CN109600061B - Novel fixed-frequency model prediction current control method based on dynamic weight - Google Patents

Abstract

The invention belongs to the field of power converter control, and relates to a dynamic variable weight re-type fixed-frequency model predictive current control method, which controls the system according to the equivalent switching circuit of the triple-change converter and the integrated control theory. The traditional fixed-frequency model predictive current control accurately calculates the voltage vector required at the next moment according to the system discrete mathematical model, so it has the characteristics of rapid response in the dynamic process, because it is affected by the double frequency of the DC bus voltage in the triple converter system. Influenced by fluctuations, it does not have good control performance in the steady-state process, and the proportional resonance (PR) control algorithm can realize the AC signal tracking without static error in the steady-state process, but the PR control algorithm has control in the dynamic process. Disadvantages of regulation hysteresis. Therefore, the present invention combines the advantages of the two control algorithms, takes the PR control algorithm as the compensation link of the model predictive current control, constructs a dynamic variable weight re-type fixed frequency model to predict the current control, and uses the d-axis current error value as the judgment basis. The Gaussian distribution function in the curve, the design time-varying weight function m. According to the change law of the m function, the two algorithms can be smoothly converted, thereby weakening the influence of the double frequency fluctuation in the DC side bus voltage on the performance of the control system, ensuring that the control system has good performance in both dynamic and steady-state processes.

Description

A Novel Fixed Frequency Model Predictive Current Control Method Based on Dynamic Weights

technical field

The invention belongs to the field of power converter control, and relates to a novel fixed-frequency model predictive current control method based on dynamic weights, which can be applied to the fields of motor speed regulation, renewable energy power generation and the like.

Background technique

With the improvement of the power level of the permanent magnet synchronous motor system, the topology structure composed of traditional power devices can no longer meet the requirements of high voltage and high power applications. In order to meet the actual engineering needs, domestic and foreign researchers have adopted the multiplexed converter technology to combine the traditional two-level six-switch topology to form a multiplexed converter topology. Due to the characteristics of the topology itself, when the system is running, there are double frequency fluctuations in the bus voltage of the multiplexed converter structure. Therefore, when controlling the system, if the traditional fixed frequency model is used to predict the current control, when the mathematical model of the system is accurate, it has the ability to respond quickly, but it cannot weaken the influence of the bus voltage fluctuation on the control performance in the steady state. Proportional resonant control (PR for short) can realize AC signal tracking without static error. Therefore, the PR control algorithm can weaken the influence of bus voltage fluctuation on the control performance. However, the PR control algorithm has the disadvantage of adjustment lag in the dynamic process.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the above-mentioned deficiencies of the prior art, and to provide a novel fixed-frequency model predictive current control method based on dynamic weights, which can realize that the control system has good control performance in both dynamic and steady-state processes, and is easy to implement. , the advantages of high reliability. A novel fixed-frequency model predictive current control method based on dynamic variable weights of the present invention includes the following steps:

Step 1, at time k, sample the signals required by the control system, which specifically include: three groups of DC bus voltages U _dc1 , U _dc2 , U _dc3 , grid side input three-phase voltages e _A , e _B , e _C and three-phase voltages Current i _A , i _B , i _C ;

Step 2: Set the q-axis given current to zero, and obtain the d-axis current given value ^{id ref} through the proportional integral controller according to the difference between the voltage outer loop given value and the feedback value; _id ^ref and _{id d} The expressions of ^ref are:

where i _d ^ref and i _q ^ref are the given current values of the dq axis respectively, K _p is the proportional coefficient of the voltage outer loop PI controller, K _i is the integral coefficient of the voltage outer loop PI controller, and U _eq ^ref is the voltage given value , U _eq is twice the average feedback voltage.

In step 3, the voltages e _A , e _B and e _C obtained by k sampling are subjected to phase-locked loop operation to obtain the position information θ, and the AC input voltage signal (obtained by sampling) and the current signal (obtained by sampling and calculation) are transformed through coordinate transformation. Theoretically, it is transformed into the voltage and current components under the α-β axis system; the expression of this step is as follows:

In the formula, e _α (k) and e _β (k) are the voltage components of the AC side input voltage at time k under the α-β axis system, and e _A (k), e _B (k), and e _C (k) are the time k. AC side input three-phase voltage;

In the formula, i _α (k) and i _β (k) are the current components of the AC side input current under the α-β axis system at time k, and i _A (k), i _B (k), and i _C (k) are the time k. AC side input three-phase current;

where i _α ^ref and i _β ^ref are the current components of the given current in the α-β axis system, respectively, id ^ref and i _q ^ref are the given current values of the _dq axis, respectively, and θ is the position information;

Step 4: Use the current and voltage sampling values at time k to calculate the model predictive control delay compensation current value according to the discrete mathematical model of the system; its expression is as follows:

