Direct voltage prediction control method of permanent magnet synchronous motor driving system

Abstract

The invention discloses a direct voltage prediction control method for a permanent magnet synchronous motor driving system, which is used for a motor side converter and a grid side converter which are connected in a grid-connected manner, and comprises the following steps of: evaluating a newly-defined cost function, selecting a voltage vector in the cost function, and calculating the direct voltage of the permanent magnet synchronous motor driving system; the selected voltage vector can generate a DC bus voltage closer to a set point at the moment of k + 2, and can reduce the active power of a subsequent voltage error, thereby realizing the adjustment of the DC link voltage and power flow; and aiming at the steady-state error of the driving system, a Kalman filter is added, and disturbance caused by parameter mismatching is estimated through the Kalman filter and is used for correcting a power expression, so that the steady-state power error is compensated.

Description

Technical Field

[0001] The invention relates to power electronics technology, and more specifically to a direct voltage prediction control method for a permanent magnet synchronous motor drive system. Background Art

[0002] Back-to-back (BTB) converters are often used in medium and large power motor drive systems. In the prior art, PI-MPC is one of the common solutions for controlling back-to-back converter-fed PMSG systems. It uses an additional PI controller to generate an active power reference for the internal power controller to track. However, this control method requires an external PI controller to form a PI control loop to generate a power set point (i.e., power reference) for the internal MPC controller. However, there is a common integral saturation problem in the PI control loop, and the dynamic response of the system is slow.

[0003] In view of this, the applicant conducted in-depth research on this basis, which led to the emergence of this case. Summary of the invention

[0004] The purpose of the present invention is to provide a direct voltage prediction control method for a permanent magnet synchronous motor drive system, which does not require an external PI control loop, avoids performance degradation due to the common integral saturation problem in the PI control loop, and improves dynamic response.

[0005] To achieve the above object, the solution of the present invention is: A direct voltage prediction control method for a permanent magnet synchronous motor drive system, the permanent magnet synchronous motor drive system comprising a motor-side converter and a grid-side converter connected to the grid, the control method comprising the following steps: S1. Establish the cost function as follows, , where is the motor side cost function, , and They are the active component of the cost function, the bus voltage component of the cost function and the reactive component of the cost function; Indicates the reactive power reference on the grid side; is the DC link voltage set point; is the DC bus voltage at time k+2, which is based on time The expression of The discrete model of time moves forward by one get; Adjustment coefficient, represents the set of positive real numbers; is the weighting factor used to adjust reactive power; represents the active power constraint, represents the constraint on reactive power; is the active power generated by the permanent magnet synchronous motor, = , is the torque reference, is the angular velocity of the permanent magnet synchronous motor, Calculated in the position estimator by the feedback encoder signal on the motor shaft from the permanent magnet synchronous motor; The active power value at time k+2 is obtained by shifting the instantaneous power formula forward by one The obtained expression predicts that; represents the estimated grid-side reactive power at time k+2; S2. Selecting a voltage vector from the cost function, the specific method is as follows: Adjust the term in the cost function to approach zero to track the reference variable, assuming =0 and =0, if the voltage vector middle , then in > The voltage vector will be was selected; If the voltage vector middle , then the voltage vector Select the one that produces less than of The voltage vector of exist Reaching steady state hour, , select The voltage vector.

[0006] Based on the disturbance compensation steady-state error in the correction power expression, the specific method is: S1-1. The mathematical model of the Kalman filter is as follows: (1), (2), In the formula, represents the active power disturbance, represents the reactive power disturbance, represents the active power disturbance at time k-1, represents the reactive power disturbance at time k-1, represents the active power at time k, is the reactive power at time k, represents the active power at time k-1, represents the reactive power at time k-1, and are the state vectors at time k and time k-1 respectively, is the sampling period; is the state matrix, , , , , , , is an artificially defined matrix symbol; where , , = , , , ; , , both are commonly used mathematical matrices; Represents the resistance in the grid-side converter The nominal value of represents the inductance in the grid-side converter The nominal value of is the sampling frequency, is the grid-side electrical angular velocity; is the covariance matrix Process noise: ,in Is a mathematical symbol representing a random variable Following the mean of 0, the covariance matrix is Normal distribution of S1-2: Based on the mathematical model of step 1-1, the prior estimate of the state vector is as follows, (3), In the formula, Represents the state vector A priori estimate of Represents the interval [ ]The state vector calculated within The posterior estimate of ; S1-3: Define the precision covariance of the prior estimate as follows, (4), where E stands for mathematical expectation and T stands for matrix transpose, both of which are mathematical symbols; The precision covariance is calculated as follows, , where ; Indicates that in the interval [ ] is the precision covariance of the posterior estimate of the state vector obtained in; S1-4: Kalman gain and the posterior estimate of the state vector The calculation formulas are as follows: , , The corrected power prediction expression is: ,in is updated for the next sampling period; is the identity matrix, which is a square matrix with all elements on the diagonal from the upper left corner to the upper right corner being 1 and all other elements being 0.

