Rectification control method of photovoltaic energy storage bidirectional inverter

Abstract

The invention discloses a rectification control method of a photovoltaic energy storage bidirectional inverter. The method comprises the following steps: initializing control parameters of the photovoltaic energy storage bidirectional inverter; an MPC and PI fusion control T-type three-level rectifying circuit is adopted, AC / DC rectification is carried out on power grid current, and the power grid current is input to a direct current bus; an MPC and an NADRC are fused to control a CLLLC resonant circuit, DC / DC conversion is carried out on a DC bus of an input end, and the DC bus is input to charge a battery; a parameter detection evaluation and self-adaptive parameter updating algorithm is adopted to carry out optimization control on the AC / DC rectification and DC / DC conversion process, and the calculation reference value in the rectification and conversion control process in the charging and discharging process of the battery is updated in real time.

Description

Technical Field

[0001] The invention relates to a rectification control method for a photovoltaic energy storage bidirectional inverter. Background Art

[0002] The current energy transformation trend has promoted the deep integration of photovoltaic power generation, energy storage systems and power grids. With the increase in the proportion of renewable energy, the structure of the power system is changing, resulting in new challenges to the stability and reliability of the power grid. In order to achieve "peak shaving and valley flattening" for the power grid and solve the "power storage and access" problem, photovoltaic energy storage bidirectional inverter is one of the key and core.

[0003] Photovoltaic energy storage bidirectional inverters absorb energy from low-price power grids at night to replenish part of the power for energy storage devices, so that they can provide stable power support to the power grid in cooperation with photovoltaic inverters during the day, thereby alleviating the intermittent and instability of photovoltaic power generation. However, photovoltaic energy storage bidirectional inverters face problems such as harmonic distortion, low power factor, insufficient response speed and dynamic performance in the bidirectional energy flow management between the power grid and the internal energy storage devices.

[0004] In order to better adapt to the bidirectional energy exchange between the photovoltaic energy storage system and the power grid, the photovoltaic energy storage bidirectional inverter adopts a T-type three-level topology in the bidirectional AC / DC module. Compared with the traditional two-level topology, it can effectively improve the rectification and inversion performance, reduce losses and harmonic content, and improve power density and system efficiency. The CLLLC resonant circuit is used in the bidirectional DC / DC module. Compared with the traditional bidirectional DC / DC circuit, it has greater advantages in high efficiency, bidirectional energy flow, high power density and wide voltage range adaptability. However, in the process of using the T-type three-level circuit AC / DC rectification in the front stage and the CLLLC resonant DC / DC circuit in the back stage to realize the energy storage charging, the midpoint potential offset during the rectification control of the T-type three-level circuit will cause the output voltage waveform to be distorted. At the same time, the high-frequency electromagnetic interference generated by the front-stage rectification and the back-stage CLLLC resonant circuit will affect the system power factor and harmonic performance. Especially when the grid voltage fluctuates or the load changes greatly, the midpoint voltage balance problem is particularly prominent, which directly affects the bus power quality of the photovoltaic energy storage bidirectional inverter system and indirectly affects the control stability and power quality.

[0005] In order to solve these problems, the present invention proposes a rectifier control method for a photovoltaic energy storage bidirectional inverter, which eliminates the interference of input voltage and load changes on bus voltage and charging and discharging current based on nonlinear active disturbance rejection control (NADRC). In the front-stage rectifier control, NADRC is used to stabilize the bus voltage; in the back-stage CLLLC control, it can be used to accurately track the charging and discharging current of the battery. Summary of the invention

[0006] The present invention is to solve the problems existing in the above-mentioned prior art and provide a rectification control method for a photovoltaic energy storage bidirectional inverter. The photovoltaic energy storage bidirectional inverter can take into account the dynamic performance, accuracy of the front-stage rectification and the rear-stage DCDC conversion control and the overall efficiency of the photovoltaic energy storage bidirectional inverter system in rectification and energy storage charging control.

[0007] The technical solutions adopted in the present invention are:

[0008] A rectification control method for a photovoltaic energy storage bidirectional inverter, the photovoltaic energy storage bidirectional inverter comprising a T-type three-level rectification circuit, a CLLLC resonant circuit and an LCL filter, the LCL filter being connected to the T-type three-level rectification circuit, the T-type three-level rectification circuit and the CLLLC resonant circuit being connected via a bus, the control method comprising the following steps:

[0009] S1: Initialize the control parameters of the photovoltaic energy storage bidirectional inverter;

[0010] S2: MPC and PI fusion control T-type three-level rectifier circuit is used to realize AC / DC rectification of grid current and input it into the DC bus;

[0011] S3: MPC and NADRC are combined to control the CLLLC resonant circuit to achieve DC / DC conversion of the DC bus at the input end and input it to charge the battery;

[0012] S4: Optimize the control of AC / DC rectification and DC / DC conversion process by using parameter detection evaluation and adaptive parameter update algorithm, and update the calculated reference values ​​in the rectification and conversion control process during the battery charging and discharging process in real time.

