Improved Model Predictive Vector Control Method for Grid-Connected Inverters Based on Evaluation Function

Abstract

This invention relates to a grid-connected inverter control method based on improved evaluation functions and model predictive vector control, belonging to the power industry. Building upon SVPWM (Single Vector Width Modulation) grid-connected inverters, it employs a model predictive control (MPC) algorithm to provide reference values ​​for the switching arm port voltages. During the use of MPC, the original evaluation function is improved by proposing an algorithm for evaluating the DQ-axis current model error at eight time points within each sector cycle. This invention results in a higher degree of agreement between the predicted model and the actual controlled object, and a faster convergence rate.

Description

Technical Field

[0001] This invention relates to the field of power, and specifically to a grid-connected inverter control method using model predictive vector control with an improved evaluation function. Background Technology

[0002] With the development of distributed power generation technology and intelligent control technology, Model Predictive Control (MPC) algorithms have been applied to grid-connected inverters using SVPWM (Short-Side Vector Width Modulation). However, existing MPC model error evaluation functions only use information from the start and end nodes of each sector cycle of SVPWM, neglecting the multiple changes in output current within a sector cycle. The MPC error evaluation function is a crucial basis for rolling optimization and correction of the prediction model; existing functions discard much information from intermediate processes, resulting in slow convergence of the prediction model. Improvements to the evaluation function are urgently needed. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a grid-connected inverter control method based on improved evaluation functions and model predictive vector control. Building upon a vector control (SVPWM) grid-connected inverter, it employs a model predictive control (MPC) algorithm to provide reference values ​​for the switching arm port voltages. During the use of MPC, the original evaluation function is improved by proposing an algorithm for evaluating the DQ-axis current model error at eight time points within each sector cycle.

[0004] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0005] A grid-connected inverter control method with improved evaluation function and model predictive vector control includes the following steps:

[0006] Step 1: Collect the three-phase line voltages Uab, Ubc, and Uca at the grid connection point, and collect the three-phase currents Ia, Ib, and Ic. Convert the three-phase line voltages into three-phase phase voltages Ua, Ub, and Uc using the following formulas:

[0007] Ua = (Uab – Uca) / 3

[0008] Ub=(Ubc–Uab) / 3

[0009] Uc = (Uca – Ubc) / 3

[0010] Step 2: Based on the given D-axis current reference value Idref and Q-axis current reference value Iqref, the model prediction algorithm is used to deduce the inverter bridge output port reference voltages Vpwma, Vpwmb, and Vpwmc; the improved evaluation function is used during the execution of the model prediction algorithm.

[0011] Step 3: Based on the reference voltages Vpwma, Vpwmb, and Vpwmc at the inverter bridge output port, use a vector control algorithm to determine the on / off state and time of the six switching transistors.

[0012] Step 4: The switching cycle ends and the next switching cycle begins, repeating from Step 1.

[0013] The present invention has the following beneficial effects:

[0014] 1. Compared with existing methods, the improved evaluation function of this method considers the output of the model within the sector period, making the prediction model closer to the physical model of the controlled object and the control convergence rate faster.

[0015] 2. No additional hardware is required; the algorithm can be applied simply through software upgrades, effectively reducing costs. Attached Figure Description

[0016] Figure 1 This is a flowchart of the grid-connected inverter control method based on model prediction vector control of the improved evaluation function of the present invention.

[0017] Figure 2 This is a timing diagram of the operation within one switching cycle of the present invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0019] like Figure 1 As shown, the present invention provides a grid-connected inverter control method based on model predictive vector control with an improved evaluation function, comprising the following steps:

[0020] Step 1: Collect the three-phase line voltages Uab, Ubc, and Uca at the grid connection point, and collect the three-phase currents Ia, Ib, and Ic. Then, convert the three-phase line voltages into three-phase phase voltages Ua, Ub, and Uc according to the following formulas.

[0021] Ua = (Uab – Uca) / 3

[0022] Ub=(Ubc–Uab) / 3

[0023] Uc = (Uca – Ubc) / 3

[0024] Step 2: Based on the given D-axis current reference value Idref and Q-axis current reference value Iqref, the model prediction algorithm is used to deduce the inverter bridge output port reference voltages Vpwma, Vpwmb, and Vpwmc. An improved evaluation function is involved during the execution of the model prediction algorithm.

[0025] Step 3: Based on the reference voltages Vpwma, Vpwmb, and Vpwmc at the inverter bridge output port, a vector control algorithm is used to determine the on / off state and time of the six switching transistors.

[0026] Step 4: The switching cycle ends and the next switching cycle begins, repeating from Step 1.

[0027] Specifically, the specific steps of the model prediction algorithm in step 2 include:

[0028] Step 2.1: Given the D-axis current reference value Idref and the Q-axis current reference value Iqref, after processing by the feedforward controller, the controlled object, namely the inverter and the prediction model, is simultaneously input.