[0007] In step S2, Reaching steady state When Add to middle.

[0008] In step S1, the instantaneous power is: , where is the sampling frequency, is the grid-side electrical angular velocity, represents inductance, and They are respectively represented by grid-side voltage, grid-side current, inductive reactance and inductive resistance, where: represents the α component of the grid-side voltage at time k+1, represents the β component of the grid-side voltage at time k+1, represents the α component of the grid-side voltage at time k+1, Represents the β component of the grid-side voltage at time k+1.

[0009] After adopting the above structure, the present invention has the following beneficial effects: the present invention evaluates the newly defined cost function so that the selected voltage vector will not only generate a DC bus voltage closer to its set point at the moment k+2, but also further generate active power to reduce subsequent voltage errors, thereby realizing the regulation of the DC link voltage and power flow within a cost function.

[0010] Furthermore, the present invention estimates the disturbance caused by parameter mismatch through a Kalman filter and uses it to correct the power expression to eliminate steady-state deviation even under variable parameters, thereby improving system robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a back-to-back permanent magnet synchronous motor drive system.

[0012] Figure 2 This is the control schematic diagram of the traditional PI-MPC.

[0013] Figure 3 This is a control principle diagram of the dynamic predictive control method of the present invention.

[0014] Figure 4 It is a schematic diagram of experimental waveforms related to DC bus voltage and reactive power regulation in the dynamic predictive control method of the present invention. DETAILED DESCRIPTION

[0015] In order to further explain the technical solution of the present invention, the present invention is described in detail below through specific embodiments.

[0016] The present invention provides a direct voltage prediction control method for a permanent magnet synchronous motor drive system. Figure 1 As shown, the permanent magnet synchronous motor adopts a conventional back-to-back permanent magnet synchronous motor. Accordingly, the permanent magnet synchronous motor drive system adopts a conventional back-to-back permanent magnet synchronous motor drive system. The drive system includes a permanent magnet synchronous motor (PMSG), a motor-side converter (MSC) and a grid-side converter (GSC). The motor-side converter and the grid-side converter are connected to the grid in a conventional manner, and the permanent magnet synchronous motor (PMSG) is connected to the motor-side converter (MSC) in a conventional manner.

[0017] like Figure 1-4 As shown, the direct voltage prediction control method provided in this embodiment is performed based on the above-mentioned drive system, and the control method includes the following steps: S1. Establish the cost function as follows, , where is the motor side cost function, is the optimal voltage vector; is the active component of the cost function, is the bus voltage component of the cost function, is the reactive component of the cost function; where , and The expression is as follows: , = , .

[0018] The above is the DC link voltage set point, which is a manually set value; Indicates the reactive power reference on the grid side, which is a manually set value; Adjustment coefficient, represents the set of positive real numbers; is the weighting factor used to adjust reactive power; represents the active power constraint, Represents the reactive power constraint.

[0019] is the DC bus voltage at time k+2, which is based on time The expression of The discrete model of time moves forward by one The discrete model adopts the conventional first-order Euler discretization, which is the conventional technology, and the above It is the input quantity measured by the voltage sensor.

[0020] represents the active power generated by the permanent magnet synchronous motor, where = , is the torque reference, torque reference It is set by external artificial means; is the angular velocity of the permanent magnet synchronous motor, The angular velocity is calculated in the position estimator by the feedback encoder signal on the motor shaft of the permanent magnet synchronous motor; The calculation of is an existing conventional calculation, so it will not be described in detail.

[0021] The active power value at time k+2 is obtained by shifting the instantaneous power formula forward by one The obtained expression predicts that; Represents the estimated grid-side reactive power at time k+2.

[0022] Among them, the instantaneous power formula is: , In the formula, is the sampling frequency, is the grid-side electrical angular velocity, represents inductance, and They are respectively represented by grid-side voltage, grid-side current, inductive reactance and inductive resistance, where: represents the α component of the grid-side voltage at time k+1, represents the β component of the grid-side voltage at time k+1, represents the α component of the grid-side voltage at time k+1, Represents the β component of the grid-side voltage at time k+1.

