A construction method of a flat wire permanent magnet wheel hub motor inverse push type model predictive controller

Abstract

The application discloses a construction method of a flat wire permanent magnet wheel hub motor backstepping model predictive controller, first derives an angular velocity deviation, constructs a corresponding Lyapunov function one, calculates q-axis current, obtains a q-axis current estimation value according to a calculation formula of the q-axis current and a torque estimation error, and constructs a backstepping controller with the angular velocity deviation input and the q-axis current estimation value output; then constructs a corresponding Lyapunov function two according to the angular velocity deviation and a current tracking error, solves acceleration, constructs an acceleration control module with the angular velocity deviation and the current tracking error input and the acceleration output, and finally evaluates each d-axis current prediction value, q-axis current prediction value and acceleration prediction value through a value function formula, selects a voltage vector corresponding to a prediction value that makes the value function minimum; the method realizes rapid and accurate acquisition of motor given current under complex working conditions, improves the response capability of a hub driving system, and is favorable to improvement of collaborative control performance of a distributed driving system.

Description

Technical Field

[0001] This invention relates to a controller for a flat wire permanent magnet hub motor for electric vehicles, belonging to the field of motor control technology, which controls the flat wire permanent magnet hub motor to achieve stable operation of electric vehicles. Background Technology

[0002] Distributed in-wheel drive electric vehicles (DEVs) offer short drive chains and flexible handling, making them a key research area in electric vehicle development. However, space constraints necessitate complex cooling systems for in-wheel motors, increasing system costs and vehicle unsprung mass, which further reduces vehicle control flexibility. Flat wire windings, employing quadrilateral cross-sections, offer a 30% higher slot fill factor and 150% better heat dissipation compared to traditional round windings. Power density increases from 3.5 kW / kg to 4.5 kW / kg, while volume decreases by 30%. Furthermore, flat wire motors achieve over 90% efficiency, significantly higher than traditional round wire motors. Moreover, their low-speed, high-torque conversion efficiency is far superior to round wire motors, making them ideal for low-speed direct-drive in-wheel motors. Applying flat wire winding technology to in-wheel motors reduces weight by approximately 12%, lowers temperature rise under rated conditions by 10%, and effectively simplifies the cooling system. In summary, flat-wire hub motors can significantly reduce vehicle unsprung mass and drive system costs, which is conducive to the popularization and application of distributed hub drive electric vehicles.

[0003] Current in-wheel motor speed stabilization commonly employs PI controllers. However, the integrator in a PI controller is prone to saturation, resulting in slow speed response of the in-wheel motor under uneven road conditions. Furthermore, under conditions of strong external interference, in-wheel motor speed overshoot becomes a significant issue. These problems lead to noticeable differences in the speeds of multiple in-wheel motors, causing distributed electric vehicles to deviate from their intended trajectory and experience instability during cornering.

[0004] Improving the rapid response and resistance to external disturbances of hub motor drive systems is an effective measure to achieve stable operation of distributed electric vehicles. Model Predictive Control (MPC) has a fast response speed and is used in fields such as motor drives. Traditional MPC obtains the q-axis current through a PI controller in the speed loop and uses the current as a value function constraint to achieve speed control. However, the integrator in the PI controller is prone to saturation, resulting in slow dynamic condition response, which makes it difficult to meet the high-performance driving requirements of distributed electric vehicles under varying road conditions. To address these issues, there are currently two types of model prediction improvement strategies: one method constructs a value function with torque as the variable, achieving stable speed control by improving torque response. However, hub motors have large inertia, and speed changes lag significantly behind changes in varying road conditions (torque changes). The motor speed stability is poor on uneven roads. Furthermore, since torque control obtains the torque setpoint through the PI output of the speed loop, its integral lag characteristic makes the hub response untimely, which also affects the coordinated control of the entire vehicle under varying road conditions. Another approach abandons the traditional dual-loop cascaded structure of speed and torque, and constructs a direct speed controller that is not cascaded with speed and current, intending to solve the problems of large motor speed overshoot and poor dynamic capability. However, due to the large inertia of the hub motor, there is a time lag between speed and current, making it difficult to effectively and directly construct the relationship between speed and current in the value function. As a result, the actual effect is difficult to meet the operation requirements of distributed electric vehicles.

