, k is the sampling time;

[0009] Step 3): Based on the flux linkage estimate and inductance value L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N Calculate N sets of predicted dq-axis current values Thus, the predicted value of motor copper loss is obtained. Predicted value of iron loss branch current This allows for the calculation of the predicted values ​​for motor iron loss and total loss. ;

[0010] Step 4): Define the nominal estimate To obtain the total loss prediction value The optimal current angle for minimizing the value Objective function:

[0011] ,

[0012] To find the expected value, Let x be the predicted value of the state variable x at time k+1, and let x = [x1, x2, x3, x4] = [i d i q i cd i cq ], U is the value of the independent variable that minimizes the objective function. The range is 90°~180°, R c R s These are the motor stator resistance and the motor iron loss resistance, respectively. Let m be the nominal predicted expected value of the current. Let $k$ be the prediction error value of the m-th current component at time $k+1$ from time $k$.

[0013] Step 5): The mechanical angular velocity ω of the motor and the given velocity ω * The deviation is used to obtain the given stator current i via the velocity loop. s * The optimal current angle β *Multiplied by current i s * Generate the optimal current allocation value for the next time step, and then compare the optimal current allocation value with the current i. d i q The deviation is input current looped and modulated, and then the inverter outputs three-phase voltage to drive the motor.

[0014] Furthermore, in step 3), according to formula

[0015] N sets of predicted dq-axis current values ​​were calculated. :i s Stator current, current β0 is the set initial current angle, Δβ is the step size, and ω e Let T be the electric angular velocity. s The sampling interval is denoted as .

[0016] Furthermore, based on the q-axis torque component current i of the motor wq Establish its differential equation Constructed gradient cost function

[0017] ω e ѱ is the electric angular velocity. f For motor magnetic flux;

[0018] flux linkage estimate d represents the gradient, and α represents the learning rate, which is no more than 10. -2 .

[0019] Furthermore, based on N sets of dq-axis current prediction values... The predicted value of motor copper loss was obtained. Iron loss branch current prediction value Then, the predicted value of the motor iron loss is calculated. and total loss prediction .

[0020] The present invention has the following beneficial effects:

[0021] 1) This invention introduces DCEE theory into the field of electromechanical systems for the first time, and adopts an autonomous efficiency-optimal control method for VLF-PM motor systems. Without human intervention, the motor can operate autonomously at the highest efficiency state under the current working conditions, truly realizing the motor's adaptive capability to operate in uncertain environments.

[0022] 2) Unlike the traditional direct calculation method for the minimum loss model, this invention proposes an autonomous current angle optimization method, which avoids the approximation error caused by solving higher-order equations and obtains more accurate efficiency optimization results.

[0023] 3) Unlike traditional parameter identification methods that only estimate changing parameters, this invention utilizes an N-dimensional estimator to quantify the impact of uncertainties in operating conditions. This avoids high estimation errors caused by improper initial settings, ultimately improving efficiency and optimization accuracy. Furthermore, this uncertainty quantification method is also applicable to other parameters affected by external operating conditions, demonstrating strong practicality.

[0024] 4) This invention targets VLF-PM motors, a type of multi-condition motor, and proposes a novel objective function with both exploration and utilization effects, based on the characteristic that its electromagnetic parameters are subject to uncertain changes in external operating conditions. This function can improve efficiency while reducing the impact of external uncertainties.

[0025] 5) Based on the principles of statistics and information theory involved in uncertainty quantification methods, this invention proposes an N-dimensional magnetic flux linkage parameter estimator. Unlike traditional intelligent efficiency optimization algorithms, it does not require the establishment of a large dataset or offline training of multiple sets of data. It has a certain degree of real-time performance and can achieve high accuracy.

[0026] 6) This invention innovatively introduces an N-dimensional estimator algorithm to quantify uncertainty. Compared with the DCEE strategy based on particle filter algorithm in the field of automatic control, it effectively reduces the computational complexity in the process and is more conducive to engineering and practical application. Attached Figure Description

[0027] Figure 1 A simulation of the dq-axis inductance of a VLF-PM motor as a function of load, taking iron loss into account.

[0028] Figure 2 A simulation of the d-axis flux linkage of a VLF-PM motor as a function of load, taking iron loss into account.

