A control method and device of an inverter, the inverter and a medium

Abstract

The application discloses a control method and device of an inverter, the inverter and a medium. The control method is used for adjusting the output of the inverter, and comprises the following steps: collecting power information of a power grid side, wherein the power information comprises current data and voltage data of the power grid side; identifying a current working condition type used for representing a current running environment state of the inverter according to the power information; collecting a key performance indicator data set of the inverter, wherein the key performance indicator data set comprises a quantitative indicator used for evaluating the current anti-inrush control effect and a deviation amount of the quantitative indicator from a target value; and determining a control parameter set of the inverter based on the current working condition type and the key performance indicator data set. The application can improve the adaptability of the anti-inrush control parameter to different running working conditions to a certain extent.

Description

Technical Field

[0001] This application relates to the field of new energy grid connection control technology, and in particular to a control method, device, inverter and medium for an inverter. Background Technology

[0002] During the grid-connected operation of new energy equipment such as photovoltaics and energy storage, to prevent excess electricity generated by the system from being fed back into the grid, current transformers or electricity meters are typically used to collect power information such as voltage and current from the grid side. The inverter output is then adjusted based on the deviation between the grid-connected power target and the actual power. Existing anti-reverse current control typically uses pre-set control parameters, which are mostly determined manually and remain fixed during inverter operation.

[0003] However, the operating environment of inverters changes with factors such as light intensity, load power and grid conditions. Different operating conditions have different requirements for anti-reverse current control parameters. Fixed control parameters are difficult to match with the current operating conditions and the actual anti-reverse current control effect, resulting in poor adaptability of existing anti-reverse current control. Summary of the Invention

[0004] This application provides multiple embodiments of an inverter control method, device, inverter, and medium, which can improve the adaptability of anti-reverse flow control parameters to different operating conditions to a certain extent.

[0005] A first aspect of the present invention provides a control method for an inverter, the control method being used to adjust the output of the inverter, the control method comprising: Collect power information from the power grid side; the power information includes current data and voltage data from the power grid side; The current operating condition type is identified based on the power information collected from the grid side; the current operating condition type is used to characterize the current operating environment status of the inverter. Collect a dataset of key performance indicators for the inverter; the dataset includes quantitative indicators used to evaluate the effectiveness of the inverter's current anti-backflow control and the deviation from the target value. The control parameter set of the inverter is determined based on the current operating condition type and the key performance indicator dataset.

[0006] A second aspect of the present invention provides a control device for an inverter, the control device including an adaptive body, the adaptive body comprising: The acquisition submodule is used to acquire power information from the power grid side, including current data and voltage data from the power grid side. The operating condition sensing submodule is used to identify the current operating condition type based on the power information; the current operating condition type is used to characterize the current operating environment state of the inverter. The indicator monitoring submodule is used to acquire the operating data of the inverter during the anti-reverse current control process, and determine the key performance indicator dataset of the inverter based on the operating data; the key performance indicator dataset includes quantitative indicators used to evaluate the current anti-reverse current control effect of the inverter and the deviation from the target value. The parameter optimization submodule is used to determine the control parameter set of the inverter based on the current operating condition type and the key performance index dataset.

[0007] A third aspect of the present invention provides a non-transient computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method described above.

[0008] In a fourth aspect of the present invention, an inverter is provided, including a power conversion unit and a control unit. The control unit includes a control device as described above, the control device being used to adjust the output of the power conversion unit, and the control device being used to adjust the output of the power conversion unit based on a determined set of control parameters.

[0009] In this embodiment of the invention, by collecting current and voltage data from the grid side, the current operating condition type, which characterizes the current operating environment of the inverter, is identified based on the real-time current and voltage data. A dataset of key performance indicators, including quantitative indicators of the anti-reverse current control effect and their deviation from the target value, is also collected. Then, based on the current operating condition type and the dataset of key performance indicators, the control parameter set of the inverter is determined. This enables dynamic determination of control parameters according to the current operating environment of the inverter and the actual anti-reverse current control effect, thereby improving the adaptability of the anti-reverse current control to different operating conditions. Attached Figure Description

[0010] The embodiments of this application are further described below with reference to the accompanying drawings and specific implementation details.

[0011] Figure 1 This is a schematic diagram of the grid-connected system provided in the embodiments of this application.

[0012] Figure 2 This is a schematic diagram of the steps in a control method for an inverter provided in an embodiment of this application.

[0013] Figure 3 This is a schematic diagram of the control device provided in the embodiments of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the embodiments of this application.

[0015] In the description of the embodiments of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0016] In grid-connected operation scenarios of new energy equipment such as photovoltaics and energy storage, anti-reverse current control is usually required for inverters to prevent excess electricity from being fed back into the grid. Specifically, the anti-reverse current control system can collect current, voltage, or power information from the grid side in real time, and perform closed-loop adjustment based on the collected grid-side power information to control the inverter output, keeping the power at the grid connection point within the allowable range and reducing or preventing electricity from being fed back into the grid.

[0017] However, the operating conditions of inverters in actual operation are usually constantly changing. For example, light intensity may change abruptly, load power may fluctuate, and the grid conditions may also change. Under different operating conditions, the inverter's operating environment varies, and the requirements for the response speed, stability, and adjustment accuracy of anti-reverse current control also differ. If the anti-reverse current control still uses preset fixed control parameters, it may lead to a mismatch between the control parameters and the current operating conditions. For example, when the operating conditions change rapidly, insufficient control parameter response may lead to lag in reverse current suppression; when the operating conditions are relatively stable, excessively strong control parameters may cause output fluctuations or oscillations.

[0018] Furthermore, existing anti-reverse current control typically relies primarily on grid-side power feedback for adjustment, rarely incorporating quantitative indicators of the inverter's current anti-reverse current control effectiveness to dynamically determine control parameters. In other words, even if the inverter exhibits poor anti-reverse current control performance, such as long settling time, large overshoot, and high steady-state error, the system struggles to promptly re-determine control parameters adapted to the current state based on these key performance indicators, resulting in insufficient adaptability and effectiveness of the anti-reverse current control.

[0019] In this embodiment, the inverter control method is used to adjust the inverter's output so that the power exchange between the grid-connected system where the inverter is located and the grid meets the backflow prevention control requirements. For example... Figure 1 As shown, a grid-connected system may include photovoltaic (PV) generators, a battery, a hybrid inverter, a load, a grid, and a meter. The PV generators and battery can be connected to the DC side of the inverter, while the load and grid can be connected to the AC side. The meter can be installed between the grid-connected system and the grid to detect current, voltage, or power on the grid side. The inverter can be a hybrid inverter or a power storage converter (PCS).