[0005] In conclusion, under complex and uneven road conditions, simply improving the speed or torque response is insufficient to meet the stability requirements of distributed electric vehicles. Given the characteristic of flat-wire permanent magnet hub motors where the large inertia causes speed to lag behind torque, it is necessary to propose a controller that comprehensively considers both torque and speed to achieve stable operation of electric vehicles under complex conditions. Summary of the Invention

[0006] To address the aforementioned problems and effectively overcome the drawbacks of PI controllers being prone to saturation and having slow response speed under dynamic operating conditions, this invention proposes a reverse-propulsion model predictive controller for flat wire permanent magnet hub motors and its construction method. This improves the response speed of the speed loop of the flat wire permanent magnet hub motor drive system, enabling rapid response of the flat wire permanent magnet hub motor's speed and torque under varying road conditions.

[0007] To achieve the above objectives, the present invention provides a method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor, employing the following technical solution:

[0008] (1) Construct a system with angular velocity deviation e as input and q-axis current estimate as input. For the output reverse controller:

[0009] Differentiate the angular velocity deviation e and construct the corresponding Lyapunov function V1 = e 2 / 2 and differentiate;

[0010] Based on the Lyapunov stability condition dV1 / dt<0, the q-axis current is calculated as follows:

[0011] Calculate the load torque estimation error T is the estimated load torque value. L This is the nominal value of the load torque;

[0012] Based on the formula for calculating the q-axis current and the torque estimation error ΔT L Obtain the q-axis current estimate

[0013] (2) Constructing a system based on angular velocity deviation e and current tracking error For input, i d and i q An acceleration control module that outputs acceleration 'a' as the current component:

[0014] Based on angular velocity deviation e and current tracking error e q e d Construct the corresponding Lyapunov function V2 = (e 2 +e q 2 +e d 2 +ΔT L 2 Given dV2 / dt < 0 and r1 > 0, solve for the acceleration a based on the Lyapunov stability condition dV2 / dt < 0.

[0015] (3) Based on the mathematical model of the motor, obtain the predicted values ​​of the d-axis and q-axis currents i at the next sampling time k+1. d (k+1),i q (k+1) and the acceleration prediction value a(k+1), and the current prediction values ​​i of each d and q axis. d (k+1),i q (k+1) and the acceleration prediction value a(k+1) are expressed by the value function. The evaluation is performed, where A is the weighting coefficient. The voltage vector corresponding to the predicted value that minimizes Err(i) is selected and controlled by the inverter to control the motor.

[0016] Furthermore, the acceleration mentioned Coefficients k1>0, k2>0, L d L q These are the d-axis and q-axis stator inductances, respectively; u d uq The stator voltages are d-axis and q-axis, respectively; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the angular velocity; ψ f It is a permanent magnet flux linkage.

[0017] Furthermore, the derivative of the Lyapunov function is: The q-axis current i q =2(Bω+T) L +k3Je) / (3Pψ f ).

[0018] Furthermore, the estimated q-axis current value J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the angular velocity; ψ f is the permanent magnet flux linkage, and k3 is the Lyapunov coefficient.

[0019] The advantages of this invention using the above technical solution are:

[0020] 1. This invention replaces the traditional PI speed loop with reverse-propagation control, and obtains the q-axis current setpoint using the speed error as the state variable. It effectively overcomes the shortcomings of PI controllers, such as easy saturation, slow response, and poor dynamic quality under dynamic conditions. It has the advantages of fast speed response and less parameter adjustment, and realizes the rapid and accurate acquisition of the setpoint current of the flat wire permanent magnet hub motor under complex conditions. It improves the response capability of the hub drive system and is conducive to improving the collaborative control performance of the distributed drive system.

[0021] 2. The acceleration control model in this invention can effectively reflect the dynamic characteristics of the hub motor and is the link between the vehicle kinematic model and the hub motor control. Its rationality directly affects the fast response performance of the hub motor drive system. Therefore, an adaptive acceleration control model for the flat wire permanent magnet hub motor is constructed based on the load torque to improve the observation accuracy of the load torque.