[0013] Furthermore, in S2, MPC and PI are used to integrate and control the T-type three-level rectifier circuit, and the process is as follows:

[0014] S21) Construction and discretization of the state space model of T-type three-level rectifier circuit;

[0015] S22) calculating the bus voltage error, and inputting the calculated error into the PI controller to achieve error correction;

[0016] S23) The output voltage corrected by the PI controller is integrated with the MPC control output, and finally the prediction error is calculated.

[0017] Further, the process of S21) is:

[0018] Establish a description of the output bus DC capacitor voltage of the T-type three-level rectifier circuit And the state space equation of the input grid current i1 is shown in equation (3):

[0019]

[0020] Where: x j is the state of the photovoltaic energy storage bidirectional inverter at the current moment, as shown in formula (4):

[0021]

[0022] u j is the control quantity, that is, the bridge arm switch control signal of the T-type three-level rectifier circuit, as shown in formula (5):

[0023]

[0024] Where: S1∈(0,1) is the positive arm switch state of the T-type three-level rectifier circuit, 0 means connected to the midpoint voltage 0V, 1 means connected to S2∈(0,1) is the switch state of the negative bridge arm, 0 means connected to the midpoint voltage 0V, 1 means connected to y j The output bus DC capacitor voltage of the T-type three-level rectifier circuit is The DC capacitor current of the output busbar of the T-type three-level rectifier circuit

[0025] The state transfer matrix A is shown in formula (6):

[0026]

[0027] The input matrix B is shown in formula (7):

[0028]

[0029] The output matrix C is shown in formula (8):

[0030]

[0031] Further, in S22), the bus voltage error calculation formula is:

[0032]

[0033] in, is the bus voltage reference value;

[0034] Input the error e(t) into the PI controller and dynamically adjust the proportional gain K 1P And the integral gain K 1I , to adapt to grid voltage fluctuations and load changes, the formula is:

[0035]

[0036] Among them, u PI is the output voltage corrected by the PI controller, K 1P Increase to speed up error correction, K 1I Reduce to avoid integral windup and oscillation.

[0037] Furthermore, in S23), the output voltage u after correction by the PI controller is PI and MPC control output u MPC Combine by weight:

[0038] x j,p =α·u PI +(1-α)·x j (11)

[0039] Among them, α is the weight;

[0040] Based on the currently sampled bus voltage and grid voltage, the compensated x j,p and control input u j , predict the state x for multiple time steps in the future j+1,p ,x j+2,p ,…,x j+N,p .

[0041] Furthermore, the switch combination state of the T-type three-level rectifier circuit is controlled by fusion of MPC and PI, and the process is:

[0042] Calculate the objective function J1:

[0043]

[0044] Among them: λ, Q1, Q2 are weight coefficients, x 4,p [k+1] is the predicted bus voltage at time k+1, x 3,p [k+1] is the predicted bus current at time k+1, is the reference value of bus voltage, is the reference value of the output current, which is set to be in phase with the grid voltage; To predict the voltage difference between the two DC bus sides at k+1,

[0045] Solve the optimization problem and obtain the optimal control switch sequence {u j ,u j+1 ,…,u j+N}(13).

[0046] Furthermore, in S3, the MPC and NADRC are integrated to control the CLLLC resonant circuit to perform DC / DC conversion on the DC bus at the input end, specifically:

[0047] (S31) Construction and discretization of the state space model of CLLLC resonant circuit;

[0048] Based on the dynamic relationship between voltage and current of the CLLLC resonant circuit, a discrete time state space model is established:

[0049] x′(k+1)=A′·x′(k)+B′·u′(k)(14)

[0050] Among them, x′(k) is the state of the photovoltaic energy storage bidirectional inverter at time k, which is expressed as follows:

[0051]

[0052] Among them, i r is the resonant current; v c is the resonant capacitor voltage; v o is the transformer output voltage; i o is the transformer secondary current; u′(k) is the control input, which is determined by the switch state of the CLLLC resonant circuit input bridge at time k, u′(k)=v indc ;

[0053] The state transfer matrix A′ is expressed as follows:

[0054]

