A robust deadbeat model predictive method for permanent magnet synchronous motor

Abstract

The application relates to the field of motor control and discloses a robust model predictive control method for a permanent magnet synchronous motor, which comprises the following steps: sampling stator current and stator voltage of the permanent magnet synchronous motor and outputting the stator current and the stator voltage to an MRAS observer; calculating a rotating speed estimation value and a resistance estimation value according to the obtained parameters by the MRAS observer; receiving corresponding parameters of the motor and the parameters output by the MRAS observer by using a deadbeat predictive controller, calculating an ideal voltage vector in each sampling period, and outputting the ideal voltage vector to a discrete space vector modulation module; determining a standardized triangle area where the ideal voltage vector is located through processing of the discrete space vector modulation module, taking three vertex vectors of the standardized triangle area as candidate voltage vectors, and selecting an optimal control vector to output to an inverter. The application has the advantages of high efficiency, low pulsation output, small calculation amount, high-precision voltage prediction and control under the condition of parameter uncertainty, and improved robustness.

Description

Technical Field

[0001] This invention relates to the field of motor control, and more specifically to a robust method for predicting deadbeat models of permanent magnet synchronous motors. Background Technology

[0002] In traditional permanent magnet synchronous motor control strategies, finite control set model predictive control (FCIMM) is used. FCS-MPC This method has attracted widespread attention due to its rapid dynamic response and excellent steady-state performance. However, its inherent drawbacks include high torque ripple, large computational load, and its control performance is easily affected by the accuracy of motor model parameters.

[0003] Another common control scheme is single-vector model predictive torque control (MPC). MPTC This scheme applies only one voltage vector to the system in each sampling period. Although it can maintain a fast dynamic response speed, it will result in large torque ripple and current harmonics, making it difficult for the system to achieve ideal control accuracy and stability during steady-state operation.

[0004] Furthermore, deadbeat control theoretically possesses superior dynamic response characteristics and stronger steady-state performance; however, this method is more sensitive to changes in motor parameters, and its system robustness is inferior to traditional finite set model predictive control. FCS-MPC )method.

[0005] In view of this, this application has conducted in-depth research on this basis, resulting in this case. Summary of the Invention

[0006] The purpose of this invention is to provide a robust deadbeat model prediction method for permanent magnet synchronous motors, which can achieve high efficiency, low ripple output, and low computational load. At the same time, it can realize high-precision voltage prediction and control under parameter uncertainty conditions, while improving robustness.

[0007] To achieve the above objectives, the solution of the present invention is: a robust method for predicting deadbeat models of permanent magnet synchronous motors, comprising the following steps: Step 1, sampling the permanent magnet synchronous motor in... The stator current and stator voltage at time t are denoted as . and And output it to the MRAS observer;

[0008] Step 2: The MRAS observer calculates the rotor speed estimate and stator resistance estimate for each sampling period based on the acquired parameters;

[0009] Step 3: Receive the reference torque and flux linkage amplitude of the permanent magnet synchronous motor using a deadbeat predictive controller. The stator current at each time point, along with the rotor speed estimate and stator resistance estimate output by the MRAS observer, are used to calculate the ideal voltage vector in each sampling period and output it to the discrete space vector modulation module.

[0010] Step 4: In the discrete space vector modulation module, a process is formed in... α - β The voltage plane in the coordinate system is divided into multiple standardized triangular regions according to the direction of the parent boundary line. The standardized triangular regions are determined by calculating the geometric distance between the ideal voltage vector and the three selected parent boundary lines. The voltage vectors corresponding to the three vertices of the standardized triangular regions are used as candidate voltage vectors. Then, the optimal control vector is selected according to the minimization criterion.

[0011] Step 5: Calculate the switching duration corresponding to the optimal control vector and output it. PWM The signal is sent to the inverter, and the inverter outputs... Stator voltage at time and The stator current is fed into the MRAS observer at a certain time, and then step 2 is repeated.