[0022] 3. Based on the working characteristics of distributed drive hub motors, this invention designs and optimizes a value function based on adaptive acceleration. The value function is constructed by combining the acceleration control model and the current relationship. The optimal voltage vector is selected to act on the inverter through model predictive control, and the driver current limit is used as a constraint to achieve rapid response of motor speed and torque. Attached Figure Description

[0023] Figure 1 This is a structural block diagram of a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to the present invention;

[0024] Figure 2 for Figure 1 Block diagram illustrating the structural principle of the mid-thrust controller;

[0025] Figure 3 for Figure 1 Block diagram illustrating the structural principle of the acceleration control module;

[0026] Figure 4 A block diagram illustrating the construction principle of the prediction module and value function optimization;

[0027] Figure 5 This is a schematic diagram of the simulated current and speed change waveforms of a traditional vector controller;

[0028] Figure 6 This is a schematic diagram of the simulated current and speed change waveforms of the reverse-engineering model predictive controller of the present invention.

[0029] Figure 7 This is a schematic diagram of the simulated current and variable carrier waveform of a traditional vector controller;

[0030] Figure 8 This is a schematic diagram of the simulation current and variable carrier waveform of the reverse-engineering model predictive controller of the present invention. Detailed Implementation

[0031] like Figure 1 The flat wire permanent magnet hub motor reverse-propagation model predictive controller shown herein uses model predictive control as its core. It combines a reverse-propagation controller and an acceleration control model to construct a value function, selecting the optimal voltage vector applied to the inverter to improve the speed and torque response capability of the flat wire permanent magnet hub motor under complex road conditions. The invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0032] This invention first constructs a reverse-engineering controller with angular velocity deviation as input and q-axis current estimate as output: First, the angular velocity deviation is differentiated, and the corresponding Lyapunov function is constructed and differentiated. Based on the Lyapunov stability condition, the q-axis current is calculated. The load torque estimation error is calculated, and the estimated q-axis current value is obtained based on the q-axis current calculation formula and the torque estimation error. This constructs a reverse-engineering controller with angular velocity deviation as input and q-axis current estimate as output. Next, based on the angular velocity deviation and current tracking error, a second Lyapunov function is constructed. Based on the Lyapunov stability condition, the acceleration is solved. This constructs an acceleration control module with angular velocity deviation and current tracking error as input and acceleration as output. Finally, the predicted d-axis and q-axis current values ​​and acceleration prediction values ​​are evaluated using a value function, and the voltage vector corresponding to the predicted value that minimizes the value function is selected to control the motor via the inverter. The specific method is as follows:

[0033] Step 1: Constructing the mathematical model of the flat wire permanent magnet hub motor

[0034] The mathematical model for the flat wire permanent magnet hub motor based on the synchronous rotating rotor coordinate dq is as follows:

[0035]

[0036]

[0037]

[0038] In the formula: u d u q These are the stator voltages along the d and q axes, respectively; i d i q These are the d-axis and q-axis stator currents, respectively; R is the stator resistance; L d L q These are the d-axis and q-axis stator inductances, respectively; T L ψ is the nominal load torque; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f It is a permanent magnet flux linkage.

[0039] Step 2: Constructing the thrust reverser controller

[0040] like Figure 2 As shown, the reverse controller selects a portion of the states of the flat wire permanent magnet hub motor drive system to form a subsystem and the corresponding Lyapunov function, and designs a virtual control function to solve for i. q To meet the load torque T L Real-time changes in demand, setting estimated load torque values, and improving q-axis current enhance the speed and accuracy of the control system. Specifically:

[0041] 1. Taking the angular velocity deviation e as input, and differentiating it, we can obtain equation (3) in the mathematical model as follows:

[0042]

[0043] Where e = ω* - ω, ω* is the given value of the rotor's mechanical angular velocity, ω is the actual mechanical angular velocity of the rotor, and L d L q These are the d-axis and q-axis stator inductances, respectively; T L ψ is the nominal value of the load torque; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; f It is a permanent magnet flux linkage.

[0044] 2. Construct the corresponding Lyapunov function: Lyapunov function V1 = e 2 / 2, taking its derivative, based on the angular velocity deviation e, we can obtain from formula (4):

[0045]

[0046] 3. If the Lyapunov function is stable, the flat wire permanent magnet hub motor can achieve global asymptotic tracking of speed under dynamic operating conditions, thus achieving the goal of high-precision speed control. In the above formula, the Lyapunov stability condition is dV1 / dt < 0, and the selected current i d Under motor control conditions where q = 0, the q-axis current can be obtained by combining equation (5):

[0047] i q =2(Bω+T) L +k3Je) / (3Pψ f (6)

[0048] In the formula: k3 is the Lyapunov coefficient.