In the formula, i _α (k+1) and i _β (k+1) are the current components of the AC side input current under the α-β axis system at the time of k+1, that is, the delay compensation current value. T _s is the system control period, L _gx is the equivalent AC side inductance, u _α (k-1), u _β (k-1) are the voltage of the output voltage vector at the time of k-1 under the α-β axis system, respectively weight;

Step 5: According to the delay compensation value obtained in step 4, replace the delay compensation value i _{α, β} (k+1) with the current value i _{α, β} (k) at time k, and substitute it into the system discretization mathematical model to solve α. Under the -β axis system, the model predictive control output voltage vector; its expression is as follows:

where u _α ^p (k) and u _β ^p (k) are the voltage components of the model predictive control output under the α-β axis system;

Step 6: According to the discrete mathematical model of the system, solve the output voltage vector of the PR controller under the α-β axis system; its expression is:

where u _α ^pr (k) and u _β ^pr (k) are the output voltage components of the PR controller under the α-β axis, and G _PR (s) is the transfer function of the PR controller, as shown in the following formula:

where K _p and K _r are the proportional coefficient and resonance coefficient of the PR controller, respectively, and ω _c and ω ₀ are the cut-off frequency and resonance frequency, respectively;

In step 7, the PR control is used as the compensation link of the model predictive current control, and the dynamic variable weight re-type fixed frequency model predictive current control is constructed. Taking the d-axis current error value as the judgment basis, according to the Gaussian distribution function in the bell curve, the time-varying weight function m is designed, and the maximum value of m is 1 to limit the amplitude; its expression is:

In the formula, u _α ^end (k), u _β ^end (k) are the final reference voltage vector, m is the time-varying weight function, and its expression is:

In the formula, Δi _d is the d-axis current error value, σ is the standard deviation, which is always greater than 0; assuming that the d-axis current difference fluctuates between ±0.2A when the control system is in a steady state, when the m function is parameterized, The following rules apply:

1) When the system is in a steady state process and there is no sampling error, the d-axis current difference fluctuates between ±0.2A. At this time, PR control is desired, so 0.2A is taken as a turning point. When the absolute value of the error is less than or equal to 0.2 When A, m is always 1.

2) When the system is in a steady state process and there is a sampling error, the absolute value of the current difference will be greater than 0.2A. At this time, it is hoped that m varies slowly between 0 and 1, and the two control algorithms work at the same time to achieve fast response and weakening. The purpose of the influence of bus voltage fluctuation on control performance.

3) When the system is in a dynamic process and there is no sampling error, it is assumed that the current difference in the dynamic process changes to 1A. At this time, it is hoped that the model predicts the current control to play a leading role, and m should decrease to 0 quickly.

Step 8: Divide the finally obtained reference voltage vectors u _α ^end (k) and u _β ^end (k) by 2 according to the position information θ obtained at time k and the three groups of bus voltages, respectively, which are carried out by three SVPWM calculation units. In the duty cycle calculation, when the SVPWM calculation unit calculates, the carrier phases of the three SVPWM calculation units are different, and the carrier phases of the second and third SVPWM calculation units lag behind the calculation unit - 1/3 control cycle at the same time, so as to perform PWM Calculate the duty cycle, recombine the obtained PWM signal, update the duty cycle at the time k+1, and repeat steps 1 to 7 at the time k+1, so as to cycle.

Compared with the prior art, the effect that the present invention has is:

(1) The present invention adopts model predictive current control in the dynamic process, which has the advantage of rapid dynamic response compared with PR control.

(2) Affected by the double frequency fluctuation of the DC bus voltage of the triple system, the model predicted current control cannot obtain good steady-state performance. The present invention uses PR control to compensate the model predicted current control, which can weaken the effect of the double frequency fluctuation of the bus voltage on the system. Compared with the traditional model, the current control has good steady-state performance.

Description of drawings

Figure 1: Topological diagram of line voltage cascade triple converter;

Figure 2: Equivalent circuit of line voltage cascade triple converter;

Figure 3: System Control Structure Diagram

Detailed ways

When the triple converter is used as a grid-side rectifier circuit, its topology is shown in Figure 1. Ignoring the influence of the input line resistance on the AC side on the system, and generalizing the loop formed by unit 1 and unit 2 in Figure 1, the following relationship can be obtained:

In the formula, U _AB , U _BC , and U _CA are the line voltages of the triple LVC-VSC AC side, E _AB , E _BC , and E _CA are the input line voltages of the power grid, and I _ai , I _bi , and I _ci are the respective phases in the constituent unit i. The phase currents U _aibi , U _bici , and U _ciai are the fundamental wave components of the phase-to-phase line-to-phase voltages on the AC side ab, bc, and ca in the constituent unit i, respectively, where i=1, 2, and 3. U _b1a2 , U _c2b3 , and U _a1c3 are respectively the voltages connecting the current-limiting inductors between the units.