[0029] Figure 3 Equivalent circuit of the d-axis of the VLF-PM motor;

[0030] Figure 4 Equivalent circuit of the q-axis of the VLF-PM motor;

[0031] Figure 5 A schematic block diagram of the system for implementing the adaptive efficiency control method for variable leakage magnet permanent magnet motors that takes into account environmental uncertainties, as described in this invention;

[0032] Figure 6 for Figure 5 A schematic block diagram of the current angle prediction module for optimizing the current angle.

[0033] Figure 7 This is the curve showing the change of current angle with external load under the control of this invention;

[0034] Figure 8This is the load curve of the motor being increased from light load to heavy load under the control of the present invention;

[0035] Figure 9 Loss variation curves of motor under various loads under traditional loss minimization control;

[0036] Figure 10 The curves show the loss variation of the motor under various loads under the control of this invention. Detailed Implementation

[0037] To make the above-mentioned objectives, technical solutions, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0038] The electromagnetic parameters of the VLF-PM motor used in this invention change with external operating conditions, meaning that these electromagnetic parameters also have uncertainty when the external operating conditions are uncertain. For example... Figure 1 The inductance of the dq axis of the VLF-PM motor considering iron loss varies with i. q The changing trend of (load conditions); such as Figure 2 The d-axis flux linkage considering iron loss as shown by i q The trend of (load conditions) changes. It can be seen that these motor parameters change significantly with the change of operating conditions, and these changes will have a significant impact on the current control state of the motor.

[0039] like Figure 3 and Figure 4 As shown, u d u q For the equivalent dq-axis voltage, i d i q For the equivalent total dq-axis current, i wd i wq For the dq-axis torque component current, i cd i cq R is the iron loss branch current of the dq axis. c R s These are the stator resistance and the motor iron loss resistance, respectively, which are inherent parameters of the motor. d L q For dq axis inductance, ѱ f For permanent magnet flux linkage, ω e ω is the electric angular velocity.

[0040] like Figure 5 As shown, this invention employs a DCEE-based optimal control strategy for the VLF-PM motor efficiency. First, it detects the three-phase current i of the VLF-PM motor. a i b i cand three-phase voltage u a u b u c , where i a i b i c Collectively referred to as Figure 5 The i shown abc u a u b u c Collectively referred to as Figure 5 u shown abc .

[0041] The three-phase current i a i b i c and three-phase voltage u a u b u c After Clark and Park transformations, the current i in the dq-axis rotating coordinate system is obtained. d i q and voltage u d u q , where i d i q Collectively referred to as i dq u d u q Collectively referred to as u dq Simultaneously, the rotor position θ of the VLF-PM motor is detected by a rotary encoder, and its mechanical angular velocity ω and electrical angular velocity ω' are obtained by differentiating the rotor position θ. e The obtained electric angular velocity ω e Current i in a rotating coordinate system d i q and voltage u d u q The inputs are respectively fed into the N-dimensional estimator module and the current angle β prediction module. The Clark transformation matrix is:

[0042]

[0043] The Park transformation matrix is:

[0044]

[0045] i abc u abc Represents current and voltage in the natural coordinate system; i αβ u αβ Represents current and voltage in a stationary coordinate system; i dq u dqRepresents the current and voltage in the d / q coordinate system; i0 and u0 represent the zero-sequence components.

[0046] Since changes in motor flux linkage have the most significant impact on efficiency optimization control, accurate flux linkage values ​​are required. This leads to a method for quantifying the uncertainty of external operating conditions on the flux linkage value, taking into account variations in operating conditions without needing the absolute true value of the flux linkage. First, the iron loss resistance R needs to be considered. c The motor model is used to calculate the dq-axis torque current component i using the following formula. wd i wq With iron loss branch component current i cd i cq :

[0047]

[0048] The expressions for a, b, and c are:

[0049]

[0050] Among them ѱ f For motor flux linkage, L d L q Let i be the inductance along the dq axis. Then the iron loss branch current i of the motor... cd i cq It can be represented as:

[0051]

[0052] Secondly, based on the Maxwell simulation software for motor design, the dq axis inductance L is estimated. d L q The range of variation is divided into N points, resulting in N-1 segments, thus obtaining N groups of inductance L. dq，i / N L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N N groups of inductors L dq，i / N As one of the inputs to the N-dimensional estimator module, i here refers to the number of inductors, 1 < i ≤ N, as shown in Table 1 below:

[0053] Table 1

[0054]

[0055] The value of N should be between 10 and 30. If it is too small, an effective distribution cannot be obtained, and if it is too large, it will increase the computational burden of the estimator.

[0056] The obtained N sets of inductors L dq，i / N The inputs are respectively fed into the N-dimensional estimator module and the current angle β prediction module.