[0012] Step 2 includes the following steps;

[0013] Step 2-1: Establish a reference model and calculate based on the reference model. The reference flux linkage value is constantly monitored and transmitted to the error calculation module; wherein, the reference model is... In the formula, express The flux linkage value calculated from the reference model at each time step. express Stator voltage vector at time , Indicates nominal resistance. express Stator current vector at any given moment Indicates the sampling period of the control system;

[0014] Step 2-2: Establish an adaptive prediction model and calculate the results based on the adaptive prediction model. Stator flux linkage estimate at time and The estimated stator current value at time t is obtained and transmitted to the error calculation module described below, wherein the adaptive prediction model is:

[0015] ,

[0016] ;

[0017] In the formula, They represent Time and The estimated value of the stator flux linkage at time t. express Estimated stator resistance at time [time] This represents the rotor speed estimate output by the MRAS observer during the current sampling period. express Stator current estimate at time [time] Represents the imaginary unit. This represents the mapping function used to calculate the stator current estimate from the stator flux linkage estimate and the rotor speed estimate;

[0018] Steps 2-3: Obtain the parameter error using the error calculation formula, which is as follows:

[0019] , ;

[0020] In the formula, express Momentary current error, express Momentary flux error express The stator reference current value at that moment;

[0021] Steps 2-4, using Pi The adaptive adjustment model adaptively adjusts the parameter error and then adjusts the adaptively adjusted parameter. Rotor speed estimate at time and The estimated stator resistance values ​​at each time step are fed back to the adaptive prediction model for state prediction in the next sampling period. Pi The adaptive adjustment model is as follows:

[0022] ,

[0023] ;

[0024] In the formula, , This represents the adjustment coefficient. Indicates passage Pi The adaptive adjustment model was calculated to obtain Estimated rotor speed at any given time , This represents the integral adjustment coefficient. Indicates a discrete-time index;

[0025] Steps 2-5: The MRAS observer outputs the updated current sampling period. and , This represents the estimated rotor speed. This represents the estimated value of the stator resistance.

[0026] In step 3, the discretized mathematical model of the permanent magnet synchronous motor is established as follows:

[0027] In the formula, Indicates the first Stator current vector within each sampling period, This represents the ideal voltage vector. This represents the output of the MRAS observer. Estimated stator resistance at time [time] This indicates the value estimated by the MRAS observer based on the stator flux linkage during the current sampling period. and rotor speed estimate The calculated back electromotive force, Indicates stator inductance, Indicates the sampling period of the control system;

[0028] Next, in the The stator current vector is obtained through the current sampling device within each sampling period. A prediction model is established based on the stator current vector. To achieve deadbeat tracking of the reference current by the stator current vector in the next sampling period, the predicted current value in the discretized mathematical model is set to... Compared with reference current value Equal to each other, and substituted into the discretized mathematical model, the ideal voltage vector is obtained as follows:

[0029] .

[0030] In step 4, α - β In the coordinate system, the voltage plane formed by the inverter output voltage is hexagonal in shape. The voltage plane includes three sets of parallel parent boundary lines, with three non-parallel parent boundary lines serving as directional baselines.

[0031] Includes the following steps:

[0032] Step 4-1: Determine the standardized triangular region. Let the coordinates of the endpoints of the ideal voltage vector be ( ). u α , u β The equations for the three non-parallel parent boundary lines in the voltage plane are established as follows:

[0033] ,

[0034] ,

[0035] ;

[0036] In the formula, , , These represent the three non-parallel parent boundary lines of the voltage plane. u α , u β They represent respectively Clarke After coordinate transformation α - β Two-phase equivalent voltage components in a coordinate system This indicates the DC bus voltage of the inverter;

[0037] Then, the perpendicular distance from the endpoint of the ideal voltage vector to the three parent boundary lines is calculated using the following formula: , , ;

[0038] Next, , , The normalized distance integers are obtained by quantization, and the distance integers are... ,in, , Represents the normalization coefficient. This represents the number of equal divisions of the hexagon within a sampling period;

[0039] Then, the standardized triangular region is determined, and its number is... ,in, , , These represent the normalized integer coordinates of the three sides, respectively.