According to the above analysis, the triple converter can be equivalent to a traditional two-level power converter, as shown in Figure 2. In the figure, L _gx is the equivalent AC side inductance, and its expression is:

In the formula, k _L is the ratio between L _g and L _x , namely:

Since the AC side line voltage of the equivalent switching circuit is equal to the sum of the line voltages of two adjacent cascaded VSC units, the equivalent DC bus voltage U _eq is equal to twice U _av , and U _av represents the average of the three groups of DC bus voltages value. According to Figure 2, the mathematical model expression of the equivalent switching circuit of the triple converter under the α-β axis system can be obtained as:

In the formula, e _α , e _β , i _α , and i _β are the grid voltage and grid current under the α-β axis system, respectively. The acquisition method is as described in step 3, and can be expressed as:

In the formula, e _α (k) and e _β (k) are the voltage components of the AC side input voltage at time k under the α-β axis system, and e _A (k), e _B (k), and e _C (k) are the time k. AC side input three-phase voltage.

In the formula, i _α (k) and i _β (k) are the current components of the AC side input current under the α-β axis system at time k, and i _A (k), i _B (k), and i _C (k) are the time k. Three-phase current is input on the AC side.

In step 2, the given current value of the d-axis of the control system is calculated by the voltage outer loop PI controller, which can be expressed as:

In the formula, i _d ^ref and i _q ^ref are the given current values of the dq axis on the grid side respectively, K _p is the proportional coefficient of the voltage outer loop PI controller, K _i is the integral coefficient of the voltage outer loop PI controller, and U _eq ^ref is the voltage supply A fixed value, U _eq is twice the average value of the feedback voltage. The coordinate transformation of formula (7) as described in step 3 can be obtained:

In the formula, i _α ^ref and i _β ^ref are the current components of the given current in the α-β axis system, respectively, id ^ref and i _q ^ref are the given current values of the _dq axis, respectively, and θ is the position information.

In the fourth step, the model predictive control delay compensation is calculated, that is, i _{α, β} (k+1) is solved according to the formula (4), and the formula (4) is discretized and solved. You can get:

In the formula, i _α (k+1) and i _β (k+1) are the current components of the AC side input current under the α-β axis system at the time of k+1, that is, the delay compensation current value. T _s is the system control period, and u _α (k-1) and u _β (k-1) are the voltage components of the given voltage vector at the time of k-1 under the α-β axis system, respectively.

In step 5, the model predictive control output voltage vectors u _α ^p and u _β ^p are calculated according to the delay compensation current value obtained in step 4, which can be expressed as:

In step 6, the output voltage vectors u _α ^pr (k) and u _β ^pr (k) of the PR control compensation link are calculated, which can be expressed as:

where G _PR (s) is the transfer function of PR control, as shown in the following formula:

In the formula, K _p and K _r are the proportional coefficient and resonance coefficient of the PR controller, respectively, and ω _c and ω ₀ are the cut-off frequency and the resonance frequency, respectively.

In step 7, a dynamic variable weight remodeling fixed frequency model is constructed to predict the current control, which can be expressed as:

where u _α ^end (k) and u _β ^end (k) are the final reference voltage vector, m is the time-varying weight function, and its expression is:

where Δid is the _d -axis current error value, and σ is the standard deviation, which is always greater than 0. At the same time, a suitable σ can be selected to construct the function m according to the requirements of the actual operating conditions. Since different σ may make m greater than 1, it is necessary to limit the maximum value of m to 1.

Step 8: According to the structural characteristics of the triple converter and the existing carrier phase-shift modulation strategy, when controlling the system, three SVPWM calculation units are required, and each SVPWM calculation unit adopts 2U _av /3 as the space vector coordinate. The modulo length of the coordinate axis in the system, where U _av is the average value of the three groups of capacitor voltages on the DC side. Therefore, the reference voltage vectors u _α ^end (k) and u _β ^end (k) obtained at time k need to be multiplied by 1/2 as the final reference voltage vector of each SVPWM calculation unit. When the SVPWM calculation unit calculates, the three SVPWM The carrier phases of the calculation units are different, and the carrier phases of the second and third SVPWM calculation units lag behind the calculation unit by 1/3 control period at the same time, so as to calculate the PWM duty cycle, and recombine the obtained PWM signals. The duty cycle is updated at time k+1 and steps 1 to 7 are repeated, and the cycle is repeated. At this time, the system control block diagram is shown in Figure 3.