[0049] Substituting the q-axis current from equation (6) into the Lyapunov function from equation (5), we get: dV1 / dt=-k3e 2 When k3>0, the system satisfies the Lyapunov stability condition.

[0050] 4. To meet the load torque T L Real-time changing requirements, calculate load torque estimation error T is the estimated load torque value. L Let the nominal value of the load torque be the derivative with respect to the load torque estimation error, then:

[0051]

[0052] Therefore, based on the q-axis current calculation formula (6), the q-axis current control function is selected to obtain the estimated value of the q-axis current. as follows:

[0053]

[0054] In the formula: J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f It is a permanent magnet flux linkage.

[0055] Therefore, a method was constructed that uses the angular velocity deviation e as input and the q-axis current estimate as input. The output is a reverse controller.

[0056] Step 3: Construct the acceleration control module

[0057] To meet the rapid response control requirements of flat wire permanent magnet hub motors under dynamic operating conditions, an adaptive acceleration-velocity model based on load torque is proposed. The construction process is as follows: Figure 3 As shown, it specifically includes:

[0058] 1. Estimated q-axis current based on the output of the reverse controller Construct the corresponding state equations.

[0059] in i d Under the control condition of 0, the estimated value of the q-axis current in equation (8) above is... Substitute into equation (4) to obtain the initial acceleration model:

[0060] de / dt=-dω / dt=-ke-ΔT L / J (9)

[0061] In the formula, J is the moment of inertia; ω is the mechanical angular velocity of the rotor.

[0062] 2. Detect the three-phase current i of the flat wire permanent magnet hub motor. a i b i c The three-phase current i a i b i c After sequential Clark and Park transformations, the current components i in the two-phase rotating coordinate system are obtained. d and i q To achieve rapid torque tracking, the estimated q-axis current output from the reverse-engineered controller will be used. With current i q In comparison, the difference is the current tracking error e. q Choose the current tracking error as the state variable: This is the estimated value of the d-axis current. The given d-axis current is 0, therefore...

[0063] For current tracking error e q e d Differentiate each equation separately and construct the state equation by combining equations (1) and (2). and

[0064]

[0065]

[0066] In the formula: L d L q These represent the stator inductances of the d and q axes, respectively; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f It is a permanent magnet flux linkage.

[0067] 3. Based on the angular velocity deviation e and the current tracking error e q e dConstruct the corresponding Lyapunov function V2 to achieve velocity tracking and current tracking:

[0068] V2=(e 2 +e q 2 +e d 2 +ΔT L 2 / r1) / 2 (12)

[0069] Where r1>0, according to the stability condition of the Lyapunov function, differentiate equation (12), combine it with equations (7) and (9), and substitute it into the state equations of (10) and (11). and We can obtain:

[0070]

[0071] In the formula: L d L q These represent the stator inductances of the d and q axes, respectively; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f For permanent magnet flux linkage; u d u q These are the d-axis and q-axis voltages, respectively.

[0072] Equation (13) includes the acceleration *a* of the flat wire permanent magnet hub motor. To achieve stability of the Lyapunov function equations, equation (13) must satisfy dV² / dt < 0. Based on this constraint, an acceleration control function is selected, and the acceleration *a* is solved:

[0073]

[0074] Where: coefficients k1>0, k2>0, and the adaptive law of load torque is: In the formula L d L q These represent the stator inductances of the d and q axes, respectively; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f The term refers to the permanent magnet flux linkage. This equation indicates that the acceleration model constructed in this invention can quickly map changes in the load of the flat wire permanent magnet hub motor, which is beneficial for subsequent control to improve the hub torque response speed.

[0075] This constructs a system based on angular velocity deviation e and current tracking error. For input, An acceleration control module that outputs acceleration 'a'.

[0076] Further verification showed that the acceleration model could achieve global asymptotic tracking of the subsystem, thus enabling stable operation of the electric vehicle. Details are as follows:

[0077] The acceleration a and the adaptive law of load torque in equation (14) Substituting into equation (13), we get:

[0078] dV2 / dt=-ke 2 -k1e q 2 (15)

[0079] Given the conditions (15) and k, k1, we can obtain: dV2 / dt<0, which indicates that the Lyapunov function equation based on equation (12) is stable and can achieve global asymptotic tracking of rotational speed and dq axis current.