[0020] In this embodiment, reverse current prevention refers to limiting the reverse transmission of electrical energy generated by the grid-connected system to the grid, ensuring that the reverse transmission power does not exceed the allowable value. The target power for reverse current prevention can be zero, or a non-zero power value or allowable power range set according to grid connection requirements. The inverter can regulate the power exchange between the grid-connected system and the grid by adjusting at least one of the photovoltaic side power, energy storage side power, and AC side output power.

[0021] Figure 2 This is a schematic diagram illustrating the steps of a control method for an inverter provided in an embodiment of this application. For example... Figure 2 As shown, this control method is used to regulate the output of the inverter, and the control method includes the following steps: Step S1: Collect power information from the power grid side; the power information includes current data and voltage data from the power grid side; Step S2: Identify the current operating condition type based on the power information collected from the grid side; the current operating condition type is used to characterize the current operating environment status of the inverter; Step S3: Collect the key performance index dataset of the inverter; the key performance index dataset includes quantitative indicators used to evaluate the current anti-reverse current control effect of the inverter and the deviation of the quantitative indicators from the target values; Step S4: Determine the control parameter set of the inverter based on the current operating condition type and the key performance index dataset.

[0022] Specifically, power information from the grid side is collected. This grid-side power information characterizes the energy exchange status between the grid-connected system and the grid, including current and voltage data from the grid side. Current and voltage data can be obtained through current transformers (CTs), electricity meters (Meters), or electrical quantity acquisition circuits within the inverter, all located on the grid side. The term "acquisition" here refers to continuously or periodically acquiring real-time current and voltage data according to a preset sampling period during inverter operation, without requiring absolute temporal continuity. Current data may include instantaneous values, effective values, or sampling sequences of the current in each phase on the grid side, and voltage data may include instantaneous values, effective values, or sampling sequences of the voltage in each phase on the grid side. Based on the current and voltage data, one or more of the following can be determined: active power, reactive power, power change, power change rate, voltage fluctuation, current fluctuation, grid frequency, or harmonic distortion rate. Taking active power as an example, the active power on the grid side can be determined based on real-time current and voltage data within the same sampling time or the same sampling period, and the trend of grid-side power change can be determined based on the active power on the grid side over multiple consecutive sampling periods. The sampling frequency of grid-side voltage and current can be determined based on the rate of change of electrical quantities and the accuracy of operating condition identification. For example, the sampling frequency can be 10kHz, 15kHz, 20kHz, 30kHz, or 50kHz, or other sampling frequencies not lower than 10kHz. The above values ​​are only used to illustrate one method of electrical quantity acquisition and do not constitute a limitation on the sampling frequency.

[0023] The current operating condition type is identified based on the collected power information from the grid side. The current operating condition type characterizes the current operating environment state of the inverter and reflects the impact of changes in photovoltaic output, load power, or grid conditions on the inverter's operating status. The current operating condition type is not limited to a fixed classification method and can be determined based on at least one of the following: the magnitude, amplitude, and rate of change of grid-side power, and the magnitude or fluctuation of grid-side voltage. In addition to grid-side power information, one or more of the following can be collected: output power of each controllable port of the inverter, energy storage battery status, environmental data, and equipment status signals. This collected data is used to assist in identifying the current operating condition type. The output power of each controllable port of the inverter can include photovoltaic MPPT power, energy storage battery charging and discharging power, load power, and inverter output power. Energy storage battery status can include State of Charge (SOC), allowable charging power, allowable discharging power, or full charge limit power state. Environmental data can include irradiance, temperature, or wind speed. Equipment status signals can include inverter start / stop status, relay position, or fault codes.

[0024] Time-series features can be extracted from continuously collected operational data. These features may include one or more of the following: power change rate dP / dt, voltage fluctuation standard deviation σ_V, load fluctuation amplitude, or load periodicity. For example, load periodicity can be used to characterize day-night peak-valley patterns, and power change rate can be used to characterize the rate of change of photovoltaic power, load power, or grid-side power.

[0025] Preferably, before identifying the current operating condition type, the collected data can be cleaned and feature-processed. Data cleaning may include reducing noise through moving average filtering and removing outliers caused by current transformer saturation, communication anomalies, or detection device malfunctions. Feature processing may include normalization, statistical processing, and feature vector construction. For example, feature vectors for operating condition identification can be formed based on one or more of the following: rate of change of illumination, load fluctuation amplitude, grid harmonic distortion rate, power change rate, voltage fluctuation degree, energy storage battery state of charge, and equipment operating status.

[0026] The current operating condition type can also be identified through preset judgment rules. For example, the current operating condition type can be identified based on whether the rate of change of illumination exceeds a preset threshold for illumination change, whether the load fluctuation amplitude exceeds a preset threshold for load fluctuation, whether the grid harmonic distortion rate exceeds a preset threshold for harmonics, or whether the grid voltage exceeds a preset voltage boundary.

[0027] Lightweight machine learning models can also be used to identify the current operating condition type. These lightweight machine learning models are those with a low parameter count, low storage resource consumption, and low computational resource consumption suitable for online operation of inverters, PCS, EMS, or edge computing devices. The lightweight machine learning model can be pre-trained based on sample data under different operating conditions. The sample data can include grid-side power and voltage collected over different time periods, and corresponding operating condition labels. After training, real-time collected grid-side power and voltage, or features extracted based on the grid-side power and voltage, can be input into the lightweight machine learning model, which then outputs the current operating condition type. Lightweight machine learning models can employ random forest models, LSTM models, lightweight XGBoost models, one-dimensional convolutional neural network models, or other models capable of classification based on time-series electrical data. Lightweight XGBoost models can reduce computational cost by limiting the number of decision trees, decision tree depth, or feature dimensions; one-dimensional convolutional neural network models can reduce model size by reducing the number of convolutional layers, convolutional kernels, or channels. The above models are alternative implementation methods.