Claims (7)

1. A dynamic variable weight novel fixed frequency model prediction current control method is characterized by comprising the following steps:

step one, at the time k, a control system samples a required signal, and the method specifically comprises the following steps: three groups of direct current bus voltages and three-phase voltage e input by power grid side_A，B，CAnd three-phase current i_A，B，C；

Step two, enabling the q-axis given current to be zero, obtaining a d-axis current given value i through a proportional-integral controller according to a difference value between a voltage outer ring given value and a feedback value, wherein the feedback value is two times of an average value of three groups of direct-current bus voltage sampling values_d ^ref；

Step three, the voltage e obtained by sampling_A，B，CObtaining position information theta through phase-locked loop operation, and converting the alternating current input voltage signal and the current signal obtained by sampling in the step one into α - β shafting lower voltage and current components through a coordinate transformation theory;

step four, current and voltage sampling values at the moment k are utilized, wherein the current sampling value at the moment k is the current i of the three-phase power grid_A，B，CThe voltage sampling value at the moment k is the three-phase power grid voltage e_A，B，C(ii) a Calculating the delay compensation value of model prediction current control according to the discretization mathematical model of the system,i.e. the current value i at the moment k +1_α，β(k+1)；

Step five, delaying the compensation value i_α，β(k +1) instead of the current value i at time k_α，β(k) Substituting the voltage vector into a system discretization mathematical model, solving α - β shafting when adopting model prediction current control, and outputting a given voltage vector u by the model prediction current control_α ^p、u_β ^p；

Step six, according to the discretization mathematical model of the system, the output given voltage vector u is obtained by the PR controller when the PR control is adopted_α ^pr、u_β ^pr；

Step seven, constructing a time-varying function m according to a Gaussian distribution function in a bell-shaped curve so as to realize smooth conversion between model prediction current control and PR control, wherein the value of m is between 0 and 1;

step eight, according to the position information theta obtained at the moment k and the three groups of bus voltages, obtaining a finally obtained reference voltage vector u_α ^end(k)、u_β ^end(k) Dividing by 2, performing PWM wave duty ratio calculation by three SVPWM calculation units respectively, updating the duty ratio at the moment of k +1, and repeating the steps from one step to seven at the moment of k +1, thereby performing circulation.

2. The method for controlling the predictive current of the dynamic variable-weight novel fixed-frequency model according to claim 1, wherein the calculation formula in the second step is as follows:

in the formula i_d ^ref、i_q ^refRespectively giving current values, K, to d-q axes on the power grid side_pIs the proportional coefficient, K, of a voltage outer loop PI controller_iFor voltage outer loop PI controller integral coefficient, U_eq ^refIs given value of voltage, U_eqIs twice the average value of the feedback voltage.

3. The method for controlling the predictive current of the dynamic variable weight novel fixed frequency model according to claim 1, wherein the calculation formulas of the coordinate transformation of each signal in the three steps are respectively as follows:

in the formula e_α、e_βIs the voltage component of the grid voltage in the α - β axis system, e_A、e_B、e_CThe three-phase voltages of the power grid are respectively;

in the formula i_α、i_βInputting current components i of three-phase current in α - β shafting for the side of the power grid_A、i_B、i_CInputting three-phase current for the power grid respectively;

in the formula i_α ^ref、i_β ^refRespectively, the current components of the given current in the α - β shafting.

4. The method for controlling the predicted current of the novel dynamic variable weight constant frequency model according to claim 1, wherein the delay compensation value in the fourth step is calculated according to the following formula:

in the formula i_α，β(k +1) is the current value of the current at the moment of k +1 in the axis system of α - β, namely the delay compensation value u_α，β(k-1) is the component of a given voltage vector at the time of k-1 in the axis line of α - β, L_gxIs equivalent AC side inductance, T_sA system control cycle.

5. The method for predicting the current control by the dynamic variable weight novel fixed frequency model as claimed in claim 1, wherein the model prediction current control output voltage vector in the fifth step can be calculated according to the following formula:

in the formula u_α ^p(k)，u_β ^p(k) Predicting the current control output for the model, T_sA system control cycle.

6. The method as claimed in claim 1, wherein the output of the PR controller in step six is calculated according to the following formula:

in the formula u_α ^pr(k)，u_β ^pr(k) Is the output of the PR controller, G_PR(s) is a PR controller.

7. The method according to claim 1, wherein the novel fixed-frequency model predictive current control in step seven is calculated according to the following formula:

in the formula u_α ^end(k)、u_β ^end(k) For the final reference voltage vector, m is a time-varying duty ratio function, and the expression is as follows:

in the formula,. DELTA.i_dIs d-axis current error value, sigma is standard deviation, constantGreater than 0; meanwhile, according to the requirements of actual operating conditions, a proper sigma can be selected to construct the function m, and because different sigma can enable m to be larger than 1, amplitude limitation needs to be carried out on the maximum value of m to be 1.

Images