[0080] Further proof of the global uniform stability of the Lyapunov functional equations:

[0081] From equation (15) and the given conditions of k and k1, we know that dV2 / dt ≤ -ke 2 Therefore, we can conclude that:

[0082]

[0083] At the same time, since V2 is bounded, according to Barbalat's corollary, we can obtain... It has been proven that the tracking error of the permanent magnet flat wire hub motor drive system asymptotically approaches zero, and the control system satisfies global uniform stability.

[0084] From the above construction process of the adaptive acceleration control module based on load torque, it can be seen that: when the system satisfies the adaptive law of acceleration a and load torque... The flat wire permanent magnet hub motor drive system can simultaneously achieve global asymptotic tracking and global consistent stability of speed and current.

[0085] Step 4: Construct the prediction module for the flat wire permanent magnet hub motor. Specific steps are as follows: Figure 4 As shown, the process is as follows:

[0086] 1. Based on the mathematical model of the motor, and using the Euler approximation of formulas (1) and (2), the discrete d-axis and q-axis current prediction formulas are obtained:

[0087] i d (k+1)=i d (k)+T s [u d (k)-Ri d (k)+ω re (k)L q i q (k)] / L d(17)

[0088] i q (k+1)=i q (k)+T s [u q (k)-Ri q (k)-ω re (k)L d i d (k)-ω re (k)ψ f ] / L q (18)

[0089] In the formula: i d (k+1),i q (k+1) represent the predicted d-axis and q-axis current values ​​at the next sampling time k+1, respectively; i d (k), i q (k) represents the d-axis and q-axis current feedback values ​​at the current time k; E d (k) and E q (k) represents the back electromotive force along the d and q axes at the current time k; u d (k), u q (k) represents the d-axis and q-axis voltages at the current time k; ω re (k) represents the rotor's electric angular velocity; T s L is the controller data sampling period. d L q These are the stator inductances along the d and q axes, respectively; ψ f It is a permanent magnet flux linkage.

[0090] A two-level inverter has a total of 6 non-zero voltage vectors and 2 zero voltage vectors (7 effective voltage vectors in total), each corresponding to a different u. d (k) and u q (k), according to equations (17) and (18), the current value i at the next sampling moment of the dq axis under different voltage vectors can be obtained. d (k+1),i q (k+1).

[0091] ② To achieve high-performance operation of the flat-wire permanent magnet hub motor under complex road conditions, this invention incorporates acceleration variables into the value function for model predictive control. The specific method is as follows:

[0092] First, determine the predicted value of acceleration a(k+1) at the next moment. Substituting this value into equation (4) according to the current prediction model, we can obtain:

[0093] a(k+1)=dω / dt=3Pψ f i q (k+1) / (2J)-(Bωre +T L ) / J-3P(L q -L d )i q (k+1)i d (k+1) / (2J) (19)

[0094] In the formula: a(k+1) is the predicted acceleration value at the next moment; ω re (k) represents the rotor's electric angular velocity; L d L q These are the d-axis and q-axis stator inductances, respectively; T L ψ is the load torque; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f It is a permanent magnet flux linkage.

[0095] Secondly, the MPC value function is constructed. Due to the controller current limitation, considering the weighted relationship between the dq-axis current and acceleration, an acceleration variable weighting coefficient A is introduced. Combining equations (17), (18), and (19), the value function is selected as follows:

[0096]

[0097] In the formula, i d (k+1),i q (k+1) represents the predicted d-axis and q-axis current values ​​at the next sampling time; a(k+1) represents the predicted acceleration value at the next time; and a represents the acceleration setpoint. This is the estimated value of the q-axis current; This is the estimated value of the d-axis current.

[0098] When the weighting coefficient A > 0, it represents a reverse-push MPC for a flat wire permanent magnet hub motor based on the acceleration value function. Using the controller's rated current as a constraint, the response capability of the flat wire permanent magnet hub motor and i are considered. q Current pulsation; select appropriate weighting coefficient A.

[0099] Finally, the value function is optimized. A two-level inverter has a total of 6 non-zero voltage vectors and 2 zero voltage vectors (7 effective voltage vectors), each corresponding to a different u. d and u q Based on equations (17), (18), and (19), seven predictions are performed to obtain the predicted current value i at the next sampling time of the dq axis under different voltage vectors. d (k+1),i q (k+1) and the predicted acceleration value a(k+1) at the next moment; the estimated dq-axis current value The acceleration 'a' obtained in step three is used as the given value input for model predictive control, and the predicted values ​​under different voltage vectors are substituted into the value function (20) for evaluation; the voltage vector control inverter corresponding to the predicted value that minimizes Err(i) is selected. Figure 4 S a,b,c U represents the inverter switching state corresponding to this voltage vector. a,b,c This refers to the three-phase phase voltages corresponding to this voltage vector.

