Model prediction optimization control method for photovoltaic grid-connected inverter

Abstract

The invention provides a model prediction optimization control method for a photovoltaic grid-connected inverter, and relates to the technical field of new energy. The model prediction optimization control method of the photovoltaic grid-connected inverter specifically comprises the following steps of S1, mathematical modeling of the photovoltaic grid-connected inverter, S2, establishment of a prediction model and a value function, S3, a three-vector model current prediction optimization method, and S4, compensation of time delay through a two-step prediction method. By establishing a mathematical model of the photovoltaic grid-connected inverter, predicting the current of the grid-connected inverter and based on a control target, optimized three-vector model prediction control is provided, so that optimized control of the photovoltaic grid-connected inverter is achieved, the electric energy quality and the system stability are improved, the system cost is reduced, and the system efficiency is improved. While the control precision of the power grid current is improved, the harmonic content of the current can be effectively reduced, the neutral-point potential balance speed is accelerated, and the dynamic performance under various working conditions is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy, and specifically to a model predictive optimization control method for a photovoltaic grid-connected inverter. Background Art

[0002] The grid-connected inverter is the core equipment of the photovoltaic power generation system, which converts direct current into alternating current and incorporates it into the power grid. However, due to the nonlinear characteristics of the photovoltaic power generation system itself, it is difficult for traditional control methods to take into account various indicators such as the harmonic content of voltage and current, the balance of the midpoint potential on the DC side, and the dynamic performance under disturbance conditions. Therefore, in recent years, model predictive control, as an advanced non-linear control strategy, has been gradually applied to the optimization control of photovoltaic inverters.

[0003] Currently, model predictive control can be divided into single-vector and multi-vector model predictions according to the number of output vectors. Currently, the more widely used ones are single-vector and multi-vector controls. Traditional single-vector model predictive control has problems such as a fixed voltage vector direction, a fixed amplitude, and a small number of optimization times, resulting in a large pulsation of the grid-connected current. Although the dual-vector model predictive control can change the amplitude of the inverter output voltage, the output voltage direction is still fixed, the coverage range of the output voltage vector is limited, and the current pulsation is large, resulting in low current tracking accuracy and a high harmonic content of the grid-connected current. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a model predictive optimization control method for a photovoltaic grid-connected inverter, which improves the control accuracy of the grid current, reduces the harmonic content of the current, speeds up the midpoint potential balance speed, and improves the dynamic performance under various conditions.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A model predictive optimization control method for a photovoltaic grid-connected inverter specifically includes the following steps:

[0006] S1. Mathematical modeling of the photovoltaic grid-connected inverter

[0007] The photovoltaic inverter includes a photovoltaic panel, two DC-side capacitors, a T-type three-level topology structure, and an output terminal. The two DC-side capacitors are respectively marked as C1 and C2. The output side includes an LC filter and the grid voltage. Each arm of the T-type three-level inverter has three switching states, namely the P state, the O state, and the N state. The switching-on situation of each switching state corresponding to the switching tubes is as follows:

[0008]

[0009] Where S i is the switching state, and S ix represents the four switching tubes on each phase arm; therefore, the three-phase three-level inverter has 3 3= 27 working states, corresponding to 27 basic voltage vectors;

[0010] Assume that all power switches, all capacitors, and all inductors are ideal components, and the grid voltage is three-phase symmetrical. Then, the mathematical model of the PV grid-connected inverter in the three-phase stationary coordinate system can be expressed as:

[0011]

[0012] The formula for converting current from the three-phase stationary coordinate system to the two-phase stationary coordinate system through Clark transformation is:

[0013]

[0014] The steps for converting the current in the three-phase stationary coordinate system to the two-phase stationary coordinate system through Clark transformation include: combining the α-β coordinate system with the mathematical model of the grid-connected inverter to obtain the inductor current in the two-phase stationary coordinate system as:

[0015]

[0016] Further, convert the current in the two-phase stationary coordinate system to the two-phase rotating coordinate system through Park transformation. The formula is:

[0017]