[0028] As an example, a lightweight machine learning model is used to identify the current operating condition type in real time. The specific implementation includes the following steps: First, collected power and voltage data from the grid side are used as raw input. Before being fed into the lightweight machine learning model, the data is cleaned and feature-engineered to extract key feature vectors reflecting the system's operating state. These feature vectors include, but are not limited to: grid-side active power, reactive power, effective voltage value, power change rate, voltage fluctuation standard deviation, and frequency domain energy distribution extracted by short-time Fourier transform (STFT) or wavelet transform. When the system is equipped with environmental sensors, non-electrical features such as light intensity and temperature can also be incorporated as model input. A sliding time window is applied to the continuous feature data to form the time dimension of the model input samples. Each time window contains feature sequences from the current moment and multiple historical continuous sampling points. The window length is set according to the dynamic characteristics of the operating condition changes, generally covering hundreds of milliseconds to several seconds. The window sliding step size is synchronized with the operating cycle of the adaptive computing unit (e.g., ≤100ms) to ensure the real-time nature of operating condition identification. The feature sequences within each time window are then input into the pre-trained lightweight machine learning model. The model has learned data distribution patterns under different operating conditions through offline training, enabling it to directly map input samples to discrete operating condition category labels. The model's output classification results include at least broad categories such as steady-state, transient, and extreme conditions, further expanded into multiple subcategories, such as twelve specific operating conditions including "PV cloud shading transient," "energy storage full-charge power limitation," "load step transient," and "grid frequency anomaly." The output format is the current operating condition label and its corresponding confidence score. The confidence score can be obtained based on the class probability, class score, or ensemble voting ratio output by the model; normally, a confidence score ≥95% is required. When the highest confidence score output by the model is lower than a preset threshold (e.g., 95%), it indicates that the current operating state is at the boundary of the operating condition or that an unknown pattern not included in the training set has appeared. At this point, the system does not directly adopt the low-confidence classification result, but instead initiates a conservative safety strategy: for the operating condition label, it reverts to the operating condition type confirmed with high confidence in the previous frame, or adopts the control parameters corresponding to the preset safe operating condition; if the grid-side power, current, or voltage is simultaneously detected to exceed the allowable physical limits or safety specification boundaries, it is preferentially identified as an extreme operating condition or the corresponding safety protection strategy is triggered; simultaneously, this identification is marked as a low-confidence event in the safety verification layer, and additional parameter stability constraints are added (such as tightening the parameter adjustment range and reducing the parameter update step size) until the operating condition identification returns to a high-confidence state. Furthermore, the low-confidence event will be logged as an anomaly for subsequent model iterations and rule optimization.

[0029] In some implementations, the current operating condition type may include one of steady-state operating condition, transient operating condition, and extreme operating condition, or it may be a composite operating condition label formed by combining the steady-state operating condition, transient operating condition, or extreme operating condition with corresponding sub-operating conditions. A steady-state operating condition is when the grid-side power and voltage remain within a preset fluctuation range for a preset time; a transient operating condition is when at least one of the grid-side power and voltage exceeds the preset fluctuation range, but has not yet exceeded the permissible physical limit or safety specification boundary; an extreme operating condition is when at least one of the grid-side power and voltage exceeds the permissible physical limit or safety specification boundary.

[0030] The preset time can be set according to the inverter's control cycle, the rate of change of the grid state, and the accuracy of operating condition identification. For example, the preset time can be 0.1 seconds, 0.2 seconds, 0.5 seconds, 1 second, 2 seconds, 5 seconds, or 10 seconds, or any other time between 0.1 seconds and 10 seconds.

[0031] The preset fluctuation range can be expressed as a percentage change relative to a reference value. For example, the percentage change in grid-side power relative to the average power over a preset time period can be between -5% and 5%, and the percentage change in grid-side voltage relative to the rated voltage can be between -5% and 5%, which can be used as the preset fluctuation range corresponding to steady-state operating conditions. The boundaries of the preset fluctuation range can also be ±1%, ±2%, ±3%, ±4%, ±5%, ±8%, or ±10%.

[0032] Operating conditions with fluctuations less than 5% can be identified as steady-state conditions, those with fluctuations between 5% and 30% as transient conditions, and those with fluctuations greater than 30% or those experiencing grid faults as extreme conditions. For example, the fluctuation percentages corresponding to transient conditions could be 5%, 10%, 15%, 20%, 25%, or 30%. The above values ​​are only used to illustrate one method of classifying operating conditions.

[0033] Permissible physical limits or safety specification boundaries may include maximum permissible grid-side power, maximum permissible grid-side current, upper voltage limit, lower voltage limit, or permissible voltage fluctuation boundaries. For example, the upper voltage limit may be 105%, 110%, 115%, 120%, or 125% of the rated voltage, and the lower voltage limit may be 95%, 90%, 85%, 80%, or 75% of the rated voltage.

[0034] When the power and voltage on the grid side exhibit large fluctuations and exceed the allowable physical limits or safety specification boundaries, the current operating condition can be preferentially identified as an extreme operating condition; when the power and voltage on the grid side do not exceed the allowable physical limits or safety specification boundaries, but exceed the preset fluctuation range, it can be identified as a transient operating condition; when the power and voltage on the grid side remain within the preset fluctuation range, it can be identified as a steady-state operating condition.

[0035] Sub-conditions can be further set based on steady-state, transient, and extreme operating conditions. For example, sub-conditions may include sudden increases in solar irradiance, transient photovoltaic cloud shading, sudden load increases, sudden load decreases, weak grid, full-charge power limitation of energy storage, grid voltage dips, or grid faults. Different implementations may set up twelve sub-conditions or other numbers of sub-conditions.

[0036] The operating condition identification result may also include a confidence level corresponding to the current operating condition type. The confidence level is used to characterize the reliability of the operating condition identification result. For example, the confidence level can be 90%, 92%, 95%, 97%, or 99%, or other values ​​not lower than 95%. When the confidence level is lower than the preset confidence threshold, the operating condition type can be maintained, operating data can be re-acquired, or the control parameters corresponding to the preset safe operating condition can be used.

[0037] A dataset of key performance indicators (KPIs) for the inverter is collected. This KPI dataset refers to a collection of data used to quantitatively evaluate the current anti-reverse control effect of the inverter. It includes at least one quantitative indicator for assessing the current anti-reverse control effect, and the deviation between this quantitative indicator and its corresponding target value. The quantitative indicator is used to convert the anti-reverse control effect into comparable and calculable data, and the target value represents the expected control level achieved by the corresponding quantitative indicator. The specific content of the data collection may include collecting the inverter's operating data and calculating, statistically analyzing, or estimating the corresponding quantitative indicators based on the operating data.

[0038] In some implementations, the key performance indicator (KPI) dataset may include stability indicators, dynamic performance indicators, and economic indicators. The actual quantified values ​​of each indicator can be obtained separately, and these values ​​can be compared with their corresponding target values ​​to determine the deviation. The KPI dataset may also include the real-time values, historical trends, and deviations from the target values ​​for each indicator.

[0039] Stability metrics may include one or more of the following: phase margin and gain margin of the anti-reverse current control loop, oscillation amplitude and attenuation rate of the inverter output current, and oscillation amplitude and attenuation rate of the inverter output power. Phase margin characterizes the allowable increase in phase lag of the anti-reverse current control loop before reaching a critical stable state, while gain margin characterizes the allowable increase in loop gain of the anti-reverse current control loop before reaching a critical stable state.