[0040] Step 4-2: Determine three candidate voltage vectors;

[0041] First, calculate the coordinates of the reference point in the standardized triangular region. , The calculation formula is as follows:

[0042] , , In the formula, This indicates that the point is at α Coordinates on the axis This indicates that the point is at β Coordinates on the axis Indicates the basic voltage step size;

[0043] Then, based on the coordinates of the reference point ( , Given the basic voltage step size Δ, calculate the candidate voltage vectors corresponding to the three vertices of the standardized triangular region. The complex form of these vectors is as follows:

[0044] , , , Represents the imaginary unit;

[0045] Step 4-3: Calculate the future response of each candidate voltage vector using a cost function, which is:

[0046] ,

[0047] In the formula, Indicates the reference electromagnetic torque. Indicates the reference stator flux linkage amplitude; This represents the electromagnetic torque estimate obtained in the next sampling period under the applied candidate voltage vector. These represent the stator flux linkage estimates obtained in the next sampling period under the applied candidate voltage vector; , , , These represent the weighting coefficients for torque, flux linkage, switching losses, and current constraints, respectively. This indicates that the inverter has switched from its current switching state to the candidate voltage vector. V ( i The number of switching actions required for the corresponding switching state is used to measure switching losses. Indicates the current constraint penalty term; The index number represents the candidate voltage vector, used to distinguish different candidate voltage vectors. Indicates the first One candidate voltage vector;

[0048] Step 4-4: Establish the optimal control vector selection formula, and select the candidate voltage vector with the smallest comprehensive performance index as the optimal control vector. The optimal control vector selection formula is as follows:

[0049] .

[0050] The inverter is a three-phase two-level inverter, and the inverter has six effective switching states.

[0051] After step 4-2, according to and + The parity correction sign, if and + If the parity is the same, then the vertex coordinates corresponding to the candidate voltage vector are negative. and + If the parity of the candidate voltage vector is different, a positive sign is taken when calculating the vertex coordinates corresponding to the candidate voltage vector. The negative sign and the negative sign are used to distinguish the directionality of adjacent regions during the numbering of the standardized triangular region.

[0052] After adopting the above method, the present invention has the following beneficial effects: 1. The dual-layer control structure is adopted: the first layer realizes the prediction and control of voltage under the condition of parameter uncertainty through the deadbeat prediction model that does not rely on velocity estimation; the second layer adopts an improved discrete space vector modulation module, which transforms the traditional method of enumerating multiple virtual voltage vectors into only evaluating three candidate voltage vectors, thereby completing the synthesis of predicted voltage in a finite set of candidate voltage vectors, reducing the amount of computation, and helping to reduce output ripple. The two parts are functionally independent and cooperate with each other during operation. The former provides an ideal voltage reference, and the latter ensures the continuity of physical realization, thereby improving the real-time performance and robustness of the system.

[0053] 2. This invention employs an MRAS observer based on Lyapunov theory, which can simultaneously acquire real-time stator resistance estimates and rotor speed estimates, thereby effectively reducing drift errors and ensuring system stability. Attached Figure Description

[0054] Figure 1 This is a system flowchart of the present invention.

[0055] Figure 2 This is a control block diagram of the MRAS observer in this invention.

[0056] Figure 3 This is the voltage plane containing the ideal voltage vector endpoints in this invention. Detailed Implementation

[0057] To further explain the technical solution of the present invention, the present invention will be described in detail below through specific embodiments.

[0058] A robust deadbeat-free model prediction method for permanent magnet synchronous motors (PMSMs) is proposed. This method is based on common control systems used in PMSM control, such as... Figure 1 As shown, the control system includes an inverter, which is a conventional two-level three-phase inverter. A conventional two-level three-phase inverter has eight switching states, and in each of these eight switching states, it generates six non-zero voltage vectors. U 1~U 6) and two zero-voltage vectors. These non-zero voltage vectors are the basic voltage vectors. In other words, the two-level three-phase inverter has six effective switching states. These six effective switching states are all the inverter's conventional switching states, and therefore will not be described further. In this embodiment, the six basic voltage vectors generated by the three-phase two-level inverter under six different switching states are as follows: Figure 3 As shown, they are 60° apart in space, and their endpoints together form a regular hexagon. This regular hexagon represents the limit range of the inverter's output voltage, and the space formed is the voltage plane, which is used for subsequent space vector modulation.

[0059] In this embodiment, as Figures 1-3 As shown, the deadbeat model prediction method for permanent magnet synchronous motors includes the following steps.

[0060] Step 1: Sample the permanent magnet synchronous motor to obtain the current stator current and current stator voltage. In this embodiment, the current sampling time is defined as... At time , the next sampling time is At that moment, the current stator current and the current stator voltage correspond to respectively: and Then sampled and The output is sent to the MRAS observer.