[0018] Combining the two-phase rotating coordinate system with the mathematical model of the grid-connected inverter, the mathematical model of the three-phase PV grid-connected inverter in the two-phase rotating coordinate system is:

[0019]

[0020] S2. Establishment of prediction model and value function

[0021] S201. Establish prediction current model

[0022] After discretizing and organizing the mathematical model using the forward Euler method, the following formula can be obtained:

[0023]

[0024] Among them, T s is the control period. In the two-phase rotating coordinate system, u dn( (k) and u qn( (k) are the voltage vectors output in different switching states at time k, u d( (k) and u q( (k) are the output voltages at time k, i d( (k) and i q (k), i d( (k + 1) and i q(The inductor currents at the k-th and (k + 1)-th instants are represented as such. To ensure that the midpoint voltage does not shift, a midpoint potential prediction model is established. Assume that the DC-side capacitors C1 = C2 = C dc , combined with the forward Euler method and the mathematical model of the capacitor voltage, the capacitor voltage difference can be expressed as:

[0025]

[0026] Combined with the three-phase NPC inverter topology, the capacitor current i c , the midpoint current i o and the three-phase current i x have the following relationships:

[0027] i c1 (k) - i c2 (k) = i o (k) = (1 - |S x (k)|)i x (k)

[0028] Combined with the midpoint potential prediction model, the capacitor voltage difference expression formula, and the relationship expression formulas of the capacitor current i c , the midpoint current i o and the three-phase current i x , the prediction model of the midpoint voltage can be expressed as:

[0029]

[0030] S202. Establish the value function

[0031] The value function designed in the present invention is

[0032]

[0033] Assume that the current can track the given reference current at the (k + 1)-th instant. Then, combined with the Lagrange extrapolation method, the current reference value can be expressed as:

[0034]

[0035] S3. Three-vector model current prediction optimization method

[0036] S301. Simplify the voltage vectors. PPO and OON can be represented by the vector . Therefore, the 27 voltage vectors can be simplified to 19 effective vectors (V1 to V 19 );

[0037] S302. Conduct sector division to obtain 24 sectors (N1 to N24);

[0038] S303. Duty ratio calculation, that is, calculate the action time of the three basic vectors in each sector. Considering that the action time and the value function are inversely proportional, the duty ratio corresponding to each sector can be expressed as:

[0039]

[0040] S304. After calculating the duty ratio, construct a new value function, and the new value function is:

[0041]

[0042] S305. Obtain the optimal sector, that is, through the optimization of the value function, select the sector opt that makes G j the smallest;

[0043] S306. Design the switching sequence. After obtaining the optimal sector, three optimal vectors can be further determined;

[0044] S4. Compensate for the delay through the two-step prediction method

[0045] S401. Sample the controlled quantity at time k;

[0046] S402. Use the prediction model to calculate the estimated value of the controlled quantity at time k + 1;

[0047] S403. According to the estimated value and the prediction model, calculate the predicted value of the controlled quantity at time k + 2;

[0048] S404. Combine the value function formula to calculate the value function g corresponding to 19 voltage vectors i ;

[0049] S405. Calculate the duty ratio d corresponding to the voltage vectors in 24 sectors n，j ;

[0050] S406. Construct a new value function formula according to the duty ratio;

[0051] S407. Optimize and screen out the sector opt and three voltage vectors that make the value function the smallest;

[0052] S408. According to the switching sequence, output the switching state of the optimal sector to drive the circuit;

[0053] S409. Save the voltage vectors and switching states in the optimal region and input them as original data in the next cycle.

[0054] Preferably, in the mathematical model of the photovoltaic grid-connected inverter in the three-phase stationary coordinate system in S1, L is the filter inductor, ix is the inductor current, R is the equivalent resistance of the filter inductor, uxn is the output voltage of the inverter, and ex is the grid voltage (x = a, b, c).

[0055] Preferably, in the mathematical model of the three-phase photovoltaic grid-connected inverter in S1 in the two-phase rotating coordinate system, id and iq are the inductor currents in the d-q subspace of the two-phase rotating coordinate system, udn and uqn are the inverter output voltages in the d-q subspace of the two-phase rotating coordinate system, ed and eq are the grid voltages in the d-q subspace of the two-phase rotating coordinate system, and w is the angular velocity of the two-phase rotating coordinate system.

