PWM converter sensorless control method based on model reference self-adaption

Abstract

A PWM converter sensorless control method based on model reference self-adaption belongs to the technical field of power electronic converter control, inductance parameter predictive control is added in the model predictive control of a PWM converter, even if the value of the grid-side input inductance is inaccurate or fluctuates due to the influence of parameters, accurate real-time inductance can be obtained, and the accuracy of the grid-side input inductance is improved. Therefore, the accurate direct-current side voltage is obtained, and the output effect of the post-stage is ensured. Therefore, the method has robustness for the power grid change and the parameter uncertainty of the inductor at the converter side, meanwhile, the number of grid-side voltage sensors required in converter control can be reduced, and the system cost is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of power electronic converter control, and in particular relates to a sensorless control method for a PWM converter based on model reference adaptation. Background Art

[0002] With the gradual development of my country's economy, the demand for power electronic converters has gradually increased. PWM converters have many unique advantages such as bidirectional power flow, high-quality output DC voltage and low harmonic distortion input current, and have been widely used in various industrial occasions such as active power filters and STATCOM. PWM converters with filter inductors are the most common form of PWM converters at present. However, since the grid impedance and the inductance value of the previous transformer are usually unknown, changes in the grid-side parameters will cause the output accuracy of the control algorithm to decrease, and in severe cases may even cause system instability. Therefore, it is of great research significance to estimate the grid-side inductance parameter value.

[0003] Due to many technical and economic advantages, sensorless control has been studied by a large number of scholars. On the one hand, the sensor signal will be affected by noise, resulting in inaccurate signals; on the other hand, the extensive use of high-precision sensors will increase the system cost and the size of the device. Therefore, it is an ideal solution to reduce the number of sensors through control algorithms. Traditional sensorless control is often used in the field of AC motor drive applications, and is less studied in the field of power electronics. At present, the research on sensorless control at home and abroad mainly includes state observer method, sliding mode variable structure method, Kalman filter method and model reference adaptive method. Among them, the state observer has been widely used in motor control as a sensorless control motor back electromotive force estimation. Based on the state equation, the feedback matrix is ​​designed by the pole configuration method, and the adaptive mechanism is designed in combination with the Lyapunov stability theory. The speed estimation and flux linkage observation effects are achieved at the same time, but the observer bandwidth limitation will cause hysteresis. The sliding mode variable structure method uses the estimated deviation to determine the sliding mode control mechanism, and finally stabilizes the state of the control system on the designed sliding plane. Sliding mode control has good dynamic response and robustness, but the disadvantage of sliding mode variable structure method is that the estimated variables contain high-order harmonics. Although filtering can be performed, filtering usually causes phase offset. Kalman filter is an optimal linear estimation algorithm proposed in the early 1960s. Its characteristics are that it takes into account the statistical characteristics of the system model error and measurement noise. The algorithm of Kalman filter adopts recursive form and is suitable for implementation on a computer. The key to Kalman filtering is to select coefficient values ​​to obtain the best possible position estimation performance. However, the Kalman filtering method has high computational intensity, and it is difficult for the filter to determine the noise level of the actual system and the Kalman gain in the algorithm. The model reference adaptive control (MRAC) method has the advantages of simple structure and low computational complexity, and has become a commonly used sensorless control method. Traditional MRAC control often uses PI controller for parameter estimation, but the parameter adjustment of PI controller is relatively difficult, and as the system operating conditions change, it is difficult for PI parameters to be in the best state at any time.

[0004] Normally, the grid-side inductance and step-down transformer leakage inductance are unknown or inaccurate parameters, and the conventional control method using PWM converter basically requires three sensors to complete, including input AC voltage, input AC current and output DC side voltage. When the grid-side inductance parameters are not obtained accurately, the control accuracy will be seriously affected. However, the above-mentioned sensorless control is often used in the field of AC motor drive applications, and is less studied in the field of power electronics. Summary of the invention

[0005] In order to solve the above-mentioned problems, the present invention proposes: a sensorless control method for a PWM converter based on model reference adaptation, which adds inductance parameter predictive control to the traditional model predictive control of the PWM converter. Even if the value of the grid-side input inductance is inaccurate or fluctuates due to parameter influence, accurate real-time inductance can be obtained, and then accurate DC-side voltage can be obtained to ensure the output effect of the subsequent stage, thereby making it robust to grid changes and uncertainty in the inductor parameters on the converter side, and at the same time reducing the required grid-side voltage sensors.