[0055] Among them, L r is the resonant inductor, C r is the resonant capacitor, L o is the output filter inductor in the CLLLC resonant circuit, C o is the output filter capacitor, Ts is the sampling period of the CLLLC resonant circuit, and I is the unit matrix;

[0056] The input matrix B′ is expressed as follows:

[0057]

[0058] (S32) NADRC nonlinear active disturbance rejection control:

[0059] The nonlinear dynamics, external disturbances, and parameter uncertainties in the CLLLC resonant circuit are modeled as an extended state:

[0060]

[0061] Where z1 is the battery current i b ; z2 is the current change rate z3 is the unknown interference d;

[0062] Use the state observer to estimate the states z1, z2, z3 of the CLLLC resonant circuit:

[0063]

[0064] Wherein, L3 is the state observer gain coefficient, S is the control signal input of the switch state of the CLLLC resonant circuit;

[0065] (S33) Nonlinear feedback law design:

[0066] Use nonlinear feedback law to compensate for disturbances and control the output battery circuit observation value u NADRC :

[0067]

[0068] Where: K1 and K2 are the state observer feedback gain weights; i b,ref is the battery current reference value; is the estimated value of the current change rate; is the disturbance estimated by the state observer;

[0069] Use nonlinear auto-disturbance rejection control to control the output u NADRC The state-space model output of the compensated CLLLC resonant circuit o (k+1)≈i b,p (k+1), as shown in the following formula:

[0070]

[0071] (S34) Predict current trajectory:

[0072] Using the current measurement and the state x′(k) and control input u′(k) compensated by nonlinear ADRC, the battery current trajectory i at the next k+1 time is predicted b,p (k+1);

[0073] (345)Objective function evaluation:

[0074] Based on the working state of the switch tube of the CLLLC resonant circuit, a limited number of candidate control actions are defined. For each candidate control action S i , calculate the objective function J2:

[0075] J2=(i b,ref (k+1)-i b,p (k+1)) 2 (twenty two)

[0076] (346) Select the best action:

[0077] According to the calculated objective function J of each candidate control action, the optimal switch combination S that minimizes J is selected. * , as shown below:

[0078]

[0079] Furthermore, the parameter detection evaluation and adaptive parameter update algorithm are used to optimize the control of the AC / DC rectification and DC / DC conversion process. The process is as follows:

[0080] (41) Collecting circuit parameters of the photovoltaic energy storage bidirectional inverter, including bus voltage reference value, proportional gain and integral gain of the PI controller, MPC prediction time domain, battery charge and discharge current reference range, and feedback gain of NADRC control;

[0081] (42) Perform global optimization operations on the T-type three-level rectifier circuit and the CLLLC resonant circuit to obtain a global optimization result;

[0082] (43) According to the optimization results, the control parameters of the T-type three-level rectifier circuit and the CLLLC resonant circuit are adjusted to correspond to:

[0083] (1) The proportional gain and integral gain of the PI controller take optimal values ​​in normal or fluctuating conditions;

[0084] (2) Dynamic update of MPC constraints to ensure that the bus voltage is stable within the reference range;

[0085] (3) DC bus voltage Make appropriate range changes according to the operating frequency of the CLLLC resonant circuit;

[0086] (4) The state observer feedback gain weights K1 and K2 take the optimal values ​​in normal state or fluctuating state respectively;

[0087] (5) The battery charge and discharge current is updated according to demand.

[0088] The present invention has the following beneficial effects:

[0089] The present invention adopts hierarchical control, the bottom layer is from rectification to conversion circuit control, and the top layer is to optimize the control performance of the bottom layer. In the bottom front-stage T-type three-level rectifier circuit, by adopting the dual adjustment mechanism of PI compensation combined with MPC model predictive control, the system can achieve rapid response and accurate tracking of dynamic parameters, and at the same time has good anti-disturbance ability and adaptability, especially under the working conditions of large load changes, it can still maintain a stable voltage output. In the bottom rear-stage CLLLC resonant circuit, the NADRC nonlinear dynamic compensator is combined with the MPC model predictive control, which not only significantly improves the suppression ability of nonlinear disturbances, but also greatly optimizes the current tracking accuracy and the overall efficiency of the system. This dual adjustment mechanism effectively solves the limitations of traditional control methods in resonant circuits. At the same time, at the top global optimization level, the MPC-based optimization algorithm can update the system dynamic parameters in real time, and by coordinating the operating characteristics between the front stage and the rear stage, ensure the high efficiency and stability under the entire dynamic working condition. At the same time, through the continuous monitoring and iterative optimization of key indicators such as bus voltage, battery SOC (State of Charge), current accuracy and system efficiency, the comprehensive performance of the system is further improved. In general, while ensuring system stability and response speed, the operating efficiency and control accuracy are significantly improved, providing an efficient and reliable solution for the combined application of T-type three-level rectifier circuit and CLLLC resonant circuit. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 This is the circuit diagram of the photovoltaic energy storage bidirectional inverter.