[0056] Preferably, in the capacitor current i in S201 c , the neutral point current i o and the three-phase current i x in the relationship expression formula, x = a, b, c, S x represents the switch state, and the corresponding values of the states P, O, and N are 1, 0, and -1 respectively.

[0057] Preferably, in the value function in S202, i = 1 to 27, g i represents the value function corresponding to 27 basic voltage vectors, λ is the weighting factor, are the given reference values, that is, the expected output current.

[0058] Preferably, in the duty cycle expression formula corresponding to each sector in S303, d n，j is the duty cycle of the nth voltage vector in the jth sector, and g n,j is the corresponding value function value, which is calculated by the value function formula.

[0059] The present invention provides a model predictive optimization control method for a photovoltaic grid-connected inverter. It has the following

[0060] Beneficial effects:

[0061] 1. The present invention provides a model predictive optimization control method for a photovoltaic grid-connected inverter. By establishing a mathematical model of the photovoltaic grid-connected inverter, predicting the current of the grid-connected inverter, and based on the control objective, proposing an optimized three-vector model predictive control, the optimized control of the photovoltaic grid-connected inverter can be realized, improving the power quality and system stability, reducing the system cost, increasing the system efficiency. Applying the technical results of the present invention can effectively reduce the harmonic content of the current while improving the control accuracy of the grid current, accelerate the neutral point potential balancing speed, and improve the dynamic performance under various working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of the T-type three-level photovoltaic grid-connected inverter of the present invention;

[0063] Figure 2Schematic diagram of voltage vector distribution of the three-phase NPC inverter of the present invention;

[0064] Figure 3 Schematic diagram of the step flow for establishing the neutral point potential prediction model of the present invention;

[0065] Figure 4 Schematic diagram of the model prediction process of the present invention;

[0066] Figure 5 Schematic diagram of sector division of the present invention;

[0067] Figure 6 Schematic diagram of switch sequence selection of the present invention;

[0068] Figure 7 Schematic diagram of the three-vector model current prediction optimization method flow of the present invention. Detailed implementation manners

[0069] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0070] The embodiment of the present invention provides a model prediction optimization control method for a photovoltaic grid-connected inverter, which specifically includes the following steps:

[0071] S1. Mathematical modeling of the photovoltaic grid-connected inverter

[0072] A photovoltaic grid-connected inverter proposed by the present invention is as Figure 1 shown. The photovoltaic inverter includes a photovoltaic power generation panel, two DC-side capacitors, a T-type three-level topology structure and an output terminal. Figure 1 Among them, the photovoltaic power generation panel is on the far left, C1 and C2 are the two DC-side capacitors, the T-type three-level topology structure is in the middle, and the output side is on the right. The two DC-side capacitors are respectively marked as C1 and C2. The output side includes an LC filter and the grid voltage. Each arm of the T-type three-level inverter has three switching states, namely the P state, the O state, and the N state. The switching-on situation of the switching tubes corresponding to each switching state is as follows:

[0073]

[0074] Where S i is the switching state, and S ix represents the four switching tubes on each phase arm. Therefore, the three-phase three-level inverter has a total of 3 3= 27 working states, corresponding to 27 basic voltage vectors, and the 27 basic voltage vectors are as Figure 2 shown in

[0075] Assume that all power switches, all capacitors, and all inductors are ideal components, and the grid voltage is three-phase symmetrical. Then, the mathematical model of the PV grid-connected inverter in the three-phase stationary coordinate system can be expressed as:

[0076]

[0077] where L is the filter inductor, i x is the inductor current, R is the equivalent resistance of the filter inductor, u xn is the output voltage of the inverter, and e x is the grid voltage (x = a, b, c).