[0040] Phase margin and gain margin can be obtained by real-time frequency sweeping of the anti-reverse control loop, or by estimation using an observer based on the control model corresponding to the inverter. When the actual phase margin or actual gain margin is less than the corresponding target value, it indicates that the stability margin corresponding to the current control parameter set is insufficient.

[0041] The inverter output current can be continuously acquired within a preset monitoring time, and the oscillation amplitude and attenuation rate of the inverter output current can be determined based on the peak, valley, effective value, or amplitude changes of adjacent oscillation cycles. The inverter output power can also be continuously acquired, and the oscillation amplitude and attenuation rate of the inverter output power can be determined. When the oscillation amplitude of the output current or output power is large, or when the oscillation amplitude fails to decrease effectively over time, it indicates that the control strength corresponding to the current control parameter set is too high or the control stability is insufficient.

[0042] Dynamic performance indicators may include one or more of the following: reverse current adjustment time, reverse current overshoot, and steady-state error. The reverse current adjustment time is the time from the occurrence of reverse current to its recovery to normal. It can be determined that reverse current has occurred when the deviation between the grid-side power and the anti-reverse current target power exceeds a preset reverse current threshold, and that recovery to normal is determined when the deviation returns to the allowable error range. The reverse current adjustment time is determined based on the time interval between these two moments. Specifically, the grid-side power can be denoted as P_grid(t), the anti-reverse current target power as P_ref, and the power deviation e(t) can be determined according to preset power direction and sign rules. The reverse current start time T1 is when the power deviation e(t) exceeds the reverse current threshold P_th, for example, P_th can be 0.5%×P_rated, where P_rated is the rated active power of the inverter on the AC side or the reference power predetermined according to the grid connection point capacity; the reverse current recovery time T2 is when the power deviation e(t) returns to the allowable error band (e.g., ±2%×P_rated) and remains stable for more than a preset time (e.g., 20ms). The reverse current adjustment time t_s is calculated as: t_s=T2-T1.

[0043] Reverse current overshoot refers to the percentage deviation of the maximum reverse current power from the target regulation value during reverse current regulation. The reverse current power on the grid side can be continuously acquired during a single reverse current regulation process to determine the maximum reverse current flow, and the percentage deviation is determined based on the difference between the maximum reverse current flow and the target regulation value. When the target regulation value is zero or close to zero, a preset reference power, the inverter's rated power, or the allowable reverse current limit can be used as a normalization reference.

[0044] The steady-state error e_ss is the average deviation between the grid-side power and the anti-reverse current target power within a preset statistical time after the anti-reverse current regulation process enters a steady state. After the anti-reverse current regulation process enters a steady state, power data from the grid side can be continuously collected within a preset statistical time, and the power deviation between the grid-side power and the anti-reverse current target power can be determined. This power deviation is then averaged to obtain the steady-state error.

[0045] Economic indicators can include one or more of the following: solar curtailment rate, energy storage curtailment rate, and inverter lifetime losses. The solar curtailment rate is determined based on the maximum usable power output of a photovoltaic power generation unit under current irradiance conditions and the actual output power limited by anti-reverse flow control. The energy storage curtailment rate can be used to characterize the loss of energy storage availability caused by anti-reverse flow control.

[0046] Inverter lifetime loss can be calculated using a junction temperature estimation model. The current, voltage, switching frequency, ambient temperature, or heat dissipation status of the power devices in the inverter can be obtained. This data is then input into the junction temperature estimation model to obtain the junction temperature or junction temperature variation of the power devices. The inverter lifetime loss can then be estimated based on the junction temperature, junction temperature fluctuation amplitude, or number of thermal cycles. The power devices can be insulated-gate bipolar transistors or other power switching devices used to perform power conversion.

[0047] Stability, dynamic performance, and economic indicators can be compared with their corresponding target values ​​to obtain stability deviation, dynamic performance deviation, and economic deviation, respectively. These deviations are then incorporated into the key performance indicator dataset. For example, under the current operating condition of rapidly increasing light intensity, if both the reverse current settling time and reverse current overshoot exceed their corresponding target values, but the inverter output power oscillation amplitude remains within acceptable limits, the priority of dynamic performance indicators can be increased during subsequent parameter optimization. Conversely, if the reverse current settling time meets the target value, but the output current oscillation amplitude or inverter lifespan loss exceeds the corresponding target value, the priority of stability or economic indicators can be increased.

[0048] The control parameter set of the inverter is determined based on the current operating condition type and key performance indicator dataset. The control parameter set refers to a set of parameters used by the inverter's control algorithm to influence the inverter's output regulation process. In one embodiment, the control parameter set is the parameter set used by the power scheduling algorithm in the anti-reverse current control loop to generate power commands for each controllable port based on the grid-side power deviation. The control parameter set may include control gain parameters, upper limit of power command output, lower limit of power command output, control parameter change rate threshold, power command change rate limit, or other parameters that can change the inverter's output response characteristics.

[0049] A mapping library between operating conditions and parameters can be pre-established. This mapping library can be built through offline simulation and field testing, and is used to store recommended control parameter ranges, output upper limits, output lower limits, and power command change rate limits corresponding to different current operating condition types. After identifying the current operating condition type, the corresponding candidate control parameters or candidate parameter ranges can be retrieved from the mapping library to tailor the parameter space for online optimization.

[0050] The control parameters of the inverter may include at least one of the following: proportional gain Kp, integral gain Ki, and derivative gain Kd. The proportional gain Kp characterizes the adjustment gain coefficient proportional to the deviation between the current grid-side power and the anti-reverse current target power; the integral gain Ki characterizes the adjustment gain coefficient proportional to the integral of the deviation over time; and the derivative gain Kd characterizes the adjustment gain coefficient proportional to the rate of change of the deviation over time. For example, the recommended range for the proportional gain Kp under steady-state conditions is 0.8 to 1.2, the recommended range for the integral gain Ki is 0.05 to 0.1, and the power command change rate is limited to 10% / s; the recommended range for the proportional gain Kp under transient conditions is 1.5 to 2.0, the recommended range for the integral gain Ki is 0.1 to 0.2, and the power command change rate is limited to 20% / s; and the recommended range for the proportional gain Kp under extreme conditions is 0.5 to 0.8, the recommended range for the integral gain Ki is 0.02 to 0.05, and the power command change rate is limited to 5% / s. The parameter ranges and values ​​mentioned above are for illustrative purposes only and do not constitute a limitation on the values ​​of the control parameters.