[0057] Based on the q-axis torque component current i of the motor wq Its differential equation is established as follows:

[0058]

[0059] The q-axis torque component current i wq The differential value of .

[0060] Therefore, in the N groups of estimator modules, with Conditional on the convergence to zero, the gradient cost function is constructed as follows. :

[0061]

[0062] To estimate the flux linkage value, the flux linkage value under no-load conditions of the motor is typically taken as the initial value for iteration. The direction of gradient descent is chosen as the iteration direction, and then the gradient cost function is used as the initial value. The observation iteration expression for the flux linkage estimate is:

[0063]

[0064] In the formula, k represents the sampling time, d represents the gradient calculation, and α represents the learning rate, which is generally no more than 10 on the order of magnitude. -2 .

[0065] Using N sets of inductance values ​​L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N and the flux linkage value to be estimated Substituting these values ​​into equations (7) and (8) respectively, we can obtain the corresponding N sets of flux linkage estimates. ,Right now These N sets of flux linkage estimates As the output of the N-dimensional estimator module.

[0066] The N-dimensional estimator module will obtain N sets of flux linkage estimates. The current state at time k is input into the current angle β prediction module. The current angle β prediction module further utilizes the N-dimensional inductance and flux linkage values ​​to derive a current angle optimization cost function with dual effects, which serves as the basis for determining the optimal current angle. Specifically:

[0067] like Figure 6 As shown, in the simulation, the initial current angle β0 is set to 90°. First, a step size Δβ is increased and decreased, with Δβ ranging from 0.5° to 1°, to obtain... .

[0068] Calculation of predicted current: based on N sets of flux linkage estimates and its corresponding N sets of inductance values ​​L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N Based on the discretized differential equation of the dq-axis current in equation (9), N sets of predicted dq-axis current values ​​considering iron loss resistance can be calculated. :

[0069]

[0070] In the formula, i s For the stator current, it can be derived from i d i q The square root of the square is used to calculate the result. Using the previously calculated N sets of flux linkage estimates as input, the search range for the current angle is set to (90°-180°), and a fixed step size of step = 0.5°~1° is set. The calculations are then performed accordingly. and We find the minimum point by moving along the direction in which the objective function decreases. s The sampling interval is denoted as .

[0071] Based on N sets of dq-axis current prediction values The predicted value of motor copper loss was obtained. The expression is:

[0072]

[0073] Iron loss branch current prediction value The expression is:

[0074]

[0075] Thus, the predicted current is obtained, which is N sets of dq-axis current prediction values. Predicted current values ​​of N groups of iron loss branches These predicted currents are used as the basis for selecting the optimal current angle.

[0076] Based on N sets of dq-axis current prediction values Predicted value of iron loss branch current The predicted value of motor iron loss was calculated. Total loss prediction The expressions are as follows:

[0077]

[0078]

[0079] The above are the prerequisites for predicting the current angle β. The next step is to obtain the optimization cost function with dual effects:

[0080] To fully utilize N-dimensional current prediction values , , , The nominal estimate is defined as follows: :

[0081]

[0082] In the formula, This indicates the calculation of mathematical expectation. For current prediction deviation, This represents the predicted value of the state variable x at time k+1 based on the information obtained at time k. The four current components are set as the state vector x = [x1, x2, x3, x4] = [i...]. d i q i cd i cq The current prediction error value of the i-th current component is then... It can be represented as:

[0083]

[0084] This is the predicted value of the i-th current component of the state quantity at time k+1 for time k.

[0085] Therefore, equation (13) can be rewritten as:

[0086]

[0087] Taking the d-axis as an example, where and This represents the expected value and error value of the d-axis current at time k+1 based on the information obtained at time k. From formula (9), we can obtain that Substituting into formula (14) get, Substituting the two values ​​above into formula (15) yields the result. and The expected value and error value of the d-axis iron loss current at time k+1 are based on the information obtained at time k. From formula (11), we can see that the calculation process for the remaining values ​​is the same as above.

[0088] because Equation (16) can be further written as:

[0089]

[0090] Replacing the current with the variable x, and based on the expected value E above, we can predict the total loss. To find the minimum value, the substitution yields equation (18), ultimately determining the optimal current angle β for the VLF-PM motor. * The optimal control objective function is abbreviated as:

[0091]

[0092] In the formula, arg represents the value of the independent variable that minimizes the objective function. As the independent variable, The value range is 90°~180°, and m is the index subscript of the current component. For stator resistance, For iron loss resistance, Let m be the nominal predicted expected value of the current. This represents the prediction error value of the m-th current component at time k+1 based on the information at time k.