[0023] S2. Select the voltage vector from the cost function. The specific method is as follows: The terms in the cost function are adjusted to zero to track the reference variable, so it can be assumed that =0 and =0, if the voltage vector middle , then in > The voltage vector will be was selected; If the voltage vector middle , then the voltage vector Select the one that produces less than of The voltage vector of exist Reaching steady state hour, , select The voltage vector.

[0024] As a preferred method, the power expression is corrected to compensate for the steady-state error based on the disturbance in the corrected power expression, and the specific method is as follows: S1-1. The mathematical model of the Kalman filter is as follows: (1), (2), In the formula, represents the active power disturbance, represents the reactive power disturbance, represents the active power disturbance at time k-1, represents the reactive power disturbance at time k-1, represents the active power at time k, is the reactive power at time k, represents the active power at time k-1, represents the reactive power at time k-1, and are the state vectors at time k and time k-1 respectively, is the sampling period.

[0025] The above is the state matrix, , , , , , , It is a manually defined matrix symbol with no specific meaning. , , = , , , ,in, , , both are commonly used mathematical matrices; here Indicates the resistance in the grid-side converter The nominal value of represents the inductance in the grid-side converter The nominal value of is the sampling frequency, is the grid-side electrical angular velocity.

[0026] It should be noted that in the above expression (1), Meaning , Meaning , similarly, expressions (1) and (2) express meaning in this way.

[0027] Further speaking, represents the inductance in the grid-side converter The nominal value of is the covariance matrix Process noise: ,in, Is a mathematical symbol representing a random variable Following the mean of 0, the covariance matrix is The normal distribution of .

[0028] S1-2, based on the mathematical model of step S3-1, the prior estimate of the state vector is as follows: (3), In the formula, Represents the state vector A priori estimate of Represents the interval [ ]The state vector calculated within The posterior estimate of ; S1-3. Define the precision covariance of the prior estimate as follows, (4), where E stands for mathematical expectation and T stands for matrix transpose, both of which are mathematical symbols.

[0029] The precision covariance is calculated as follows, , where is a matrix The transpose operation of ; Indicates that in the interval [ ] is the precision covariance of the posterior estimate of the state vector obtained in .

[0030] S1-4, Kalman gain and the posterior estimate of the state vector The calculation formulas are as follows: , , The corrected power prediction expression is: ,in is updated for the next sampling period; is the identity matrix, which is a square matrix with all elements on the diagonal from the upper left corner to the upper right corner being 1 and all other elements being 0.

[0031] It should be noted that if Figure 2-3 As shown, the setting and / or acquisition of other parameters of the present invention except for the aforementioned improvement of the present invention are all prior art, that is, they can be obtained by using the existing PI-MPC control method used in the permanent magnet synchronous motor drive system, for example, in the above step S4 and The acquisition of is the motor rotor angle, obtained by the position sensor; is the motor rotor speed, obtained by the speed sensor; is the DC bus voltage value at time k, which is obtained by sampling the voltage sensor. In addition, represents the estimated motor side switch state at time k, Represents the estimated motor side switch state at time k. In this embodiment, the voltage vector corresponding to the cost function is obtained according to step S2 to output the motor side switch state and the grid side switch state of the optimal solution.

[0032] Further, in step S2, the regulation process is described as follows using the PMSG power generation method as an example, wherein the negative PMSG active power is The PMSG grid-side active power is .

[0033] In the above cost function, It helps to select the voltage vector. The selected voltage vector will be The DC bus voltage is instantaneously generated closer to its set point, so the following conditions exist.

[0034] 1. If the voltage vector middle , then in The voltage vector when the active power generated at the moment is greater than the PMSG power generation power will be was selected, that is, in > The voltage vector will be is selected, which means that not all the power generated by the PMSG is transferred to the grid to charge the DC link capacitors and increase the DC link voltage.

[0035] 2. If the voltage vector middle , then the voltage vector The tendency to choose will produce less than of In this case, not only all the power generated by the PMSG is sent to the grid, but also the DC link capacitors will be discharged to contribute their power to the grid, thereby reducing the DC link voltage.

[0036] 3. In Reaching steady state When , all the power generated by the PMSG is delivered to the grid, and ideally the DC link capacitor will not be charged or discharged, i.e. , select The voltage vector.