[0100] like Figure 1 As shown, the reverse-propulsion model predictive controller for the flat-wire permanent magnet hub motor constructed in this invention is applied to the control of the flat-wire permanent magnet hub motor. It consists of an outer-loop reverse-propulsion controller and an inner-loop predictive module and acceleration control module. In the outer-loop control section, the reverse-propulsion controller replaces the traditional speed loop PI controller of the MPC model. The speed tracking error is used as the state variable to obtain the q-axis current setpoint, effectively overcoming the shortcomings of PI controllers such as easy saturation and slow response speed under dynamic conditions, thus improving the response speed of the speed loop of the flat-wire permanent magnet hub motor drive system. In the inner-loop control section, an adaptive acceleration control model for the flat-wire permanent magnet hub motor is constructed based on the load torque. An acceleration weight coefficient is introduced to construct a value function, and the inverter voltage vector is optimized to achieve rapid response of the flat-wire permanent magnet hub motor's speed and torque under varying road conditions, ultimately achieving stable driving of the electric vehicle. First, the actual motor speed ω is measured using a rotary transformer, and the actual speed ω is compared with the given speed ω... * The angular velocity deviation e is obtained by comparing and subtracting, and is used as the input to the reverse controller; secondly, the three-phase current i of the flat wire permanent magnet hub motor is detected. a i b i c The current i in the stationary coordinate system is obtained after Clark transformation. α and i β , change i α i β The mechanical angle θ of the motor measured by the rotary transformer is used to obtain the current component i in the two-phase rotating coordinate system through Park transformation. d and i q ; Again, i d i q And ω is used as the input to the prediction module, i d i q Compared with the estimated values ​​of the dq-axis currents respectively The difference is used to obtain the current tracking error e. d and e q Current tracking error e d and e qAs the input to the acceleration control module, the reverse controller, acceleration control module and prediction module are combined to select the optimal voltage vector to act on the inverter through value function optimization, thereby controlling the flat wire permanent magnet hub motor.

[0101] See Figure 5 and Figure 6 A simulation comparison was conducted between the traditional vector controller and the inverse model predictive controller of this invention for a flat wire permanent magnet hub motor under variable speed conditions. The simulation conditions were: load torque 25 N·m, first set speed 40 rpm, second set speed 100 rpm, and third set speed 80 rpm. The simulation verified typical operating conditions such as electric vehicle start-up, acceleration, and deceleration. The results show that the variable speed response times under the traditional vector control are 0.6 s, 0.8 s, and 0.4 s, respectively, while the variable speed response times of the controller proposed in this invention are 0.3 s, 0.4 s, and 0.1 s, respectively. Compared with the traditional controller, the inverse model predictive controller for the flat wire permanent magnet hub motor based on the acceleration value function proposed in this invention has a shorter response time, and the flat wire permanent magnet hub motor has a faster response speed.

[0102] See Figure 7 and Figure 8 Simulation waveforms of a flat-wire permanent magnet hub motor under sudden load changes on an uneven road surface were compared between a conventional vector controller and the controller proposed in this invention. Simulation conditions were: flat-wire permanent magnet hub motor speed 80 rpm, initial load torque 10 N·m, first set load torque 20 N·m, and second set load torque 36 N·m. Simulation results showed that under conventional vector control, the speed drops during load changes were 4 rpm and 6 rpm, respectively, while the speed drops of the controller proposed in this invention were 2 rpm and 4 rpm, respectively. These results demonstrate that the flat-wire permanent magnet hub motor reverse-push MPC proposed in this invention, based on the acceleration value function, can effectively reduce the impact of load torque changes on the performance of the flat-wire permanent magnet hub motor, exhibiting faster speed and torque response.

[0103] Therefore, this invention improves the rapid response and anti-interference capabilities of the flat wire permanent magnet hub motor under complex road conditions. The controller incorporates a reverse-engineering controller into model predictive control, effectively avoiding the problems of integral saturation, overshoot, and poor dynamic quality inherent in traditional PI controllers, thus achieving rapid response of the flat wire permanent magnet hub motor under sudden speed changes. Simultaneously, an adaptive acceleration model prediction value function based on load torque is constructed to improve the rapid and accurate tracking of the drive torque of the flat wire permanent magnet hub motor under load variations.