[0006] The present invention solves the above technical problems through the following technical solutions: a sensorless control method of a PWM converter based on model reference adaptation, comprising the following steps:

[0007] Step 1: Sampling the input current of the PWM converter at the kTth moment is i s (k), and uses the PWM converter voltage V at the kTth moment determined in the previous moment PWM (k) and V s (k);

[0008] Step 2: Calculate the seven possible input currents i at the (k+1)Tth moment under different switching states s (k+1);

[0009] Step 3: According to the AC line current i at time k s (k), the line voltage v can be inferred s (k)

[0010] Step 4: Use seven possible input currents to predict the next input active and reactive power P in (k+1) and Q in (k+1);

[0011] Step 5: Based on the cost function g for evaluating system performance, select the switch combination corresponding to the minimum value of g, and act on the PWM converter at the next moment.

[0012] The sensorless control method for a PWM converter based on model reference adaptation described in the present invention first constructs a cost function based on a control variable reference value and a control variable output prediction value, selects the optimal output level through a finite control set, continuously performs rolling optimization based on sampled values ​​to approach the command value, and selects the optimal voltage vector for the next sampling period as a control action based on a sorted minimum value through a cost function including a control variable error.

[0013] Furthermore, in step 2, the grid-side current i at time k+1 is s The calculation process of (k+1) is as follows:

[0014] Step 2.1: Calculate the grid-side voltage and current vector expressions:

[0015] v s =2 / 3(v sa +αv sb +α 2 v sc )

[0016] i s =2 / 3(i sa +αi sb +α 2 i sc )

[0017] Where α = e j2π / 3 , v sa 、v sb and v sc are the three-phase grid-side phase voltages, i sa 、i sb and i sc They are the phase currents on the three-phase grid side, L s is the grid-side filter inductor, R s It is the sum of the equivalent resistance and the equivalent resistance of the power switch tube loss, and C is the DC side filter capacitor.

[0018] Step 2.2: Calculate the voltage vector of the PWM converter as follows:

[0019] v afe =2 / 3(S1+αS2+α 2 S3)V dc

[0020] Among them, S1, S2, and S3 are the switch states of the upper tube of the PWM converter, 1 means closed, and 0 means open. According to KVL, the following formula can be obtained in the stationary coordinate system αβ, where L s and R s are the grid-side inductance and equivalent resistance, V dc is the DC side voltage.

[0021] Step 2.3: The expression after discretizing the grid-side current using the forward Euler formula is as follows:

[0022]

[0023] Among them, T s is the sampling period of the control system.

[0024] Furthermore, in step 3, the line voltage v s (k) The calculation process is as follows:

[0025] Step 3.1: According to the instantaneous power theory of three-phase circuit, predict the instantaneous active power and reactive power P at the next moment in (k+1) and Q in (k+1) is expressed in the αβ coordinate system as:

[0026]

[0027] in, for i s (k+1) conjugate current. In addition, since the sampling frequency is much higher than the fundamental frequency, the voltages of the two samples can be approximately equal, that is, v s (k+1)≈v s (k).

[0028] Step 3.2: Use the predicted values ​​P of active power and reactive power at time k+1 in (k+1) and Q in (k+1), and the measured AC line current i at time k s (k), the line voltage v can be inferred s (k) is as follows:

[0029]

[0030] Furthermore, in step 4, the specific expression of the cost function g is:

[0031]

[0032] Among them, P * in is the active power command, Q * in Reactive power command.

[0033] Furthermore, in step 2, the estimated inductance value can be obtained by the following model reference adaptive control method:

[0034] The estimated current It can be expressed as:

[0035]

[0036] The “^” indicates an estimated value. And u=v s -v pwm , then the error expression can be obtained as:

[0037]

[0038] It can be deduced that:

[0039]

[0040] Furthermore, in step 2, the Lyapunov function is selected as:

[0041]

[0042] but

[0043]

[0044] satisfy Then MRAC is guaranteed to hold. Where V is a positive number, it only needs to satisfy And the first item - 2MR s e 2 <0 holds true, then let the following equation hold true

[0045]

[0046] So we take the adaptive law as

[0047]

[0048] Where λ represents the adaptive coefficient. As long as the above adaptive law is satisfied, the designed MRAC system can be guaranteed to converge.