[0078] The formula for converting the current from the three-phase stationary coordinate system to the two-phase stationary coordinate system through the Clark transformation is:

[0079]

[0080] The steps for converting the current in the three-phase stationary coordinate system to the two-phase stationary coordinate system through the Clark transformation include: combining the α-β coordinate system with the mathematical model of the grid-connected inverter to obtain the inductor current in the two-phase stationary coordinate system as:

[0081]

[0082] Further, the current in the two-phase stationary coordinate system is converted to the two-phase rotating coordinate system through the Park transformation, and the formula is:

[0083]

[0084] Combining the two-phase rotating coordinate system with the mathematical model of the grid-connected inverter, the mathematical model of the three-phase PV grid-connected inverter in the two-phase rotating coordinate system is:

[0085]

[0086] where i d , i q are the inductor currents in the d-q subspaces of the two-phase rotating coordinate system, u dn , u qn are the output voltages of the inverter in the d-q subspaces of the two-phase rotating coordinate system, e d , e q are the grid voltages in the d-q subspaces of the two-phase rotating coordinate system, and w is the angular velocity of the two-phase rotating coordinate system.

[0087] S2. Establishment of the prediction model and value function

[0088] To improve the power quality of the power grid, that is, to achieve effective tracking of the grid current, the present invention proposes an optimization method for three-vector model current prediction.

[0089] Generally, the steps of traditional model predictive control include:

[0090] (1) Establish a prediction model and calculate the predicted value of the controlled variable at time k + 1;

[0091] (2) Design a cost function and traverse each voltage vector to calculate the cost function;

[0092] (3) Output the optimal switching state.

[0093] Introduced into the present invention, the overall model establishment steps are as follows:

[0094] S201. Establish a predictive current model

[0095] After discretizing and arranging the mathematical model using the forward Euler method, the following formula can be obtained:

[0096]

[0097] Among them, T s is the control period. In the two-phase rotating coordinate system, u dn( k) and u qn( k) are the voltage vectors output by different switching states at time k, u d( k) and u q( k) are the output voltages at time k, i d( k) and i q (k), i d( k + 1) and i q( k + 1) respectively represent the inductor currents at time k and time k + 1. The present invention proposes that while achieving effective tracking of the grid current, ensuring that the midpoint voltage does not shift, the steps of establishing a midpoint potential prediction model are as Figure 3 shown. Assuming that the DC-side capacitors C1 = C2 = C dc , combining the forward Euler method and the capacitor voltage mathematical model, the capacitor voltage difference can be expressed as:

[0098]

[0099] Combining the three-phase NPC inverter topology structure, the relationship between the capacitor current i c , the midpoint current i o and the three-phase current i x is as follows:

[0100] i c1 (k) - i c2 (k) = i o(k) = (1 - |S x (k)|)i x (k)

[0101] where x = a, b, c, and S x represents the switch state, and the values corresponding to states P, O, and N are 1, 0, and -1 respectively;

[0102] Combined with Figure 3 the midpoint potential prediction model, the capacitor voltage difference expression formula, and the capacitor current i c and the midpoint current i o and the relationship expression formula between the three-phase current i x The prediction model of the midpoint voltage can be expressed as:

[0103]

[0104] S202. Establish the value function

[0105] The value function designed in the present invention is

[0106]

[0107] where i = 1 to 27, and g i represents the value functions corresponding to 27 basic voltage vectors, and λ is the weight factor, which are the given reference values, i.e., the desired output current;

[0108] Assuming that the current can track the given reference current at the k + 1 moment, then combined with the Lagrange extrapolation method, the current reference value can be expressed as:

[0109]

[0110] The flowchart of this control algorithm is as Figure 4 shown. To ensure that the grid current is accurately tracked while the midpoint voltage does not shift, a value between 0 and 1 can be taken, which can be determined according to specific situations.