[0091] Figure 2 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0092] The present invention will be further described below in conjunction with the accompanying drawings.

[0093] like Figure 1 and Figure 2 The present invention provides a rectification control method for a photovoltaic energy storage bidirectional inverter, wherein the photovoltaic energy storage bidirectional inverter comprises a T-type three-level rectification circuit, a CLLLC resonant circuit and an LCL filter, wherein the LCL filter is connected to the T-type three-level rectification circuit, and the T-type three-level rectification circuit and the CLLLC resonant circuit are connected via a bus. The control method of the present invention comprises the following steps:

[0094] S1: Initialize the control parameters of the photovoltaic energy storage bidirectional inverter;

[0095] S2: MPC and PI fusion control T-type three-level rectifier circuit is used to realize AC / DC rectification of grid current and input it into the DC bus;

[0096] S3: MPC and NADRC are combined to control the CLLLC resonant circuit to achieve DC / DC conversion of the DC bus at the input end and input it to charge the battery;

[0097] S4: Optimize the control of AC / DC rectification and DC / DC conversion process by using parameter detection evaluation and adaptive parameter update algorithm, and update the calculated reference values ​​in the rectification and conversion control process during the battery charging and discharging process in real time.

[0098] In S1, specifically:

[0099] 1. Initialize system hardware

[0100] (1) T-type three-level rectifier circuit: Input voltage: three-phase 380VAC, frequency 50Hz. DC bus voltage target value is set to 400V. Switching frequency: 10kHz Midpoint voltage reference: half of the DC bus voltage (about 200V). Filter inductance value: 4mH Bus side filter capacitor value: 220μF.

[0101] (2) CLLLC resonant circuit: Resonant frequency: 100kHz. Resonant parameters: Resonant capacitance C r :22nF。 Switching frequency range: 90kHz~110kHz (charge and discharge power regulation). Input voltage range (bus voltage): 380V~420V。 Output voltage range (battery end): 40V~58V。

[0102] 2. Set control parameters

[0103] (1) Bus voltage reference value: 400 V (steady-state target).

[0104] The allowable deviation range of steady state is ±2% (about 8V).

[0105] (2) Battery charge and discharge current range:

[0106] Charging current limit: maximum 50A (corresponding to 1C~2C charging rate);

[0107] Discharge current limit: Maximum 60A (considering high power load).

[0108] (3) Control target priority:

[0109] Priority 1: Stabilize the DC bus voltage (pre-stage T-type three-level rectifier circuit).

[0110] Priority 2: Accurately control the battery charge and discharge current (post-CLLLC resonant circuit).

[0111] Priority 3: Suppress high-frequency interference and optimize power factor and harmonic performance.

[0112] 3. Initialize control algorithm parameters

[0113] (1) PI controller parameters:

[0114] Proportional gain K 1P =0.9. Integral gain K 1I =0.02. Adaptive adjustment range: K 1P Dynamic range: 0.6~1.5. K 1I Dynamic range: 0.01~0.03.

[0115] (2) Model predictive control (MPC) parameters:

[0116] Prediction time domain: Front-stage control (bus voltage): 5 sampling cycles. Back-stage control (charging and discharging current): 3 sampling cycles.

[0117] Control time domain: 1 sampling period.

[0118] Constraints: Bus voltage constraint: 380V~420V. Output current constraint: ±60A (depending on the charge and discharge direction). Midpoint voltage balance constraint: ±5V (offset range).

[0119] (3) Nonlinear active disturbance rejection control (NADRC) parameters:

[0120] Extended State Observer (ESO) parameters: Gain coefficient: L3 = 150 (enhanced response to medium and low voltage disturbances). Filter time constant: 5ms. Nonlinear feedback gain: dynamic range 0.1 to 3 (automatically adjusted according to disturbance changes).