[0049] The beneficial effects of the present invention are as follows: compared with the prior art, the present invention proposes a sensorless control method for a PWM converter based on model reference adaptation, designs a model reference adaptive control according to the Lyapunov stability theorem to estimate the grid-side inductance value in real time, adds inductance parameter predictive control to the traditional model predictive control of the PWM converter, obtains the grid-side voltage value by predicting the inductance value, reduces the required grid-side voltage sensor, and obtains accurate real-time inductance even if the value of the input inductance is inaccurate or fluctuates due to the influence of parameters, thereby obtaining an accurate DC side voltage to ensure the output effect of the subsequent stage and increase the accuracy and robustness of system control. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a topological structure diagram of a PWM converter in an embodiment of the present invention;

[0051] Figure 2 7 output level state prediction control rolling optimization algorithm principle diagram of the PWM converter in the embodiment of the present invention;

[0052] Figure 3 is a control block diagram of a sensorless control method for a PWM converter according to an embodiment of the present invention;

[0053] Figure 4 is a block diagram of a model reference adaptive real-time inductance estimation process in an embodiment of the present invention;

[0054] Figure 5 1 is a simulation waveform diagram of the inductance value sudden change tracking effect test in an embodiment of the present invention; wherein, (a) actual and estimated grid-side a-phase current waveform; (b) actual and estimated grid-side a-phase error current waveform; (c) grid-side voltage and grid-side current waveform of the AEF converter; (d) actual and estimated grid-side inductance value;

[0055] Figure 6 : It is a simulation waveform diagram for comparing the MRAC control effect in the embodiment of the present invention; wherein, (a) the actual value and command value waveform of the DC side voltage when the inductance changes from 3mH to 1.5mH without MRAC control; (b) the actual value and command value waveform of the DC side voltage when the inductance changes from 3mH to 1.5mH with MRAC control; (c) the actual value and command value waveform of the DC side voltage when the inductance changes from 1.5mH to 3mH without MRAC control; (d) the actual value and command value waveform of the DC side voltage when the inductance changes from 1.5mH to 3mH with MRAC control. DETAILED DESCRIPTION

[0056] In order to make the technical means and objectives of the present invention easy to understand, the present invention is further described below in combination with specific implementation methods. A sensorless control method for a PWM converter based on model reference adaptation is as follows: Figure 1 As shown, the following steps are included:

[0057] 1. The input current of the PWM converter at the kTth moment is(k), and the PWM converter voltage VPWM(k) and Vs(k) at the kTth moment determined in the previous moment are used to calculate the seven possible input currents is(k+1) at the (k+1)th moment under different switching states according to the mathematical model of the PWM converter;

[0058] like Figure 1 As shown in the figure, the PWM converter consists of 6 switching tubes forming 3 bridge arms. Only one of the upper and lower power devices of a single bridge arm can be turned on. When the upper tube is turned on and the lower tube is turned off, the bridge arm state is 1, and when the lower tube is turned on and the upper tube is turned off, the bridge arm state is 0. The PWM converter has 7 different switch state combinations, namely V0 (0, 0, 0), V1 (1, 1, 1), V2 (1, 1, 0), V3 (0, 1, 0), V4 (0, 1, 1), V5 (1, 0, 0), and V6 (1, 0, 1). Under different switch state combinations, the output level of the converter is different.

[0059] Calculate the grid-side voltage and current vector expression:

[0060] v s =2 / 3(v sa +αvsb +α 2 v sc )

[0061] i s =2 / 3(i sa +αi sb +α 2 i sc )

[0062] Where α = e j2π / 3 , vsa, vsb and vsc are the three-phase grid-side phase voltages respectively, isa, isb and isc are the three-phase grid-side phase currents respectively, Ls is the grid-side filter inductor, Rs is the sum of the equivalent resistance and the equivalent resistance of the power switch tube loss, and C is the DC side filter capacitor.