[0061] Furthermore, permanent magnet synchronous motors are equipped with current sampling devices (such as current sensors) and voltage sensors. The current sensor samples the current stator current, and the voltage sensor samples the current stator voltage. It is worth noting that configuring current and voltage sensors on conventional permanent magnet synchronous motors is standard practice in the motor industry, and therefore will not be elaborated upon further.

[0062] Step 2: The MRAS observer, based on the acquired data... and Calculate the estimated rotor speed. ( Stator resistance estimate ( ).

[0063] Specifically, step 2-1: Establish a reference model, and calculate based on the reference model. The magnetic flux linkage value is constantly referenced and transmitted to the error calculation module.

[0064] To elaborate, the reference model is based on the stator voltage equation of the motor, which is as follows: In the formula, Represents the stator voltage vector. Represents the stator current vector. Indicates stator resistance. Represents the stator flux linkage vector. Indicates the rotor speed. Represents the imaginary unit, used to denote... α - β The voltage vector in the coordinate system is represented in complex form.

[0065] because The velocity coupling term makes the flux linkage estimation sensitive to rotor speed changes. Therefore, to eliminate the above effects, the components related to speed and its high-frequency disturbances are classified into the same function. Defined as: In the formula, This represents the high-frequency and nonlinear error terms. From this, a velocity-independent reference equation can be obtained, which is: In the formula, express Stator voltage vector at time , Indicates nominal resistance. express Stator current vector at any given moment This represents the reference flux linkage value at the current moment. This indicates the sampling period of the control system.

[0066] Furthermore, due to the above function The impact of this can be ignored in the modeling, so the above reference model is: .in, Indicates the sampling period. This represents the calculated next time step (i.e. (Time) Reference flux linkage value.

[0067] It is worth mentioning that the above reference model relies on measurable voltage and current, but does not include an unknown velocity term. Based on this reference model, the following can be calculated: Constantly referencing magnetic flux linkage value Furthermore, this is combined with the magnetic flux linkage equation. Calculate the next moment (i.e.) (Time) Reference current value Finally, the obtained reference flux linkage value and reference current value are transmitted to the error calculation module described below.

[0068] Step 2-2: Calculate using an adaptive prediction model. Stator flux linkage estimate at time t and The estimated value of the stator current at any given time is transmitted to the error calculation module described below.

[0069] To elaborate, the adaptive prediction model is an online observer based on a mathematical model of an electric motor, whose goal is to utilize the current time (i.e., The estimated parameters at time (i.e., time step 1) predict the motor's position at the next time step (i.e., time step 2). The flux linkage and current estimates at time ( ) are based on state predictions of the discretized voltage equations, as detailed below.

[0070] The flux linkage prediction equation is established as follows:

[0071] ,

[0072] In the formula, They represent Time and The estimated value of the stator flux linkage at time t. express Estimated stator resistance at time [time] This represents the rotor speed estimate output by the MRAS observer during the current sampling period.

[0073] The current prediction equation is established as follows: In the formula, express Stator current estimate at time, function f This represents the flux linkage-current relationship for solving the current from the flux linkage and rotor speed, which is a mapping function for calculating the stator current estimate from the stator flux linkage estimate and the rotor speed estimate.

[0074] Steps 2-3: In the error calculation module, the parameters transmitted from the reference model and the adaptive prediction model are processed using the error calculation formula to obtain the parameter acquisition error. This parameter error is then sent to... Pi In the adaptive adjustment model.

[0075] To elaborate, the error calculation formula is as follows:

[0076] , ;

[0077] In the formula, express Momentary current error, express Momentary flux error express The stator reference current value at a given time.

[0078] Steps 2-4: Employing Lyapunov functions as the basis for... Pi The adaptive adjustment model adaptively adjusts the parameter error and then adjusts the result accordingly. Rotor speed estimate at time and The estimated stator resistance values ​​at each time point are fed back to the adaptive prediction model for state prediction in the next sampling period.

[0079] To elaborate, let's first define a Lyapunov candidate function to characterize the energy change of the system. This Lyapunov candidate function is: In the formula, Let Lyapunov function be denoted as , and let Lyapunov function be denoted as , which represents the system Total error energy at any given time; express flux linkage error at time , express Current error at any given time.

[0080] To ensure that the Lyapunov candidate function is monotonically decreasing, its difference must satisfy the following formula: In the formula, This represents the change in the Lyapunov function, from which the basic form of the parameter update rate can be obtained:

[0081] ,

[0082] ;

[0083] In the formula, , This represents the adjustment coefficient. Indicates passage Pi The adaptive adjustment model was calculated to obtain Estimated rotor speed at any given time.