[0121] In S2, MPC and PI are used to combine and control the T-type three-level rectifier circuit. The process is as follows:

[0122] S21) Construction and discretization of the state space model of T-type three-level rectifier circuit;

[0123] Based on the input side inductor L1 in the LCL filter, the KVL dynamic equation is derived, and the three-phase variables in the abc coordinate system are subjected to Clarke transformation to obtain the dynamic equation in the αβ coordinate system, as shown in equation (1):

[0124]

[0125] Wherein, L1 is the inductance at the input side of the LCL filter; i1 is the input grid current; u in is the input grid voltage; u fis the capacitor voltage of the LCL filter; R1 is the equivalent series resistance of the input side inductor L1 in the LCL filter; L2 is the output side inductor of the LCL filter; R2 is the equivalent series resistance of the output side inductor L2 in the LCL filter; u1 is the voltage input to the T-type three-level rectifier circuit through the LCL filter; i2 is the current input to the T-type three-level rectifier circuit through the LCL filter; Output bus DC capacitor voltage for T-type three-level rectifier circuit; Output bus DC capacitor current for T-type three-level rectifier circuit; is the output bus DC capacitor of the T-type three-level rectifier circuit; x dc =1, 2;

[0126] The forward Euler method is used to discretize the model, as shown in formula (2):

[0127]

[0128] Establish a description of the output bus DC capacitor voltage of the T-type three-level rectifier circuit And the state space equation of the input grid current i1 is shown in equation (3):

[0129]

[0130] Where: x j is the state of the photovoltaic energy storage bidirectional inverter at the current moment, as shown in formula (4):

[0131]

[0132] u j is the control quantity, that is, the bridge arm switch control signal of the T-type three-level rectifier circuit, as shown in formula (5):

[0133]

[0134] Where: S1∈(0,1) is the positive arm switch state of the T-type three-level rectifier circuit, 0 means connected to the midpoint voltage 0V, 1 means connected to S2∈(0,1) is the switch state of the negative bridge arm, 0 means connected to the midpoint voltage 0V, 1 means connected to y j The output bus DC capacitor voltage of the T-type three-level rectifier circuit is The DC capacitor current of the output bus of the T three-level rectifier

[0135] The state transfer matrix A is shown in formula (6):

[0136]

[0137] The input matrix B is shown in formula (7):

[0138]

[0139] The output matrix C is shown in formula (8):

[0140]

[0141] S22) calculating the bus voltage error, and inputting the calculated error into the PI controller to achieve error correction;

[0142] The bus voltage error calculation formula is:

[0143]

[0144] in, is the bus voltage reference value;

[0145] Input the error e(t) into the PI controller and dynamically adjust the proportional gain K 1P And the integral gain K 1I , to adapt to grid voltage fluctuations and load changes, the formula is:

[0146]

[0147] Among them, u PI is the output voltage corrected by the PI controller, K 1P Increase to speed up error correction, K 1I Reduce to avoid integral windup and oscillation.

[0148] S23) The output voltage corrected by the PI controller is integrated with the control output of the MPC prediction model, and finally the prediction error is calculated.

[0149] The output voltage u after PI controller correction PI and the MPC prediction model control output u MPC Combine by weight:

[0150] x j,p =α·u PI +(1-α)·x j (11)

[0151] Among them, α is the weight;

[0152] Based on the currently sampled bus voltage and grid voltage, the compensated x j,p and control input u j , predict the state x for multiple time steps in the future j+1,p ,x j+2,p ,…,x j+N,p .

[0153] The switch combination state of the T-type three-level rectifier circuit is controlled by MPC and PI fusion. The process is as follows:

[0154] Calculate the objective function J1:

[0155]

[0156] Among them: λ, Q1, Q2 are weight coefficients, x 4,p [k+1] is the predicted bus voltage at time k+1, x 3,p [k+1] is the predicted bus current at time k+1, is the reference value of bus voltage, is the reference value of the output current, which is set to be in phase with the grid voltage; To predict the voltage difference between the two DC bus sides at k+1,

[0157] Solve the optimization problem and obtain the optimal control switch sequence {u j ,u j+1 ,…,u j+N}(13).

[0158] In S3, the MPC and NADRC are combined to control the CLLLC resonant circuit to perform DC / DC conversion on the DC bus at the input end, specifically:

[0159] (S31) Construction and discretization of the state space model of CLLLC resonant circuit;

[0160] Based on the dynamic relationship between voltage and current of the CLLLC resonant circuit, a discrete time state space model is established:

[0161] x′(k+1)=A′·x′(k)+B′·u′(k)(14)

[0162] Among them, x′(k) is the state of the photovoltaic energy storage bidirectional inverter at time k, which is expressed as follows:

[0163]

[0164] Among them, i r is the resonant current; v c is the resonant capacitor voltage; v o is the transformer output voltage; i o is the transformer secondary current; u′(k) is the control input, which is determined by the switch state of the CLLLC resonant circuit input bridge at time k, u′(k)=v indc ;