[0051] In some implementations, the current grid-side power can be denoted as P_grid, the anti-reverse current target power as P_ref, and the deviation value e(t) can be determined according to preset power direction and sign rules. For example, the direction of power transmission from the grid-connected system to the grid can be defined as the positive direction, and the deviation value can be defined as e(t) = P_ref - P_grid(t); when the opposite power direction definition is used, the deviation value and the sign relationship of the control output are adjusted accordingly. The control output of the anti-reverse current controller can be determined according to the following relationship: u(t)=Kp·e(t)+Ki·∫e(t)dt+Kd·de(t) / dt.

[0052] Where u(t) represents the control quantity used to adjust the inverter output. The above relationship is used to explain the control action of the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd. In practice, discrete calculation methods can be used, and only one or more of the control actions can be used.

[0053] The proportional gain Kp primarily influences the instantaneous response of the anti-backflow controller to the current deviation value. The integral gain Ki is mainly used to reduce the persistent steady-state error. The derivative gain Kd can adjust its adjustment based on the trend of the deviation value to suppress overshoot or oscillation. The basic power control loop can use proportional-integral control, with the derivative gain Kd as an optional control parameter.

[0054] The control parameter set may include the controller's output upper limit, output lower limit, control parameter change rate threshold, and power command change rate limit. The output upper limit and output lower limit are used to limit the allowable range of power commands generated by the power scheduling algorithm, the parameter change rate threshold is used to limit the maximum change of control parameters between adjacent control cycles or within a unit of time, and the power command change rate limit is used to limit the maximum change of power commands generated based on control parameters between adjacent control cycles or within a unit of time.

[0055] The upper and lower limits of output, the threshold for the rate of change of control parameters, and the limit for the rate of change of power command can be determined based on the current operating condition. Under steady-state conditions, output boundaries and parameter rate of change thresholds suitable for fine-tuning can be adopted to reduce unnecessary output fluctuations. Under transient conditions, the output boundaries can be adjusted according to dynamic response requirements, and the allowable rate of change of control parameters and power commands can be increased. Under extreme conditions, the output range and rate of change can be limited according to the physical limits of the equipment or safety specifications. The specific adjustment direction and magnitude of the boundaries under different operating conditions can be determined according to the inverter's rated parameters and safety control requirements. When extreme operating conditions trigger the inverter's current limiting, fault ride-through, shutdown, or grid disconnection conditions, the corresponding safety protection strategy is executed first, and online control parameter optimization is suspended. Control parameters are only adjusted within the preset safety range when the inverter is allowed to continue operating in grid-connected mode.

[0056] Objective functions and constraints for control parameter optimization can be constructed based on a dataset of key performance indicators. The objective function quantifies the comprehensive control effect corresponding to different candidate control parameters, while the constraints limit the feasible range of the candidate control parameters, ensuring that the solved control parameters meet the inverter's physical capabilities, anti-reverse current control requirements, and grid connection safety requirements.

[0057] The objective function can include one or more of the following: reverse current regulation time, reverse current overshoot, steady-state error, inverter output power oscillation amplitude, solar curtailment rate, solar storage curtailment rate, or inverter lifetime loss. For quantitative indicators that are expected to be reduced, the corresponding actual value or deviation can be used as the target to be minimized; for stability margin indicators that are expected to be increased, their insufficiency relative to the target value can be used as the target to be minimized. Weights can be assigned to multiple quantitative indicators, and a comprehensive objective function can be constructed based on each quantitative indicator and its weight. For example, the objective function can be constructed according to the following relationship: .

[0058] in, Represents the objective function value. , and These represent evaluation items determined based on different quantitative indicators. , and This indicates the corresponding weight. In one implementation, , and These are the normalized countercurrent regulation time evaluation item, the countercurrent overshoot evaluation item, and the new energy output loss evaluation item. , and The sum can be 1, for example, it can be 0.5, 0.3 and 0.2 respectively, or 0.4, 0.4 and 0.2 respectively, or 0.3, 0.3 and 0.4 respectively.

[0059] The objective function can be obtained by weighted optimization of the reverse current regulation time, reverse current overshoot, and renewable energy output loss. Renewable energy output loss refers to the renewable energy power sacrificed due to reverse current prevention, which can be determined based on the difference between the available power of the renewable energy generation unit under current operating conditions and the actual output power after implementing reverse current prevention control. The reverse current regulation time, reverse current overshoot, and renewable energy output loss can be normalized separately to allow indicators with different dimensions to be weighted within the same objective function. Each weight can be dynamically set based on the current operating condition, user control requirements, or the deviation of each indicator from the target value. For example, under transient conditions, the weights corresponding to the reverse current regulation time and reverse current overshoot can be increased; under steady-state conditions, the weights corresponding to renewable energy output loss can be increased. Constraints can include one or more of the following: control parameter constraints, control output constraints, equipment operation constraints, and grid operation constraints. Control parameter constraints can be used to limit the value range of the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd; control output constraints can include upper and lower output limits; equipment operation constraints can include the inverter's maximum allowable current, maximum allowable power, the allowable charge / discharge power of the energy storage battery, and the available power of the photovoltaic power generation unit; grid operation constraints can include voltage, current, or power boundaries. The maximum allowable current can be set to 100%, 105%, 110%, 115%, or 120% of the rated current. For example, the maximum allowable current can be limited to less than 1.1 times the rated current. Specific constraint values ​​can be set according to the inverter's rated parameters, allowable overload capacity, and grid connection requirements.

[0060] Particle swarm optimization, Bayesian optimization, model predictive control, or other constrained optimization algorithms can be used to solve for the optimal values ​​of the control parameters within a parameter space limited by constraints. During the solution process, at least one of the proportional coefficient Kp, integral coefficient Ki, derivative coefficient Kd, upper output limit, lower output limit, and parameter rate of change threshold can be used as the variable to be optimized.

[0061] Preferably, an improved particle swarm optimization algorithm can be used to solve for the optimized values ​​of the control parameters. The control parameters to be optimized can be encoded as particle positions, with each particle corresponding to a set of candidate control parameters. The particle's position dimension corresponds to the type of control parameter to be optimized, such as the position vector dimension corresponding to the type of control parameter to be optimized, including at least the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd, and can also be extended to the upper limit of output, the lower limit of output, and the power command change rate limit. Multiple particles can be initialized within the candidate value range corresponding to the current operating condition type, and an initial position and initial velocity can be set for each particle. The initialization of the particle swarm is not a random scattering of points in the full parameter space, but rather the generation of an initial population within a recommended interval defined by the parameter mapping library corresponding to the current operating condition type. The parameter mapping library pre-stores the correspondence between operating conditions and control parameters, and is constructed through offline statistical analysis of historical data or expert experience. This approach can significantly compress the search space, accelerate convergence, and avoid obvious infeasible regions. In each iteration, candidate control parameters corresponding to the particle's current position and its current operating state can be input into the control model corresponding to the anti-backflow control loop. The anti-backflow control performance using the candidate control parameters is predicted based on the control model, and the corresponding key performance indicators are substituted into the objective function to obtain the fitness value for that particle. These key performance indicators may include one or more of the following: backflow adjustment time, backflow overshoot, steady-state error, oscillation amplitude, or new energy output loss. The individual optimal position obtained by each particle during the iteration process can be recorded, and the swarm optimal position can be determined from the individual optimal positions of multiple particles. Subsequently, the particle's velocity and position can be updated based on its current position, current velocity, individual optimal position, and swarm optimal position.