[0063] The calculation process of the PWM converter voltage vector is as follows:

[0064] v afe =2 / 3(S1+αS2+α 2 S3)V dc

[0065] Among them, S1, S2, and S3 are the switch states of the upper tube of the PWM converter, 1 means closed, and 0 means open. According to KVL, the following formula can be obtained in the stationary coordinate system αβ, where Ls and Rs are the grid-side inductance and equivalent resistance, respectively, and Vdc is the DC side voltage.

[0066] The expression after discretizing the grid-side current using the forward Euler formula is as follows:

[0067]

[0068] Where Ts is the sampling period of the control system.

[0069] According to the instantaneous power theory of three-phase circuits, the predicted instantaneous active power and reactive power Pin(k+1) and Qin(k+1) at the next moment are expressed in the αβ coordinate system as follows:

[0070]

[0071] in, for i s (k+1) conjugate current. In addition, since the sampling frequency is much higher than the fundamental frequency, the voltages of the two samples can be approximately equal, that is, v s (k+1)≈v s (k).

[0072] The predicted values ​​of active power and reactive power at time k+1, Pin(k+1) and Qin(k+1), are compared with the measured AC line current i at time k.s (k), the line voltage v can be inferred s (k) is as follows:

[0073]

[0074] 2. If Figure 2 As shown, P* represents the reference value of active power, Pp represents the predicted value of active power, and P represents the active power value output after performance index optimization. The vector with the smallest cost function among the seven vectors is selected as the control state at the next moment, and rolling optimization can be used to find the best vector for tracking the target instruction.

[0075] The specific expression of the cost function g is:

[0076]

[0077] Among them, P*in is the active power command, and Q*in is the reactive power command. The cost function is directly obtained based on the control variable reference value and the control variable output prediction value, ensuring that the PWM converter can effectively track the output power command, such as Figure 3 shown.

[0078] 3. In order to improve the robustness of the system, a parameter estimation method based on model reference adaptive control (MRAC) is added to the predictive control system. Lyapunov's second method is also called the direct method. Its basic idea is to use a Lyapunov function to directly judge the stability of the system's equilibrium state. The MRAC system based on the Lyapunov function adjusts the gain of the adjustable system to achieve the generalized error signal obtained by subtracting the actual output of the system from the reference model output and tends to zero, and can ensure the stability of the closed-loop control system, so it is also called adjustable gain Lyapunov-MRAC.

[0079] Assume that the state equation of the system is The equilibrium state is xe=0, satisfying f(xe)=0.

[0080] If there exists a scalar function V(x) such that:

[0081] 1) V(x) has continuous first-order partial derivatives for all x.

[0082] 2) V(x) is positive definite, that is, when x=0, V(x)=0; x≠0, V(x)>0.

[0083] 3) The time derivative of V(x) calculated along the state trajectory satisfies If V(x) is negative definite, then the equilibrium state at the origin is asymptotically stable.

[0084] like Figure 4 As shown, the estimated current It can be expressed as:

[0085]

[0086] The “^” indicates an estimated value. And u = vs-vpwm, then the error expression can be obtained as:

[0087]

[0088] It can be deduced that:

[0089]

[0090] Design the Lyapunov function as:

[0091]

[0092] but

[0093]

[0094] satisfy Then MRAC is guaranteed to hold. Where V is a positive number, it only needs to satisfy And the first item - 2MR s e 2 <0 holds true, then let the following equation hold true

[0095]

[0096] So we take the adaptive law as

[0097]

[0098] Where λ represents the adaptive coefficient. As long as the above adaptive law is satisfied, the designed MRAC system can be guaranteed to converge.

[0099] Verify the prediction effect of the proposed control method on the grid-side inductance value in MATLAB / Simulink and simulate the inductance fluctuation. Set the inductance Ls from 3mH to 1.5mH at 0.5s and observe the output effect. Figure 5 The simulation waveform of the inductance value mutation tracking effect test is shown in the figure. As can be seen from the waveform of the inductance prediction value Ls^ and the actual value Ls, the proposed control method can track the actual value of the inductance within 0.1s, which effectively improves the robustness of the system.