[0093] Equation (18) above represents the optimal current angle β. * The objective function should be minimized, i.e. Figure 6 The basis for selecting the optimal current angle. According to DCEE theory, the first term of the equation... (The mean term) emphasizes the use of a known model, namely the total motor loss model; the latter term... (Variance term) represents the quantification of the uncertain impact of external factors. The two together constitute the overall control objective, which means that while achieving the control objective, the impact of external uncertainties is reduced.

[0094] The above indicates that the current angle iteration should satisfy the minimum cost function. The following is the specific iteration process:

[0095] The calculation is performed within the current angle β prediction module, first obtaining the optimal current angle. Then return to the next iteration. In each subsequent iteration, first adjust the current angle... Move one step Δβ in each of the increasing and decreasing directions respectively, that is... The search range for the current angle is 90°~180°, with a step size of 0.5°~1°. The current current angle will be increased or decreased accordingly. Substituting into formulas (9) and (11), we compare the magnitudes of the objective functions in the two directions, move along the direction in which the objective function decreases to find the minimum point, and extract the current angle β that minimizes the objective function. *This is the optimal current angle after iteration. The same process is repeated at the next time step, and so on. Traditional minimum loss control obtains the optimal efficiency current allocation by solving a quartic equation. This method requires an approximate process, making it difficult to obtain a precise optimal efficiency operating point. This invention avoids this problem by using an automatic optimization method.

[0096] Finally, the detected mechanical angular velocity ω of the motor is compared with the given velocity ω. * The deviation input speed loop PI module obtains the given stator current i. s * The optimal current angle β obtained will be... * Multiply by the given stator current i output by the speed loop PI module s * Generate the optimal current allocation value i for the next time step. d * i q * The optimal current allocation value i d * i q * Respectively with current i d i q The deviation input current loop PI module is modulated by SVPWM, and finally the VLF-PM motor is driven by the three-phase voltage output by the inverter.

[0097] To verify the feasibility of this invention, simulation verification was performed using Matlab / Simulink software, with a simulation step size and sampling time of 0.0001s.

[0098] To verify that the control strategy of this invention can ensure stable operation of the motor under varying loads, Figure 7 The current angle distribution under various loads is presented. Specifically, the motor operates at 200 r / min, with an initial load of 5 N·m. The load is then increased by 5 N·m at 1s, 2s, and 3s, ultimately bringing the motor to a load of 20 N·m. Figure 7 As can be seen, with the sudden change in load, the control strategy of this invention can quickly generate a stable current angle distribution signal. Figure 8 The output torque curve at the motor end is given, from Figure 8 It is evident that the control strategy of this invention can provide stable torque output across low, medium, and high load regions.

[0099] Figure 9 , Figure 10The figures show the total loss curves of the motor under low, medium, and high load zones, respectively, under traditional minimum loss control and the control method of this invention. As can be seen from the figures, compared with the traditional minimum loss control method, the method of this invention reduces losses by approximately 2.9% at a 5 N·m load; approximately 11.7% at a 10 N·m load; approximately 6.9% at a 15 N·m load; and approximately 0.5% at a 20 N·m load. Overall losses are reduced, and motor control efficiency is improved.

[0100] In summary, this invention utilizes a dual-control VLF-PM motor efficiency optimal control method. First, it detects the three-phase current and voltage signals of the motor to calculate the electrical signal values ​​in the rotating coordinate system. Second, it designs an N-dimensional parameter estimator, dividing the approximate range of the motor's electromagnetic parameters into N-1 intervals (N endpoints). The inductance value is then used as an initial value input into the N-dimensional observer to obtain the N-dimensional flux linkage distribution. Constructing the motor discrete domain loss prediction formula of equation (13) As a prerequisite for predicting the current angle β, N sets of flux linkage results are obtained using an N-dimensional estimator and substituted into the calculation to obtain the predicted values ​​of N current state variables. and First, an objective function with both exploratory and exploitative effects is constructed. Second, an optimal current angle optimization rule is designed, using the objective function as an evaluation index to find a stable current distribution angle based on a set step size. Finally, the obtained optimal current angle is... The data is fed back into the motor control loop to achieve optimal efficiency control. This invention utilizes the DCEE theory to achieve optimized motor efficiency control, primarily consisting of two parts: first, an estimated distribution of the motor's affected parameters is constructed based on external uncertainties, used to obtain the motor's predicted state variable distribution to quantify the uncertainty; second, this distribution is used to design an objective function that explores and utilizes a dual effect, enabling the motor's optimal efficiency current angle to autonomously change with external operating conditions, thus reducing the impact of external uncertainties.