[0084] To improve steady-state accuracy and response speed, an integral compensation term is introduced based on the basic form of the parameter update rate, forming... Pi Renew the structure in a formula. Pi The update structure is as follows Pi Adaptive adjustment model, Pi The adaptive adjustment model is as follows:

[0085] ,

[0086] ;

[0087] In the formula, , The integral adjustment coefficient is represented by the summation term, which represents the summation from the initial time to the current time (i.e., ...). Accumulated error at (time point); This represents the discrete-time index, used to represent historical moments from the initial sampling period to the kth sampling period.

[0088] Steps 2-5: The MRAS observer is based on the above. Pi The adaptive adjustment model calculates the latest parameter estimates and outputs the updated rotor speed estimate for the current sampling period. and stator resistance estimate It is then transmitted to subsequent modules for beat-free predictive control.

[0089] Step 3: Use a deadbeat predictive controller to receive the reference torque of the permanent magnet synchronous motor. Magnetic flux amplitude , Stator current at any time And the real-time estimated motor rotor speed output by the MRAS observer. and real-time estimation of stator resistance And calculate the ideal voltage vector in each sampling period. It is then output to the discrete space vector modulation module.

[0090] To elaborate, the discretized mathematical model of the permanent magnet synchronous motor is established as follows:

[0091] ;

[0092] Will and Forced equality yields the ideal voltage vector. The formula for the ideal voltage vector is:

[0093] .

[0094] In the formula, Represents the ideal voltage vector. This represents the value estimated by the MRAS observer. Estimated stator resistance at time [time] Indicates stator inductance, This represents the back electromotive force (EMF) obtained by the MRAS observer based on the flux linkage estimate and the rotor speed estimate. This back EMF needs to be combined with the rotor electric angular velocity obtained from the MRAS observation. The calculation is performed using the following formula:

[0095] ,

[0096] In the formula, This is the estimated value of the stator flux linkage. j It is the imaginary unit.

[0097] Step 4: Forming a discrete space vector modulation module. α - βThe voltage plane in the coordinate system is divided into multiple standardized triangular regions according to the direction of the parent boundary line. The standardized triangular regions are determined by calculating the geometric distance between the ideal voltage vector and the three selected parent boundary lines. The voltage vectors corresponding to the three vertices of the standardized triangular regions are used as candidate voltage vectors. Then, the optimal control vector is selected according to the minimization criterion.

[0098] To elaborate, step 4-1 is to determine the standardized triangular region.

[0099] Because in α - β In the coordinate system, the voltage plane of the inverter output voltage is distributed in a regular hexagon. Its geometric boundary contains three pairs of parallel line segments, that is, the voltage plane includes three sets of parallel parent boundary lines. Therefore, only three non-parallel parent boundary lines need to be defined as directional baselines, and the boundary structure of the entire voltage limit space can be generated by translating along these three parent boundary lines. The regular hexagon can be represented by three non-parallel parent boundary lines. In this embodiment, the three non-parallel parent boundary lines are respectively... , , Their equations are:

[0100] ,

[0101] ,

[0102] ;

[0103] In the formula, u α , u β They represent respectively Clarke After coordinate transformation α - β Two-phase equivalent voltage components in a coordinate system This indicates the DC bus voltage of the inverter.

[0104] Let the coordinates of the endpoints of the ideal voltage vector (i.e., DB-VV) be ( u α , u β The perpendicular distances from the endpoints of the ideal voltage vector to the three selected parent boundary lines are denoted as follows: d 1, d 2, d 3. The calculation formula is as follows: , , Thus, through three sets of vertical distances ( , , Together, they determined the position of the ideal voltage vector in the regular hexagonal grid formed by the periodic translation of the parent boundary line.

[0105] To facilitate digital indexing and fast addressing, and to facilitate digital implementation, , , The normalized distance integers are obtained by quantization respectively, and these distance integers are respectively , , and with The unique identifier of the standardized triangle region containing the ideal voltage vector is represented in the form of , as follows.

[0106] The quantized distance integer is: , ,in, , This represents the normalization coefficient, used to map continuous geometric distances to integer indices; This represents the number of equal divisions of the hexagon within a sampling period.