[0062] An improved particle swarm optimization algorithm can introduce an adaptive decay mechanism for inertia weights. The inertia weights can gradually decrease as the iteration process progresses, giving the particle swarm strong global search capabilities in the early stages of iteration and strong local convergence capabilities in the later stages. The inertia weights can be adjusted based on the relationship between the current iteration number and the maximum iteration number, or based on changes in the particle swarm's fitness value. When the updated particle position exceeds the allowable range of the control parameters, the corresponding control parameters can be restricted to the corresponding boundary, or the particle can be returned to the feasible parameter space defined by the constraints. For particles that do not meet equipment operation constraints or grid connection safety constraints, a penalty term can be added to their objective function value.

[0063] The iterative process of the improved particle swarm optimization algorithm can be terminated when the maximum number of iterations is reached, the objective function value is less than the preset objective function threshold, the change in the optimal fitness of the population between adjacent iterations is less than the preset convergence threshold, or the preset solution time is reached. The control parameter corresponding to the current optimal position of the population is then determined as the optimized value of the control parameter.

[0064] In some implementations, the overall running period of the adaptive body can be 20ms, 40ms, 60ms, 80ms, or 100ms, or other periods not exceeding 100ms. The single optimization time of the improved particle swarm optimization algorithm is less than or equal to the parameter update period of the adaptive body. The specific solution time can be determined based on the processor's computing power, the number of particles, the maximum number of iterations, and the evaluation method of candidate control parameters.

[0065] For example, when the current operating condition is a transient condition corresponding to a sudden increase in illumination, and the key performance index dataset indicates that the current countercurrent adjustment time is 300ms, the target countercurrent adjustment time is no more than 150ms, the current countercurrent overshoot is 8%, and the target countercurrent overshoot is no more than 5%, the particle swarm can be initialized within the candidate value range of proportional coefficient Kp from 1.5 to 2.0 and integral coefficient Ki from 0.1 to 0.2. After multiple rounds of velocity updates, position updates, and fitness comparisons, optimized control parameter values ​​with a proportional coefficient Kp of 1.92 and an integral coefficient Ki of 0.18 can be obtained.

[0066] After obtaining the optimized control parameters, linear interpolation can be performed on the control parameters before and after optimization to determine the inverter's control parameter set. The control parameters before optimization are the control parameters currently used by the inverter, while the optimized control parameters are the control parameters obtained by solving the objective function and constraints. Through linear interpolation, one or more intermediate control parameters can be formed between the control parameters before and after optimization, allowing the control parameters to gradually transition to the optimized values.

[0067] For any control parameter in the control parameter set, the interpolated control parameter can be determined according to the following relationship: P_i=P_before+λ_i×(P_after-P_before).

[0068] Where P_before represents the control parameters before optimization, P_after represents the control parameters after optimization, P_i represents the control parameters at the i-th interpolation time, and λ_i represents the interpolation coefficients. The interpolation coefficients can be 0, 0.25, 0.5, 0.75, and 1, or 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1, respectively.

[0069] Linear interpolation can be completed within a preset transition time of 20ms, 40ms, 60ms, 80ms, 100ms, 150ms, or 200ms. The preset transition time can be determined based on the difference in control parameters before and after optimization, the parameter change rate threshold, and the inverter's dynamic response capability. The change in control parameters between adjacent interpolation moments should satisfy the control parameter change rate threshold; when the change in control parameters obtained according to the preset interpolation coefficients exceeds the control parameter change rate threshold, the number of interpolation points can be increased or the preset transition time can be extended.

[0070] In this embodiment, after obtaining the updated control parameters based on the current operating condition type and key performance indicator dataset, a safety check can be performed on the updated control parameters. The safety check is used to determine whether the updated control parameters meet the physical operating requirements of the inverter and the stability requirements of the anti-reverse current control loop before being written into the anti-reverse current controller or before they officially take effect.

[0071] Physical limit hard constraints can be performed on the updated control parameters. Physical limit hard constraints refer to the restrictions that the updated control parameters must not exceed, pre-set based on the inverter's hardware capabilities, the rated parameters of the power devices, the charging and discharging capabilities of the energy storage device, and the controller's allowable configuration range. Physical limit hard constraints may include one or more of the following: the value ranges of the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd; the upper and lower limits of the controller output; the parameter change rate threshold; and the inverter's maximum allowable current and maximum allowable power. The allowable value range of the proportional coefficient Kp can be set to be greater than or equal to 0 and less than or equal to 5, and the integral coefficient Ki can be set to be greater than or equal to 0. The above values ​​are examples; the specific value ranges can be determined based on the anti-backflow control loop model and equipment parameters.

[0072] The updated control parameters can also be used to predict their stability. The updated control parameters can be substituted into the control model corresponding to the anti-reverse current control loop, and frequency domain stability analysis or time domain stability analysis can be performed on the control model. The predicted phase margin, predicted gain margin, or closed-loop stability can be obtained using the Nyquist criterion; or the convergence trend and whether the system state deviation has converged can be determined using the Lyapunov function. The predicted values ​​for reverse current overshoot, reverse current adjustment time, output current oscillation amplitude, or output power oscillation amplitude can be obtained from the time domain response of the control model. When the predicted quantized value of the index exceeds the corresponding preset range, it can be determined that the updated control parameters lack stability.

[0073] The Nyquist criterion can be used to perform frequency domain stability analysis on the anti-backflow control loop after adopting updated control parameters, obtaining the Nyquist curve, phase margin prediction value, or gain margin prediction value. For example, the preset range of phase margin can be 30 to 60 degrees, including 30, 35, 40, 45, 50, 55, and 60 degrees. When the predicted phase margin is lower than 30 degrees, it can be determined that the updated control parameters lack stability. The Lyapunov function can be used to perform time domain stability analysis on the anti-backflow control loop after adopting updated control parameters, and the rate of change of the Lyapunov function can be used to determine whether the state deviation converges and its convergence speed. When the Lyapunov function fails to meet the preset convergence condition, or when the oscillation amplitude obtained based on the control model does not decrease or continues to increase, it can be determined that the updated control parameters lack stability. After the updated control parameters pass the physical limit hard constraint check and stability prediction, they can be written into the temporary parameter area and subjected to a limited trial run for a preset duration under conditions that restrict the inverter's output power, output current, and the variation range of control parameters. If the key performance indicators obtained during the limited trial run meet the preset requirements, the updated control parameters will officially take effect; otherwise, they will be rolled back to the previous control parameters.