[0104] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor, characterized by: (1) Construct a system with angular velocity deviation e as input and q-axis current estimate as input. For the output reverse controller: Differentiate the angular velocity deviation e and construct the corresponding Lyapunov function V1 = e 2 / 2 and differentiate; Based on the Lyapunov stability condition dV1 / dt<0, the q-axis current is calculated as follows: Calculate the load torque estimation error T is the estimated load torque value. L This is the nominal value of the load torque; Based on the formula for calculating the q-axis current and the torque estimation error ΔT L Obtain the q-axis current estimate (2) Constructing a system based on angular velocity deviation e and current tracking error For input, i d and i q An acceleration control module that outputs acceleration 'a' as the current component: Based on angular velocity deviation e and current tracking error e q e d Construct the corresponding Lyapunov function V2 = (e 2 +e q 2 +e d 2 +ΔT L 2 Given dV2 / dt < 0 and r1 > 0, solve for the acceleration a based on the Lyapunov stability condition dV2 / dt < 0. (3) Based on the mathematical model of the motor, obtain the predicted values ​​of the d-axis and q-axis currents i at the next sampling time k+1. d (k+1),i q (k+1) and the acceleration prediction value a(k+1), and the current prediction values ​​i of each d and q axis. d (k+1),i q (k+1) and the acceleration prediction value a(k+1) are expressed by the value function. The evaluation is performed, where A is the weighting coefficient. The voltage vector corresponding to the predicted value that minimizes Err(i) is selected and controlled by the inverter to control the motor.

2. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 1, characterized in that: The acceleration mentioned Coefficients k1>0, k2>0, L d L q These are the d-axis and q-axis stator inductances, respectively; u d u q The stator voltages are d-axis and q-axis, respectively; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the angular velocity; ψ f It is a permanent magnet flux linkage.

3. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 2, characterized in that: The derivative of the Lyapunov function is: The q-axis current i q =2(Bω+T) L +k3Je) / (3Pψ f ).

4. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 1, characterized in that: The q-axis current estimate J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the angular velocity; ψ f is the permanent magnet flux linkage, and k3 is the Lyapunov coefficient.

5. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 1, characterized in that: Detecting the three-phase current i of a flat wire permanent magnet hub motor a i b i c The three-phase current i a i b i c After sequential Clark and Park transformations, the current components i in the two-phase rotating coordinate system are obtained. d and i q .

6. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 2, characterized in that: The predicted d-axis and q-axis current values ​​i d (k+1),i q (k+1) and the predicted acceleration value a(k+1) are respectively: i d (k+1)=i d (k)+T s [u d (k)-Ri d (k)+ω re (k)L q i q (k)] / L d , i q (k+1)=i q (k)+T s [u q (k)-Ri q (k)-ω re (k)L d i d (k)-ω re (k)ψ f ] / L q , a(k+1)=dω / dt=3Pψ f i q (k+1) / (2J)-(Bω re +T L ) / J-3P(L q -L d )i q (k+1)i d (k+1) / (2J), i d (k), i q (k) represents the d-axis and q-axis current feedback values ​​at the current time k; u d (k), u q (k) represents the d-axis and q-axis voltages at the current time k; ω re (k) represents the rotor's electric angular velocity; T s R is the sampling period, and R is the stator resistance.

7. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 1, characterized in that: The actual motor speed ω is measured using a rotary transformer, and the actual speed ω is compared with the given speed ω. * The difference between the values ​​is used to obtain the angular velocity deviation e.

8. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 1, characterized in that: The mathematical model of the motor is: u d u q These are the stator voltages along the d and q axes, respectively; i d i q These are the d-axis and q-axis stator currents, respectively; R is the stator resistance; L d L q These are the d-axis and q-axis stator inductances, respectively; T L ψ is the nominal load torque; J is the moment of inertia; B is the coefficient of viscous friction; P is the number of pole pairs; ω is the rotor mechanical angular velocity; ψ f It is a permanent magnet flux linkage.

9. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 8, characterized in that: Based on the mathematical model of the motor, the derivative of the angular velocity deviation e is obtained:

10. The method for constructing a reverse-propulsion model predictive controller for a flat wire permanent magnet hub motor according to claim 1, characterized in that: The weighting coefficient A > 0.

Images