[0107] Then, the unique number of the above-mentioned standardized triangular region is: , h m Represents the normalized integer distance, and represents the ideal voltage vector at the _th ... m The quantization level number along the boundary line is a key parameter for achieving rapid geometric positioning. Among these parameters, the triplet ( , , It is a location index used to uniquely identify the standardized triangular region. 。

[0108] It is worth mentioning that, through the geometric positioning method, it is possible to... α - β In the coordinate system, directly based on the quantized distance integer , , Precisely locate the standardized triangular region containing the ideal voltage vector (DB-VV), a location process based on the parent boundary line. , , The periodic translational characteristics divide the regular hexagonal voltage limit region into several standardized small triangular regions, i.e., standardized triangular regions. The position index of each standardized triangular region can be determined by three quantized index integers ( , , The only certainty is that among them These represent the relative distances of the ideal voltage vector to the three parent boundary lines, and also reflect its relative geometric position within the triangle.

[0109] Step 4-2: Determine three candidate voltage vectors.

[0110] Based on step 4-1, the standardized triangular region containing the ideal voltage vector was determined, and the coordinates of the reference point and the basic voltage step size of this region were calculated. The calculation formula is as follows:

[0111] , , In the formula, This indicates that the point is at α Coordinates on the axis This indicates that the point is at β Coordinates on the axis; This represents the basic voltage step size, corresponding to the minimum voltage resolution after discrete space vector modulation (DSVM) subdivision.

[0112] Then, based on the reference point coordinates of this standardized triangular region ( a , b Given the basic voltage step size, calculate the candidate voltage vectors corresponding to the three vertices of the normalized triangular region. Their complex form is as follows:

[0113] , , , It represents the imaginary unit. It's worth mentioning that... , , The geometric relationship forms an equilateral triangular unit, whose interior points can be described by a linear combination of the three vertices, which is used to approximate a continuous ideal voltage vector in discrete space.

[0114] Furthermore, when the sampling period is divided into even numbers, to ensure that the geometric direction of the candidate voltage vector is consistent with the numbering direction, it is necessary to... and + The parity correction sign is as follows: if and + If the parity of the candidate voltage vectors is the same, then a negative sign is taken when calculating the vertex coordinates of the candidate voltage vectors. Otherwise, use "+". The positive and negative signs here are used to distinguish the directionality of adjacent areas during triangle numbering, preventing reverse deviations in vector indexing during calculations, and have no effect on voltage amplitude.

[0115] In this way, by following the steps above, the three candidate voltage vectors closest to the ideal voltage can be quickly determined without traversing all virtual voltage vectors.

[0116] Step 4-3: Calculate the future response of each candidate voltage vector using a cost function, which is:

[0117] ,

[0118] In the formula, Indicates the reference electromagnetic torque. Indicates the reference stator flux linkage amplitude; This represents the electromagnetic torque estimate obtained in the next sampling period under the applied candidate voltage vector. These represent the stator flux linkage estimates obtained in the next sampling period under the applied candidate voltage vector; , , , These represent the weighting coefficients for torque, flux linkage, switching losses, and current constraints, respectively. This indicates that the inverter is switching from its current switching state to the candidate voltage vector. The number of switching actions in the corresponding switching state is used to suppress excessively high switching frequency; This represents the current constraint penalty term, which takes a larger value when the estimated stator current exceeds the set limit, and zero otherwise, and is used to prevent overcurrent.

[0119] Step 4-4: Establish the optimal control vector selection formula, and select the candidate voltage vector with the smallest comprehensive performance index as the optimal control vector. The optimal control vector selection formula is as follows:

[0120] .

[0121] It should be noted that the optimal control vector , which is the actual output voltage of the inverter during the current sampling period.

[0122] Thus, by employing the above method, while ensuring the accuracy of torque and flux regulation, switching frequency fluctuations and overcurrent risks can be effectively suppressed, achieving the desired voltage vector. High-precision approximation and robust control.

[0123] Step 5: The discrete space vector modulation module calculates the switching duration corresponding to the optimal control vector and outputs it. PWM Signal (i.e.) Sa , Sb , Sc (This is given to the inverter, which is based on...) PWM The signal is switched on and off, and output is also performed. Stator voltage at time and The stator current is fed into the MRAS observer at a certain time, and then step 2 is repeated.