[0100] In order to verify the effect of the increase in the estimated control of the grid-side inductance value, the inductance Ls is changed from 3mH to 1.5mH and from 1.5mH to 3mH at 1s. The simulation results are as follows: Figure 6 As shown in the figure, if the inductance changes without control, the change of grid-side inductance will cause obvious fluctuations in the DC side voltage. After adding control, the fluctuation amplitude of the output DC side voltage dropped from 5V to 2V and from 5.5V to 3V, respectively, and the fluctuation situation was significantly improved, which can prove that the proposed control method effectively improves the accuracy and stability of the system output.

[0101] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A sensorless control method for a PWM converter based on model reference adaptation, characterized in that: The following steps are involved: Step 1: Sampling the input current of the PWM converter at the kTth moment is i s (k), and uses the PWM converter voltage V at the kTth moment determined in the previous moment PWM (k) and V s (k); Step 2: Calculate the seven possible input currents i at the (k+1)Tth moment under different switching states s (k+1); Step 3: According to the AC line current i at time k s (k), inversely deduce the line voltage v s (k); Step 4: Use seven possible input currents to predict the next input active and reactive power P in (k+1) and Q in (k+1); Step 5: Based on the cost function g for evaluating system performance, select the switch combination corresponding to the minimum value of g, and act on the PWM converter at the next moment.

2. The sensorless control method for PWM converter based on model reference adaptation as claimed in claim 1, characterized in that: In step 2, the grid-side current i at time k+1 s The calculation process of (k+1) is as follows: Step 2.1: Calculate the grid-side voltage and current vector expressions: v s =2 / 3(v sa +αv sb +α 2 v sc ) i s =2 / 3(i sa +αi sb +α 2 i sc ) Where α = e j2π / 3 , v sa 、v sb and v sc are the three-phase grid-side phase voltages, i sa 、i sb and i sc They are the phase currents on the three-phase grid side, L s is the grid-side filter inductor, R s is the sum of the equivalent resistance and the equivalent resistance of the power switch tube loss, and C is the DC side filter capacitor; Step 2.2: Calculate the voltage vector of the PWM converter as follows: v afe =2 / 3(S1+αS2+α 2 S3)V dc Where S1, S2, and S3 are the switch states of the upper tube of the PWM converter, 1 means closed, and 0 means open; according to KVL, the following formula is valid in the stationary coordinate system αβ, where L s and R s are the grid-side inductance and equivalent resistance, V dc is the DC side voltage; Step 2.3: The expression after discretizing the grid-side current using the forward Euler formula is as follows: Among them, T s is the sampling period of the control system.

3. The sensorless control method for PWM converter based on model reference adaptation as claimed in claim 1, characterized in that: In step 3, the line voltage v s (k) The calculation process is as follows: Step 3.1: According to the instantaneous power theory of three-phase circuit, predict the instantaneous active power and reactive power P at the next moment in (k+1) and Q in (k+1) is expressed in the αβ coordinate system as: in, for i s (k+1) conjugate current. In addition, since the sampling frequency is much higher than the fundamental frequency, the voltages of the two samples are approximately equal, that is, v s (k+1)≈v s (k); Step 3.2: Use the predicted values ​​P of active power and reactive power at time k+1 in (k+1) and Q in (k+1), and the measured AC line current i at time k s (k), inversely deduce the line voltage v s (k) is as follows:

4. The sensorless control method for PWM converter based on model reference adaptation as claimed in claim 1, characterized in that: In step 4, the specific expression of the cost function g is: Among them, P * in is the active power command, Q * in Reactive power command.

5. The sensorless control method for PWM converter based on model reference adaptation as claimed in claim 1, characterized in that: In step 2, the estimated inductance value is obtained by the following model reference adaptive control method: 1) Estimated current It can be expressed as: The "^" indicates an estimated value. And u=v s -v pwm 2) The error expression is: It can be deduced that: 3) Select the Lyapunov function as: but satisfy Then MRAC can be guaranteed to hold; where V is a positive number, it only needs to satisfy And the first item - 2MR s e 2 <0 holds true, then let the following equation hold true So we take the adaptive law as Where λ represents the adaptive coefficient. As long as the above adaptive law is satisfied, the designed MRAC system can be guaranteed to converge.

Images