[0074] In some implementations, a rollback to prior control parameters can be triggered if the updated control parameters are detected to fail the physical limit hard constraint check or lack stability. Prior control parameters refer to those used by the inverter prior to this update and determined to meet safe operation requirements. Prior control parameters can be stored in a preset storage area before the update.

[0075] When the updated control parameters have not yet taken effect, rollback may include stopping the issuance or discarding of the updated control parameters and continuing to use the previous control parameters; when the updated control parameters have been written to the temporary parameter area or have taken effect in part of the control cycle, rollback may include rewriting the previous control parameters into the anti-backflow controller.

[0076] The system can continue to monitor the key performance indicator dataset after the updated control parameters take effect. A rollback can be triggered when the reverse current overshoot, oscillation amplitude, or other quantitative indicators deteriorate compared to before the update. For example, a rollback can be triggered when the reverse current overshoot increases by more than a preset percentage, such as 10%, 20%, 30%, 40%, or 50%, relative to the baseline reverse current overshoot under the corresponding operating condition. In other embodiments, a rollback can be triggered when the reverse current overshoot increases by 30% relative to the baseline reverse current overshoot.

[0077] In some implementations, when a rollback is triggered, the current operating condition type, key performance indicator dataset, updated control parameters, types of failed safety checks, and corresponding predicted quantitative values ​​of the indicators can also be recorded. The recorded information can be used to adjust the mapping library between operating conditions and parameters, the parameter optimization range, or constraints to avoid repeatedly generating control parameters that do not meet safety requirements under the same operating conditions.

[0078] An adaptive anti-reverse current control method can operate according to a closed-loop process of perception, decision-making, and execution. First, it collects grid-side voltage, current, power, status of each controllable port, environmental data, and equipment status signals, and extracts corresponding features. Then, it identifies the current operating condition type and obtains a dataset of key performance indicators. Next, it determines updated control parameters based on the current operating condition type and the dataset of key performance indicators, and performs safety verification and smooth transition. Finally, it provides the control parameters that have passed the safety verification to the anti-reverse current control loop, and continues to collect updated operating data and key performance indicators to enter the next control cycle.

[0079] For example, when a 200 W / m² increase in light intensity is detected within 0.5 seconds, the current operating condition can be identified as a transient condition corresponding to a sudden increase in light intensity. If the current countercurrent settling time is 300 ms and the target countercurrent settling time is no more than 150 ms, and the current countercurrent overshoot is 8% and the target countercurrent overshoot is no more than 5%, then the corresponding candidate parameter range can be called, and the proportional coefficient Kp and integral coefficient Ki can be optimized to 1.92 and 0.18, respectively, using an improved particle swarm optimization algorithm. After the updated control parameters pass physical limit hard constraint checks and stability predictions, the control parameters can be smoothly switched to the optimized values ​​through linear interpolation, and the countercurrent settling time and countercurrent overshoot can be obtained again in subsequent control cycles to evaluate the control parameter update results.

[0080] In this embodiment, a control device for an inverter is also provided. The control device includes an adaptive unit, which comprises a data acquisition submodule, a condition sensing submodule, an indicator monitoring submodule, and a parameter optimization submodule. The output of the data acquisition submodule can be provided to the condition sensing submodule, the output of the condition sensing submodule can be provided to the indicator monitoring submodule, and the output of the indicator monitoring submodule can be provided to the parameter optimization submodule. The modules can exchange data via an internal data bus, shared storage area, inter-process communication interface, or message transmission interface. Figure 3 The arrangement of the sub-modules is used to illustrate the functional modules included in the adaptive body, and does not mean that the data acquisition sub-module, indicator monitoring sub-module, operating condition perception sub-module, and parameter optimization sub-module must be executed in series according to the order shown in the figure; the data acquisition sub-module, operating condition perception sub-module, and indicator monitoring sub-module can respectively provide the parameter optimization sub-module with the current operating condition type and key performance indicator dataset.

[0081] The acquisition submodule collects power information from the grid side and can communicate with current transformers, energy meters, or electrical quantity acquisition circuits located on the grid side. It adds sampling time information to the received real-time current and voltage data and provides the corresponding data to the operating condition sensing submodule according to the adaptive body's operating cycle. The operating condition sensing submodule identifies the current operating condition type based on the grid-side power information provided by the acquisition submodule. It can use preset rules or a pre-trained operating condition identification model and provides the identified current operating condition type to the parameter optimization submodule through the indicator monitoring submodule. The indicator monitoring submodule collects the inverter's key performance indicator dataset in real time. It acquires the inverter's operating data during anti-reverse current control, calculates quantitative indicators to evaluate the current anti-reverse current control effect, and compares the actual values ​​of each quantitative indicator with the corresponding target values ​​to obtain the corresponding deviation. The indicator monitoring submodule provides the key performance indicator dataset to the parameter optimization submodule. The parameter optimization submodule determines the inverter's control parameter set based on the current operating condition type and the key performance indicator dataset. The parameter optimization submodule can call the mapping library between operating conditions and parameters to obtain the initial control parameters or candidate parameter ranges, and solve for the optimized values ​​of the control parameters through parameter optimization algorithms. The adaptive body can adopt a hierarchical modular structure, encapsulating data acquisition, operating condition perception, index monitoring, and parameter optimization into mutually cooperating functional modules, and setting a safety verification layer between the parameter optimization submodule and the anti-backflow control loop. Each functional module can be implemented by multiple software modules running on the same processor, or it can be encapsulated in a microservice manner and exchange data through a high-speed bus.

[0082] In some implementations, the adaptive agent can be integrated into the inverter's EMS or PCS, or it can be configured as a host computer, coprocessor, or independent edge computing entity that communicates with the anti-reverse current control loop. The adaptive agent can operate as a higher-order optimization layer, while the anti-reverse current control loop can operate as a basic control layer.