[0111] S3. Three-vector model current prediction optimization method

[0112] S301. Simplify the voltage vectors. PPO and OON can be represented by the vector , so the 27 voltage vectors can be simplified to 19 effective vectors (V1 to V 19 );

[0113] S302. Conduct sector division, and 24 sectors (N1 to N24) can be obtained. The divided sectors are as shown in Appendix Figure 5 ;

[0114] S303. Duty ratio calculation, that is, calculating the action time of the three basic vectors in each sector. Considering that the action time is inversely proportional to the value function, the duty ratio corresponding to each sector can be expressed as:

[0115]

[0116] where d n，j is the duty ratio of the nth voltage vector under the jth sector, and g n,j is the corresponding value function value, which is calculated through Equation (11);

[0117] S304. After calculating the duty ratio, a new value function is constructed. The new value function is:

[0118]

[0119] S305. Obtain the optimal sector, that is, through value function optimization, select the sector opt that minimizes G j ;

[0120] S306. Design the switching sequence. After obtaining the optimal sector, three optimal vectors can be further determined. When designing the switching sequence in the present invention, it is ensured that only the switches of one phase bridge arm act during each state transition, which can reduce the number of switch actions and losses. The design of the switching sequence is as Figure 6 shown.

[0121] S4. Compensate for the delay through the two-step prediction method

[0122] S401. Sample the controlled quantity at time k;

[0123] S402. Use the prediction model to calculate the estimated value of the controlled quantity at time k + 1;

[0124] S403. According to the estimated value and the prediction model, calculate the predicted value of the controlled quantity at time k + 2;

[0125] S404. Combine the value function formula to calculate the value function g i corresponding to 19 voltage vectors;

[0126] S405. Calculate the duty ratio d n，j corresponding to the voltage vectors in 24 sectors;

[0127] S406. Construct a new value function formula according to the duty ratio;

[0128] S407. Optimize and screen out the sector opt and three voltage vectors that minimize the value function;

[0129] S408. According to the switching sequence, output the switching state of the optimal sector to drive the circuit;

[0130] S409. Save the voltage vectors and switch states within the optimal region and input them as original data in the next cycle.

[0131] Combining the above steps, the three-vector model current prediction optimization method proposed by the present invention is as Figure 7 shown.

[0132] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A model predictive optimization control method for a photovoltaic grid-connected inverter, characterized in that, Specifically, it includes the following steps: S1. Mathematical modeling of a photovoltaic grid-connected inverter The photovoltaic inverter includes a photovoltaic power generation panel, two DC-side capacitors, a T-type three-level topology, and an output terminal. The two DC-side capacitors are respectively labeled as C1 and C2. The output side includes an LC filter and the grid voltage. Each arm of the T-type three-level inverter has three switching states, namely the P state, the O state, and the N state. The switching-on situation of the switching tubes corresponding to each switching state is as follows: Where S i is the switch state, and S ix represents the four switching devices on each phase leg; thus, the three-phase three-level inverter has a total of 3 3 = 27 operating states, corresponding to 27 basic voltage vectors; Assuming that all power switching tubes, all capacitors, and all inductors are ideal components, and the grid voltage is three-phase symmetric, the mathematical model of the photovoltaic grid-connected inverter in the three-phase stationary coordinate system can be expressed as: The formula for converting the current from the three-phase stationary coordinate system to the two-phase stationary coordinate system through the Clark transformation is: The steps for converting the current in the three-phase stationary coordinate system to the two-phase stationary coordinate system through the Clark transformation include: combining the α-β coordinate system with the mathematical model of the grid-connected inverter to obtain the inductor current in the two-phase stationary coordinate system as: Further, the current in the two-phase stationary coordinate system is converted to the two-phase rotating coordinate system through the Park transformation, and the formula is: Combining the two-phase rotating coordinate system with the mathematical model of the grid-connected inverter, the mathematical model of the three-phase photovoltaic grid-connected inverter in the two-phase rotating coordinate system is: S2. Establishment of a prediction model and a value function S201. Establish a predictive current model After discretizing and arranging the mathematical model using the forward Euler method, the following formula can be obtained: Among them, T s is the control period. Under the two-phase rotating coordinate system, u dn( (k) and u qn( (k) are the voltage vectors output at different switch states at time k, u d( (k) and u q( (k) are the output voltages at time k, i d( (k) and i q (k), i d( (k + 1) and i q( (k + 1) respectively represent the inductor currents at time k and time k + 1. To ensure that the midpoint voltage does not shift, a midpoint potential prediction model is established. Assuming that the DC-side capacitors C1 = C2 = C dc , combining the forward Euler method and the mathematical model of the capacitor voltage, the capacitor voltage difference can be expressed as: Combined with the three-phase NPC inverter topology, the capacitor current i c , the neutral point current i o and the three-phase current i x have the following relationship: i c1 (k)-i c2 (k) = i o (k) = (1 - |S x (k)|)i x (k) Combined with the midpoint potential prediction model, the capacitance differential pressure expression formula, and the capacitance current i c , the midpoint current i o and the three-phase current i x relationship expression formula, the prediction model of the midpoint voltage can be expressed as: S202. Establish a value function The value function designed in the present invention is Assuming that the current can track the given reference current at the k + 1 moment, then combining with the Lagrange extrapolation method, the current reference value can be expressed as: S3. Three-vector model current prediction optimization method S301. Simplify the voltage vectors. PPO and OON can be represented by the vector , so the 27 voltage vectors can be simplified to 19 effective vectors (V1 to V 19 ); S302. Perform sector division to obtain 24 sectors (N1 to N24); S303. Calculate the duty cycle, that is, calculate the action time of the three basic vectors in each sector. Combining that the action time is inversely proportional to the value function, the duty cycle corresponding to each sector can be expressed as: After calculating the duty cycle, a new value function is constructed. The new value function is: S305. Obtain the optimal sector, that is, through the optimization of the value function, select the sector opt that minimizes G j to be the smallest; S306. Design the switching sequence. After obtaining the optimal sector, three optimal vectors can be further determined; S4. Compensate for the delay through a two-step prediction method S401. Sample the controlled quantity at the k moment; S402. Use the prediction model to calculate the estimated value of the controlled quantity at the k + 1 moment; S403. According to the estimated value and the prediction model, calculate the predicted value of the controlled quantity at the k + 2 moment; S404. Calculate the value function g corresponding to 19 voltage vectors in combination with the value function formula i ; S405. Calculate the duty cycle d corresponding to the voltage vectors in 24 sectors n，j ; S406. Construct a new value function formula according to the duty cycle; S407. Optimize and screen out the sector opt and the three voltage vectors that minimize the value function; S408. According to the switching sequence, output the switching state driving circuit corresponding to the optimal sector; S409. Save the voltage vectors and switching states within the optimal region and input them as original data in the next cycle.