[0037] in, Since the motor-side converter actually generates the PMSG power Power is also consumed when it is transmitted to the grid (conduction loss, switching loss, etc.), which means that the actual power that the motor-side converter can transmit to the grid side is less than If you choose to use = The voltage vector of the PMSG is that not only all the power from the PMSG will be delivered to the grid, but also the DC link capacitor will be discharged to contribute its power to the grid, which makes the regulated DC bus voltage lower than its set value. Therefore, in order to eliminate the steady-state voltage error, Add to For example, if the required reference power is 1000W, the actual reference power input to the system is 1000W+ .

[0038] The above description is only a preferred embodiment of the present embodiment, and all equivalent changes and modifications made within the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A direct voltage prediction control method for a permanent magnet synchronous motor drive system, the permanent magnet synchronous motor drive system comprising a motor-side converter and a grid-side converter connected to the grid, characterized in that: The control method comprises the following steps: S1. Establish the cost function as follows, , where is the motor side cost function, , and They are the active component of the cost function, the bus voltage component of the cost function and the reactive component of the cost function; Indicates the reactive power reference on the grid side; is the DC link voltage set point; is the DC bus voltage at time k+2, which is based on time The expression of The discrete model of time moves forward by one get; Adjustment coefficient, represents the set of positive real numbers; is the weighting factor used to adjust reactive power; represents the active power constraint, represents the constraint on reactive power; is the active power generated by the permanent magnet synchronous motor, = , is the torque reference, is the angular velocity of the permanent magnet synchronous motor, Calculated in the position estimator by the feedback encoder signal on the motor shaft from the permanent magnet synchronous motor; The active power value at time k+2 is obtained by moving the instantaneous power formula forward by one The obtained expression predicts that; represents the estimated grid-side reactive power at time k+2; S2. Selecting a voltage vector from the cost function, the specific method is as follows: Adjust the term in the cost function to approach zero to track the reference variable, assuming =0 and =0, if the voltage vector middle , then in > The voltage vector will be was selected; If the voltage vector middle , then the voltage vector Select the one that produces less than of The voltage vector of exist Reaching steady state hour, , select The voltage vector.

2. The direct voltage prediction control method for a permanent magnet synchronous motor drive system according to claim 1, characterized in that: Based on the disturbance compensation steady-state error in the correction power expression, the specific method is: S1-1. The mathematical model of the Kalman filter is as follows: （1）， （2）， In the formula, represents the active power disturbance, represents the reactive power disturbance, represents the active power disturbance at time k-1, represents the reactive power disturbance at time k-1, represents the active power at time k, is the reactive power at time k, represents the active power at time k-1, represents the reactive power at time k-1, and are the state vectors at time k and time k-1 respectively, is the sampling period; is the state matrix, , , , , , , is an artificially defined matrix symbol; where , , = , , , ; , , both are commonly used mathematical matrices; Represents the resistance in the grid-side converter The nominal value of represents the inductance in the grid-side converter The nominal value of is the sampling frequency, is the grid-side electrical angular velocity; is the covariance matrix Process noise: ,in Is a mathematical symbol representing a random variable Following the mean of 0, the covariance matrix is Normal distribution of S1-2: Based on the mathematical model of step 1-1, the prior estimate of the state vector is as follows, （3）， In the formula, Represents the state vector A priori estimate of Represents the interval [ ]The state vector calculated within The posterior estimate of ; S1-3: Define the precision covariance of the prior estimate as follows, (4), where E stands for mathematical expectation and T stands for matrix transpose, both of which are mathematical symbols; The precision covariance is calculated as follows, , where ; Indicates that in the interval [ ] is the precision covariance of the posterior estimate of the state vector obtained in; S1-4: Kalman gain and the posterior estimate of the state vector The calculation formulas are as follows: ， ， The corrected power prediction expression is: ,in is updated for the next sampling period; is the identity matrix, which is a square matrix with all elements on the diagonal from the upper left corner to the upper right corner being 1 and all other elements being 0.

3. The direct voltage prediction control method for a permanent magnet synchronous motor drive system according to claim 1, characterized in that: In step S2, Reaching steady state When Add to middle.

4. The direct voltage prediction control method for a permanent magnet synchronous motor drive system according to claim 1, characterized in that: In step S1, the instantaneous power is: , where is the sampling frequency, is the grid-side electrical angular velocity, represents inductance, and They are respectively represented by grid-side voltage, grid-side current, inductive reactance and inductive resistance, where: represents the α component of the grid-side voltage at time k+1, represents the β component of the grid-side voltage at time k+1, represents the α component of the grid-side voltage at time k+1, Represents the β component of the grid-side voltage at time k+1.

Images