[0165] The state transfer matrix A′ is expressed as follows:

[0166]

[0167] Among them, L r is the resonant inductor, C r is the resonant capacitor, L o is the output filter inductor in the CLLLC resonant circuit, C o is the output filter capacitor, Ts is the sampling period of the CLLLC resonant circuit, and I is the unit matrix;

[0168] The input matrix B′ is expressed as follows:

[0169]

[0170] (S32) NADRC nonlinear active disturbance rejection control:

[0171] The nonlinear dynamics, external disturbances, and parameter uncertainties in the CLLLC resonant circuit are modeled as an extended state:

[0172]

[0173] Where z1 is the battery current i b ; z2 is the current change rate z3 is the unknown interference d;

[0174] Use the state observer to estimate the states z1, z2, z3 of the CLLLC resonant circuit:

[0175]

[0176] Wherein, L3 is the state observer gain coefficient, S is the control signal input of the switch state of the CLLLC resonant circuit;

[0177] (S33) Nonlinear feedback law design:

[0178] Use nonlinear feedback law to compensate for disturbances and control the output battery circuit observation value u NADRC :

[0179]

[0180] Where: K1 and K2 are the state observer feedback gains; i b,ref is the battery current reference value; is the estimated value of the current change rate; is the disturbance estimated by the state observer;

[0181] Use nonlinear auto-disturbance rejection control to control the output u NADRC The state-space model output of the compensated CLLLC resonant circuit o(k+1)≈i b,p (k+1), as shown in the following formula:

[0182]

[0183] (S34) Predict current trajectory:

[0184] Using the current measurement and the state x′(k) and control input u′(k) compensated by nonlinear ADRC, the battery current trajectory i at the next k+1 time is predicted b,p (k+1);

[0185] (345)Objective function evaluation:

[0186] Based on the working state of the switch tube of the CLLLC resonant circuit, a limited number of candidate control actions are defined. For each candidate control action S i , calculate the objective function J2:

[0187] J2=(i b,ref (k+1)-i b,p (k+1)) 2 (twenty two)

[0188] (346) Select the best action:

[0189] According to the calculated objective function J of each candidate control action, the optimal switch combination S that minimizes J is selected. * , as shown below:

[0190]

[0191] When the T-type three-level rectifier circuit and the CLLLC resonant circuit are in operation, the circuit parameters of the two are collected to evaluate the performance indicators and adaptively update the reference values ​​in the front and rear stage control algorithms. The specific steps are as follows:

[0192] 1. Local control layer parameter design of hierarchical control structure:

[0193] 1.1 Control parameters of the front-stage T-type three-level rectifier circuit:

[0194] Bus voltage reference value: V bus,ref =400V;

[0195] PI control gain range: K 1p =[0.1,0.5], K 1i =[0.01,0.05];

[0196] MPC prediction time domain: N1=10.

[0197] Constraints: Bus voltage fluctuation range: ±5V; midpoint voltage offset limit: ±2V.

[0198] 1.2 Post-stage CLLLC resonant circuit control parameters:

[0199] Battery charge and discharge current reference range:

[0200] - Charging: I charge =[0,20]A;

[0201] Discharge: I discharge =[-20,0]A

[0202] Prediction time domain: N2=5.

[0203] Nonlinear active disturbance rejection control (NADRC) parameters: Extended state observer (ESO) gain: L3 = [0.1, 0.5];

[0204] Nonlinear feedback control gain: K1 = [0.2, 0.8], K2 = [0.2, 0.8].

[0205] 2 Global optimization layer operations:

[0206] Bus voltage v dc Stability is the priority goal, as shown in the following formula, with the default weight W v =0.7, charge and discharge accuracy weight W i =0.3.

[0207] J HMPC =W v ·J T3L (m+1) 2 +W i ·J CLLLC (m+1) 2 (twenty four)

[0208] In the formula, J T3L (m+1) is the optimal objective function value in the T-type three-level control at time m+1, J CLLLC (m+1) is the optimal objective function value in the CLLLC resonant circuit control at time m+1.

[0209] When the grid voltage fluctuation is ≥5%, the bus voltage priority W is increased. v =0.9; When the battery SOC is <30% or >80%, the priority of improving the charge and discharge accuracy is W i =0.6.

[0210] By monitoring bus voltage and load changes, calculate real-time voltage fluctuations and charging and discharging requirements. Allocate resources according to the optimization algorithm:

[0211] (1) When the load changes suddenly, priority should be given to ensuring the stability of the bus voltage.

[0212] (2) When the battery SOC is close to the boundary, priority is given to ensuring charging and discharging accuracy.