Claims

1. A method for adaptive efficiency optimization control of a variable leakage flux permanent magnet motor under uncertain operating conditions, characterized in that... include: Step 1): The three-phase current and voltage of the motor are transformed to obtain the current i. d i q and voltage u d u q The rotor position θ is converted into mechanical angular velocity ω, based on the iron loss resistance R. c Calculate the torque current component i of the dq axis wd i wq With iron loss branch component current i cd i cq ; Step 2): Calculate the estimated dq-axis inductance L d L q The range of variation is divided into N points on average to obtain N groups of inductance L. dq，i / N With current i wq Construct the gradient cost function with the condition that the differential of the derivative approaches zero. Then, the flux linkage estimate is calculated. and N sets of inductance values ​​L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N and the corresponding N sets of flux linkage estimates , 1 < i ≤ N, k is the sampling time; Step 3): Based on the flux linkage estimate and inductance value L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N Calculate N sets of predicted dq-axis current values Thus, the predicted value of motor copper loss is obtained. Predicted value of iron loss branch current This allows for the calculation of the predicted values ​​for motor iron loss and total loss. ; Step 4): Define the nominal estimate To obtain the total loss prediction value The optimal current angle for minimizing the value Objective function: ， To find the expected value, Let x be the predicted value of the state variable x at time k+1, and let x = [x1, x2, x3, x4] = [i d i q i cd i cq ], U is the value of the independent variable that minimizes the objective function. The range is 90°~180°, R c R s These are the motor stator resistance and the motor iron loss resistance, respectively. Let m be the nominal predicted expected value of the current. Let $k$ be the prediction error value of the m-th current component at time $k+1$ from time $k$. Step 5): The mechanical angular velocity ω of the motor and the given velocity ω * The deviation is used to obtain the given stator current i via the velocity loop. s * The optimal current angle β * Multiplied by current i s * Generate the optimal current allocation value for the next time step, and then compare the optimal current allocation value with the current i. d i q The deviation is input current looped and modulated, and then the inverter outputs three-phase voltage to drive the motor.

2. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 1, characterized in that: In step 3), according to the formula N sets of predicted dq-axis current values ​​were calculated. :i s Stator current, current β0 is the set initial current angle, Δβ is the step size, and ω e Let T be the electric angular velocity. s The sampling interval is denoted as .

3. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 2, characterized in that: First, increase or decrease the current angle by a step size Δβ to obtain the first optimal current angle, then return to perform the next iteration. In each subsequent iteration, first adjust the current angle... Move one step Δβ in each of the increasing and decreasing directions, i.e. The search range for the current angle is 90°~180°. By comparing the magnitudes of the objective functions in the two directions, we move along the direction in which the objective function decreases to find the minimum point. The current angle that makes the objective function smaller is the optimal current angle after iteration.

4. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 1, characterized in that: In step 1), the dq-axis torque current component i wd i wq With iron loss branch component current i cd i cq They are: ， ω e ѱ is the electric angular velocity. f For motor flux linkage.

5. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 1, characterized in that: In step 2), based on the q-axis torque component current i of the motor... wq Establish its differential equation Constructed gradient cost function ω e ѱ is the electric angular velocity. f For motor magnetic flux; flux linkage estimate d represents the gradient, and α represents the learning rate, which is no more than 10. -2 .

6. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 5, characterized in that: Using N sets of inductance values ​​L d,1 / N ~L d,N / N L q,1 / N ~L q,N / N Substitute the flux linkage estimate This yields N sets of flux linkage estimates. .

7. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 2, characterized in that: Based on N sets of dq-axis current prediction values The predicted value of motor copper loss was obtained. Iron loss branch current prediction value Then, the predicted value of the motor iron loss is calculated. and total loss prediction .

8. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 1, characterized in that: The value of N ranges from 10 to 30.

9. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 2, characterized in that: The initial current angle β0 is set to 90°, and Δβ is in the range of 0.5° to 1°.

10. The adaptive efficiency optimization control method for variable leakage flux permanent magnet motor under uncertain operating conditions according to claim 3, characterized in that: During iteration, the search range for the current angle is set to 90°-180°.

Images