[0029] Step 2.2: The controlled object generates an actual output, i.e., the grid-connected current, and the prediction model also generates a model output. These two are used as inputs to the improved evaluation function to obtain the model error. In this invention, the evaluation function is improved, making the model converge more quickly.

[0030] Step 2.3: Based on the model error, perform rolling optimization of the prediction model and modify the parameters of the feedforward controller to minimize the model error.

[0031] Step 2.4: Repeat steps 2.1-2.3 to reduce model error through rolling optimization, thereby achieving a high degree of consistency between the prediction model and the controlled object.

[0032] The improved evaluation function in step 2.2 specifically includes:

[0033] Step 2.2.1: The start time of any sector cycle is designated as T0. The end time of any sector cycle is designated as T8. The end time of this sector cycle coincides with the start time of the next sector cycle.

[0034] Step 2.2.2: As Figure 2 As shown, within any sector cycle, the on / off states of the 6 switching transistors (3 bridge arms) change 6 times. The time points of the first 3 state changes are labeled T1, T2, and T3, respectively, and the time points of the last 3 state changes are labeled T5, T6, and T7, respectively. The midpoint between the start time point T0 and the end time point T8 of this sector is labeled T4. Thus, any sector is labeled with T0 to T8 from the start time point to the end time point, a total of 9 points, denoted as Tm, dividing a sector into 8 time periods.

[0035] Step 2.2.3: Calculate the model error Gerr for the T(k) sector period using the following formula:

[0036]

[0037] i d_m (k) represents the actual output value of the D-axis current at time Tm within the period of sector T(k);

[0038] i q_m (k) represents the actual output value of the Q-axis current at time Tm within the T(k) sector period;

[0039] I * d_m (k+1) is the D-axis current model output value at time Tm of the next sector predicted by the period of sector T(k);

[0040] I * q_m (k+1) is the Q-axis current model output value at time Tm of the next sector predicted by the periodic prediction of sector T(k).

[0041] Step 2.2.4: The calculation time in step 2.2.3 is the end time T8 of each sector cycle, and the DQ axis current at each time point Tm must be measured and recorded within each sector cycle.

Claims

1. A grid-connected inverter control method with improved evaluation function model predictive vector control, characterized in that, Includes the following steps: Step 1: Collect the three-phase line voltages Uab, Ubc, and Uca at the grid connection point, and collect the three-phase currents Ia, Ib, and Ic. Convert the three-phase line voltages into three-phase phase voltages Ua, Ub, and Uc using the following formulas: Ua = (Uab – Uca) / 3; Ub = (Ubc – Uab) / 3; Uc = (Uca – Ubc) / 3; Step 2: Based on the given D-axis current reference value Idref and Q-axis current reference value Iqref, the model prediction algorithm is used to deduce the inverter bridge output port reference voltages Vpwma, Vpwmb, and Vpwmc; the improved evaluation function is used during the execution of the model prediction algorithm. The model prediction algorithm specifically includes: Step 2.1: Given the D-axis current reference value Idref and the Q-axis current reference value Iqref, after processing by the feedforward controller, the controlled object and the prediction model are simultaneously input; the controlled object is the inverter. Step 2.2: The controlled object generates the actual grid-connected current, and the prediction model also generates the model output. The grid-connected current and the model output are used as inputs to the improved evaluation function to obtain the model error. Step 2.3: Based on the model error, perform rolling optimization of the prediction model and modify the parameters of the feedforward controller to reduce the model error; Step 2.4: Repeat steps 2.1-2.3, and through rolling optimization, reduce the model error and achieve a high degree of consistency between the prediction model and the controlled object; The improved evaluation function includes: Step 2.2.1: The start time of any sector cycle is marked as T0, and the end time of any sector is marked as T8; the end time of this sector cycle coincides with the start time of the next sector cycle. Step 2.2.2: Within any sector cycle, the on / off state of the 6 switching transistors changes 6 times. The time points of the first 3 state changes are marked as T1, T2, and T3, respectively, and the time points of the last 3 state changes are marked as T5, T6, and T7, respectively. The midpoint between the start time point T0 and the end time point T8 of this sector is marked as T4. In this way, any sector is marked with T0~T8 from the start time point to the end time point, a total of 9 points, denoted as Tm, dividing a sector into 8 time periods. Step 2.2.3: Calculate the model error Gerr for the T(k) sector period using the following formula: The actual output value of the D-axis current at time Tm within the T(k) sector period; The actual output value of the Q-axis current at time Tm within the T(k) sector period; The output value of the D-axis current model at time Tm of the next sector is predicted for the period of sector T(k); The output value of the Q-axis current model at time Tm of the next sector is predicted for the period of sector T(k); Step 2.2.4: The calculation time in Step 2.2.3 is the end time T8 of each sector cycle, and the DQ axis current at each time point Tm is measured and recorded within each sector cycle; Step 3: Based on the reference voltages Vpwma, Vpwmb, and Vpwmc at the inverter bridge output port, use a vector control algorithm to determine the on / off state and time of the six switching transistors. Step 4: The switching cycle ends and the next switching cycle begins, repeating from Step 1.

Images