[0213] Control resource dynamic allocation:

[0214] (1) Adjust the previous stage PI gain and MPC output range to stabilize the bus voltage;

[0215] (2) Adjust the objective function weight or feedback gain of the subsequent charge and discharge controller to meet the battery requirements.

[0216] Adjust control parameters according to optimization results:

[0217] (1) PI controller gain K 1P ,K 1I Take the optimal value in normal state or fluctuating state respectively;

[0218] (2) Dynamic update of MPC constraints to ensure that the bus voltage is stable within the reference range;

[0219] (3) DC bus voltage The appropriate range is changed according to the operating frequency of the CLLLC resonant circuit.

[0220] (4) The state observer feedback gain weights K1 and K2 take the optimal values ​​in normal state or fluctuating state respectively;

[0221] (5) The reference charging current of the energy storage device is updated according to the charging demand current of the energy storage management system.

[0222] The above description is only a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements without departing from the principle of the present invention, and these improvements should also be regarded as within the protection scope of the present invention.

Claims

1. A rectification control method for a photovoltaic energy storage bidirectional inverter, wherein the photovoltaic energy storage bidirectional inverter comprises a T-type three-level rectification circuit, a CLLLC resonant circuit and an LCL filter, wherein the LCL filter is connected to the T-type three-level rectification circuit, and the T-type three-level rectification circuit and the CLLLC resonant circuit are connected via a bus, wherein: The control method comprises the following steps: S1: Initialize the control parameters of the photovoltaic energy storage bidirectional inverter; S2: MPC and PI fusion control T-type three-level rectifier circuit is used to realize AC / DC rectification of grid current and input it into the DC bus; S3: MPC and NADRC are combined to control the CLLLC resonant circuit to achieve DC / DC conversion of the DC bus at the input end and input it to charge the battery; S4: Optimize the control of AC / DC rectification and DC / DC conversion process by using parameter detection evaluation and adaptive parameter update algorithm, and update the calculated reference values ​​in the rectification and conversion control process during the battery charging and discharging process in real time.

2. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 1, characterized in that: In S2, MPC and PI are used to combine and control the T-type three-level rectifier circuit. The process is as follows: S21) Construction and discretization of the state space model of T-type three-level rectifier circuit; S22) calculating the bus voltage error, and inputting the calculated error into the PI controller to achieve error correction; S23) The output voltage corrected by the PI controller is integrated with the MPC control output, and finally the prediction error is calculated.

3. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 2, characterized in that: The process of S21) is: Establish a description of the output bus DC capacitor voltage of the T-type three-level rectifier circuit And the state space equation of the input grid current i1 is shown in equation (3): Where: x j is the state of the photovoltaic energy storage bidirectional inverter at the current moment, as shown in formula (4): u j is the control quantity, that is, the bridge arm switch control signal of the T-type three-level rectifier circuit, as shown in formula (5): Where: S1∈(0,1) is the positive arm switch state of the T-type three-level rectifier circuit, 0 means connected to the midpoint voltage 0V, 1 means connected to S2∈(0,1) is the switch state of the negative bridge arm, 0 means connected to the midpoint voltage 0V, 1 means connected to y j The output bus DC capacitor voltage of the T-type three-level rectifier circuit is The DC capacitor current of the output busbar of the T-type three-level rectifier circuit i2 is the current input to the T-type three-level rectifier circuit through the LCL filter; u f is the capacitor voltage of the LCL filter; The state transfer matrix A is shown in formula (6): The input matrix B is shown in formula (7): The output matrix C is shown in formula (8): L1 is the input side inductance of the LCL filter; R1 is the equivalent series resistance of the input side inductance L1 in the LCL filter; L2 is the output side inductance of the LCL filter; R2 is the equivalent series resistance of the output side inductance L2 in the LCL filter.

4. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 2, characterized in that: In the S22), the bus voltage error calculation formula is: in, is the bus voltage reference value; Input the error e(t) into the PI controller and dynamically adjust the proportional gain K 1P And the integral gain K 1I , to adapt to grid voltage fluctuations and load changes, the formula is: Among them, u PI is the output voltage corrected by the PI controller, K 1P Increase to speed up error correction, K 1I Reduce to avoid integral windup and oscillation.

5. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 2, characterized in that: In S23), the output voltage u after the PI controller is corrected PI and MPC control output u MPC Combine by weight: x j,p =a·u PI +(1-a)·x j (11) Among them, α is the weight; Based on the current sampled bus voltage and grid voltage, the compensated x j,p and control input u j , predict the state x for multiple time steps in the future j+1,p ,x j+2,p ,…,x j+N,p .

6. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 2, characterized in that: The switch combination state of the T-type three-level rectifier circuit is controlled by MPC and PI fusion. The process is as follows: Calculate the objective function J1: Among them: λ, Q1, Q2 are weight coefficients, x 4,p [k+1] is the predicted bus voltage at time k+1, x 3,p [k+1] is the predicted bus current at time k+1, is the reference value of bus voltage, is the reference value of the output current, which is set to be in phase with the grid voltage; To predict the voltage difference between the two DC bus sides at k+1, Solve the optimization problem and obtain the optimal control switch sequence {u j ,u j+1 ,…,u j+N }(13).

7. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 1, characterized in that: In S3, the MPC and NADRC are combined to control the CLLLC resonant circuit to perform DC / DC conversion on the DC bus at the input end, specifically: (S31) Construction and discretization of the state space model of CLLLC resonant circuit; Based on the dynamic relationship between voltage and current of the CLLLC resonant circuit, a discrete time state space model is established: x′(k+1)=A′·x′(k)+B′·u′(k)(14) Among them, x′(k) is the state of the photovoltaic energy storage bidirectional inverter at time k, which is expressed as follows: Among them, i r is the resonant current; v c is the resonant capacitor voltage; v o is the transformer output voltage; i o is the transformer secondary current; u′(k) is the control input, which is determined by the switch state of the CLLLC resonant circuit input bridge at time k, u′(k)=v indc ; The state transfer matrix A′ is expressed as follows: Among them, L r is the resonant inductor, C r is the resonant capacitor, L o is the output filter inductor in the CLLLC resonant circuit, C o is the output filter capacitor, Ts is the sampling period of the CLLLC resonant circuit, and I is the unit matrix; The input matrix B′ is expressed as follows: (S32) NADRC nonlinear active disturbance rejection control: The nonlinear dynamics, external disturbances, and parameter uncertainties in the CLLLC resonant circuit are modeled as an extended state: Where z1 is the battery current i b ; z2 is the current change rate z3 is the unknown interference d; Use the state observer to estimate the states z1, z2, z3 of the CLLLC resonant circuit: Wherein, L3 is the state observer gain coefficient, S is the control signal input of the switch state of the CLLLC resonant circuit; (S33) Nonlinear feedback law design: Use nonlinear feedback law to compensate for disturbances and control the output battery circuit observation value u NADRC : Where: K1 and K2 are the state observer feedback gain weights; i b,ref is the battery current reference value; is the estimated value of the current change rate; is the disturbance estimated by the state observer; Use nonlinear active disturbance rejection control to control the output u NADRC The state-space model output of the compensated CLLLC resonant circuit o (k+1)≈i b,p (k+1), as shown in the following formula: (S34) Predict current trajectory: Using the current measurement and the state x′(k) and control input u′(k) compensated by nonlinear ADRC, the battery current trajectory i at the next k+1 time is predicted b,p (k+1); (345)Objective function evaluation: Based on the working state of the switch tube of the CLLLC resonant circuit, a limited number of candidate control actions are defined. For each candidate control action S i , calculate the objective function J2: J2=(i b,ref (k+1)-i b,p (k+1)) 2 (22) (346) Select the best action: According to the calculated objective function J of each candidate control action, the optimal switch combination S that minimizes J is selected. * , as shown below:

8. The rectification control method of the photovoltaic energy storage bidirectional inverter according to claim 1, characterized in that: The parameter detection evaluation and adaptive parameter update algorithm are used to optimize the control of AC / DC rectification and DC / DC conversion process. The process is as follows: (41) Collecting circuit parameters of the photovoltaic energy storage bidirectional inverter, including bus voltage reference value, proportional gain and integral gain of the PI controller, MPC prediction time domain, battery charge and discharge current reference range, and feedback gain of NADRC control; (42) Perform global optimization operations on the T-type three-level rectifier circuit and the CLLLC resonant circuit to obtain a global optimization result; (43) According to the optimization results, the control parameters of the T-type three-level rectifier circuit and the CLLLC resonant circuit are adjusted to correspond to: (1) The proportional gain and integral gain of the PI controller take optimal values ​​in normal or fluctuating conditions; (2) Dynamic update of MPC constraints to ensure that the bus voltage is stable within the reference range; (3) DC bus voltage Make appropriate range changes according to the operating frequency of the CLLLC resonant circuit; (4) The state observer feedback gain weights K1 and K2 take the optimal values ​​in normal state or fluctuating state respectively; (5) The battery charge and discharge current is updated according to demand.

Images