[0124] This invention presents a robust deadbeat model prediction method for permanent magnet synchronous motors (PMSMs). It organically integrates a predicted MRAS observer with algebraic geometric discrete space vector modulation into a deadbeat model predictive control structure, forming a self-correcting predictive control method with low computational complexity, high robustness, and no speed measurement requirement. Specifically, it employs an MRAS observer based on Lyapunov theory to simultaneously predict stator resistance and rotor speed, effectively reducing drift error and ensuring stability. Furthermore, it constructs a deadbeat predictive controller that eliminates the speed term, directly avoiding control drift caused by speed estimation errors. Moreover, it uses an improved discrete space vector modulation method to reduce the number of candidate voltage vectors to three, significantly lowering computational complexity.

[0125] The above description is only a preferred embodiment of this invention. Any equivalent changes and modifications made within the scope of the claims of this invention shall fall within the scope of the claims of this invention.

Claims

1. A robust method for predicting deadbeat models of permanent magnet synchronous motors, characterized in that, The steps include: Step 1, sampling the permanent magnet synchronous motor. The stator current and stator voltage at time t are denoted as . and And output it to the MRAS observer; Step 2: The MRAS observer calculates the rotor speed estimate and stator resistance estimate for each sampling period based on the acquired parameters; Step 2 includes the following steps; Step 2-1: Establish a reference model and calculate based on the reference model. The reference flux linkage value is constantly monitored and transmitted to the error calculation module; wherein, the reference model is... In the formula, express The flux linkage value calculated from the reference model at each time step. express Stator voltage vector at time , Indicates nominal resistance. express Stator current vector at any given moment Indicates the sampling period of the control system; Step 2-2: Establish an adaptive prediction model and calculate the results based on the adaptive prediction model. Stator flux linkage estimate at time and The estimated stator current value at time t is transmitted to the error calculation module described below, wherein the adaptive prediction model is: ， ； In the formula, They represent Time and The estimated value of the stator flux linkage at time t. express Estimated stator resistance at time [time] This represents the rotor speed estimate output by the MRAS observer during the current sampling period. express Stator current estimate at time [time] Represents the imaginary unit. This represents the mapping function used to calculate the stator current estimate from the stator flux linkage estimate and the rotor speed estimate; Steps 2-3: Obtain the parameter error using the error calculation formula, which is as follows: ， ； In the formula, express Momentary current error, express Momentary flux error express The stator reference current value at time [time]. This represents the flux linkage value calculated from the reference model; Steps 2-4, using Pi The adaptive adjustment model adaptively adjusts the parameter error and then adjusts the adaptively adjusted parameter. Rotor speed estimate at time and The estimated stator resistance values ​​at each time step are fed back to the adaptive prediction model for state prediction in the next sampling period. Pi The adaptive adjustment model is as follows: ， ； In the formula, , This represents the adjustment coefficient. Indicates passage Pi The adaptive adjustment model was calculated to obtain Estimated rotor speed at any given time , This represents the integral adjustment coefficient. Indicates a discrete-time index; Steps 2-5: The MRAS observer outputs the updated current sampling period. and , This represents the estimated rotor speed. This represents the estimated value of the stator resistance; Step 3: Receive the reference torque and flux linkage amplitude of the permanent magnet synchronous motor using a deadbeat predictive controller. The stator current at each time point, along with the rotor speed estimate and stator resistance estimate output by the MRAS observer, are used to calculate the ideal voltage vector in each sampling period and output it to the discrete space vector modulation module. Step 4: In the discrete space vector modulation module, a process is formed in... α - β The voltage plane in the coordinate system is divided into multiple standardized triangular regions according to the direction of the parent boundary line. The standardized triangular regions are determined by calculating the geometric distance between the ideal voltage vector and the three selected parent boundary lines. The voltage vectors corresponding to the three vertices of the standardized triangular regions are used as candidate voltage vectors. Then, the optimal control vector is selected according to the minimization criterion. Step 5: Calculate the switching duration corresponding to the optimal control vector and output it. PWM The signal is sent to the inverter, and the inverter outputs... Stator voltage at time and The stator current is fed into the MRAS observer at a certain time, and then step 2 is repeated.