2. The model predictive optimization control method of a photovoltaic grid-connected inverter according to claim 1, characterized in that: In the mathematical model of the photovoltaic grid-connected inverter in the three-phase stationary coordinate system in S1, L is the filter inductor, ix is the inductor current, R is the equivalent resistance of the filter inductor, uxn is the inverter output voltage, and ex is the grid voltage (x = a, b, c).

3. The model predictive optimization control method for a photovoltaic grid-connected inverter according to claim 1, characterized in that: In the mathematical model of the three-phase photovoltaic grid-connected inverter in S1 in the two-phase rotating coordinate system, id and iq are the inductor currents in the d-q subspace of the two-phase rotating coordinate system, udn and uqn are the inverter output voltages in the d-q subspace of the two-phase rotating coordinate system, ed and eq are the grid voltages in the d-q subspace of the two-phase rotating coordinate system, and w is the angular velocity of the two-phase rotating coordinate system.

4. A model predictive optimization control method for a photovoltaic grid-connected inverter according to claim 1, characterized in that: The capacitive current i in S201 c , the neutral point current i o and the three-phase current i x in the relational expression formula, where x = a, b, c, and S x represents the switch state, and the corresponding values for states P, O, and N are 1, 0, and -1 respectively.

5. A model predictive optimization control method for a photovoltaic grid-connected inverter according to claim 1, characterized in that: In the value function in S202, i = 1 to 27, g i represents the value functions corresponding to 27 basic voltage vectors, and λ is a weighting factor. are respectively the given reference values, that is, the expected output current.

6. The model predictive optimization control method of a photovoltaic grid-connected inverter according to claim 1, characterized in that: In the duty cycle expression formula corresponding to each sector in S303, d n，j is the duty cycle of the nth voltage vector under the jth sector, and g n,j is the corresponding value function value, which is calculated through the value function formula.

Images