2. The robust prediction method for a deadbeat-free model of a permanent magnet synchronous motor according to claim 1, characterized in that: In step 3, the discretized mathematical model of the permanent magnet synchronous motor is established as follows: In the formula, Indicates the first Stator current vector within each sampling period, This represents the ideal voltage vector. This represents the output of the MRAS observer. Estimated stator resistance at time [time] This indicates the value estimated by the MRAS observer based on the stator flux linkage during the current sampling period. and rotor speed estimate The calculated back electromotive force, Indicates stator inductance, Indicates the sampling period of the control system; Next, in the The stator current vector is obtained through the current sampling device within each sampling period. A prediction model is established based on the stator current vector. To achieve deadbeat tracking of the reference current by the stator current vector in the next sampling period, the predicted current value in the discretized mathematical model is set to... Compared with reference current value Equal to each other, and substituted into the discretized mathematical model, the ideal voltage vector is obtained as follows: 。 3. A robust prediction method for a deadbeat-free model of a permanent magnet synchronous motor according to claim 1 or 2, characterized in that: In step 4, α - β In the coordinate system, the voltage plane formed by the inverter output voltage is hexagonal in shape. The voltage plane includes three sets of parallel parent boundary lines, with three non-parallel parent boundary lines serving as directional baselines.

4. The robust prediction method for a deadbeat-free model of a permanent magnet synchronous motor according to claim 3, characterized in that: Includes the following steps: Step 4-1: Determine the standardized triangular region. Let the coordinates of the endpoints of the ideal voltage vector be ( ). u α , u β The equations for the three non-parallel parent boundary lines in the voltage plane are established as follows: ， ， ； In the formula, , , These represent the three non-parallel parent boundary lines of the voltage plane. u α , u β They represent respectively Clarke After coordinate transformation α - β Two-phase equivalent voltage components in a coordinate system This indicates the DC bus voltage of the inverter; Then, the perpendicular distance from the endpoint of the ideal voltage vector to the three parent boundary lines is calculated using the following formula: , , ; Next, , , The normalized distance integers are obtained by quantization, and the distance integers are... ,in, , Represents the normalization coefficient. This represents the number of equal divisions of the hexagon within a sampling period; Then, the standardized triangular region is determined, and its number is... ,in, , , These represent the normalized integer coordinates of the three sides, respectively. Step 4-2: Determine three candidate voltage vectors; First, calculate the coordinates of the reference point in the standardized triangular region. , The calculation formula is as follows: , , In the formula, This indicates that the point is at α Coordinates on the axis This indicates that the point is at β Coordinates on the axis Indicates the basic voltage step size; Then, based on the coordinates of the reference point ( , Given the basic voltage step size Δ, calculate the candidate voltage vectors corresponding to the three vertices of the standardized triangular region. The complex form of these vectors is as follows: , , , Represents the imaginary unit; Step 4-3: Calculate the future response of each candidate voltage vector using a cost function, which is: ， In the formula, Indicates the reference electromagnetic torque. Indicates the reference stator flux linkage amplitude; This represents the electromagnetic torque estimate obtained in the next sampling period under the applied candidate voltage vector. These represent the stator flux linkage estimates obtained in the next sampling period under the applied candidate voltage vector; , , , These represent the weighting coefficients for torque, flux linkage, switching losses, and current constraints, respectively. This indicates that the inverter has switched from its current switching state to the candidate voltage vector. V ( i The number of switching actions required for the corresponding switching state is used to measure switching losses. Indicates the current constraint penalty term; The index number represents the candidate voltage vector, used to distinguish different candidate voltage vectors. Indicates the first One candidate voltage vector; Step 4-4: Establish the optimal control vector selection formula, and select the candidate voltage vector with the smallest comprehensive performance index as the optimal control vector. The optimal control vector selection formula is as follows: 。 5. The robust prediction method for a deadbeat-free model of a permanent magnet synchronous motor according to claim 4, characterized in that: The inverter is a three-phase two-level inverter, and the inverter has six effective switching states.

6. The robust prediction method for a deadbeat-free model of a permanent magnet synchronous motor according to claim 4, characterized in that: After step 4-2, according to and + The parity correction sign, if and + If the parity is the same, then the vertex coordinates corresponding to the candidate voltage vector are negative. and + If the parity of the candidate voltage vector is different, a positive sign is taken when calculating the vertex coordinates corresponding to the candidate voltage vector. The negative sign and the negative sign are used to distinguish the directionality of adjacent regions during the numbering of the standardized triangular region.

Images