[0083] like Figure 3As shown, the control device may further include an anti-reverse current control loop. The anti-reverse current control loop generates power commands for each controllable port based on grid-side power feedback and adjusts the inverter output according to the power commands. Grid-side power feedback can be determined based on real-time current and voltage data collected by current transformers or energy meters installed on the grid side. The anti-reverse current control loop compares the grid-side power feedback with the anti-reverse current target power to obtain the grid-side power deviation and determines the adjustment requirements for the inverter output based on the grid-side power deviation. Grid-side power feedback can be represented according to a predetermined power direction rule, for example, defining the direction of power transmission from the grid-connected system to the grid as the reverse current direction. Each controllable port may include at least one of a battery port, a photovoltaic port, and an inverter output port. The power commands for each controllable port may include at least one of a battery power setpoint Pbat_ref, a photovoltaic power setpoint Ppv_ref, and an inverter output power setpoint Pinv_ref. By adopting a heterogeneous dual-layer control structure of "basic control - high-order optimization decoupling," the entire control device is divided into two independent yet cooperative control layers in terms of operating time scale, hardware platform, and functional level: the basic control layer consists of the aforementioned anti-reverse current control loop, and the high-order optimization layer consists of the aforementioned adaptive body, which can be deployed in a host controller (such as an ARM processor, industrial PC, or edge computing gateway) with stronger floating-point operation and machine learning inference capabilities. This layer operates on an optimization cycle of milliseconds to hundreds of milliseconds, and does not directly participate in microsecond-level PWM modulation or real-time adjustment, but focuses on data-driven parameter optimization. To minimize hardware modifications and software intrusion into the existing inverter control architecture, the two layers exchange data and work collaboratively in a loosely coupled manner. The high-order optimization layer and the basic control layer operate on different time scales. The basic control layer maintains a high-speed control cycle (e.g., 50μs), while the high-order optimization layer updates parameters at a longer cycle (e.g., 100ms). The two operate in parallel without conflicting with each other and do not compete for computing resources. The two layers can exchange data via standard industrial communication interfaces (such as CAN, Modbus TCP, or RS485). The adaptive entity subscribes to grid-side power and inverter status data in real time from the controller of the anti-reverse current control loop, forming bidirectional data synchronization. Simultaneously, it sends updated control parameters to the basic control layer in the form of parameter packets. Through this heterogeneous two-layer control architecture, this invention, while maintaining the high-speed response capability of the basic anti-reverse current control loop, endows the system with higher-order intelligence, including online learning, operating condition awareness, and multi-objective optimization, significantly improving the overall control performance.

[0084] As an example, an anti-reverse current control loop can include a power dispatch layer and a power control layer. The power dispatch algorithm in the power dispatch layer can employ a single-input multiple-output (SMILE) structure, using the grid-side power deviation as input and outputting the battery power setpoint Pbat_ref, the photovoltaic power setpoint Ppv_ref, and the inverter output power setpoint Pinv_ref. The power dispatch algorithm can also allocate power regulation based on the current operating state and adjustability of each controllable port. For example, when the energy storage battery has available charging capacity, the charging power of the energy storage battery can be increased to absorb excess energy; when the energy storage battery is in a fully charged, power-limited state, the power regulation allocated to the battery port can be reduced, and the output power of the photovoltaic port can be lowered or the inverter output power adjusted.

[0085] Taking the battery port as an example, the battery power setting Pbat_ref can be compared with the battery power feedback Pbat_fb to obtain the battery power deviation; power control calculations are performed on the battery power deviation to generate the battery current setting ibat_ref; the battery current setting ibat_ref is compared with the battery current feedback, and a duty cycle command for adjusting the DC / DC power conversion stage is generated through the current control loop. The photovoltaic power setting Ppv_ref and the inverter output power setting Pinv_ref can generate the corresponding duty cycle commands for the power conversion stage through the corresponding power control loop and current control loop.

[0086] The adaptive agent provides the updated control parameter set to the anti-reverse current control loop. In one embodiment, the adaptive agent provides the updated control parameter set to at least one of the power scheduling algorithm, power control loop, or current control loop to change the response strength, control output range, or control command change rate of the anti-reverse current control loop to grid-side power deviations. The adaptive agent and the anti-reverse current control loop can operate using a time-scaled mechanism. The adaptive agent can perform operating condition identification, index monitoring, and parameter optimization according to a first control cycle, while the anti-reverse current control loop can perform power control, current control, and duty cycle adjustment according to a second control cycle shorter than the first control cycle. The adaptive agent does not directly participate in microsecond-level pulse width modulation but periodically provides control parameter update packets to the anti-reverse current control loop. The adaptive agent can provide the control parameter update packets to the anti-reverse current control loop via CAN, Modbus TCP, RS485, or the data bus inside the control device. The anti-reverse current control loop can feed back power feedback from the adaptive body to the network side, power feedback from each controllable port, currently used control parameters, and operating data during the anti-reverse current control process, thus forming a bidirectional data stream. Through bidirectional data interaction between the adaptive body and the anti-reverse current control loop, the configurable control parameters of the basic anti-reverse current control loop can be updated while maintaining the original power control, current control, and pulse width modulation logic, thereby reducing modifications to the underlying control firmware or power hardware.

[0087] In this embodiment, a non-transient computer-readable storage medium is also provided. The non-transient computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to implement the aforementioned adaptive anti-reverse current control method. The processor can be located in an inverter, PCS, EMS, or an edge computing device communicatively connected to the inverter.

[0088] In some implementations, the computer program may include one or more of the following: data acquisition program instructions, operating condition identification program instructions, indicator monitoring program instructions, parameter optimization program instructions, and safety verification program instructions. The processor can execute the corresponding program instructions to complete tasks such as acquiring grid-side power information, identifying the current operating condition type, obtaining key performance indicator datasets, determining control parameter sets, ensuring smooth transition of control parameters, and performing safety verification.

[0089] In some implementations, a non-transient computer-readable storage medium may include a read-only memory, random access memory, flash memory, solid-state drive, magnetic disk, optical disk, or other storage medium capable of storing computer programs. The term "non-transient" is used to distinguish it from carrier signals or electromagnetic signals that propagate momentarily in a communication link, and does not imply that the stored computer programs or data cannot be modified.

[0090] In this embodiment, an inverter is also provided. The inverter includes a power conversion unit and a control unit. The power conversion unit performs power conversion and adjusts the inverter's output according to control signals provided by the control unit. The control unit acquires operating data of the inverter and its grid-connected system, performs anti-reverse current control calculations, and controls the operating state of the power conversion unit. The specific functions and effects achieved by the inverter in this embodiment can be explained by referring to other embodiments of this application, and will not be repeated here.

[0091] As described above, this application identifies the current operating condition of the inverter by collecting power information from the grid side and obtains a dataset of key performance indicators for evaluating the actual anti-reverse current control effect. Then, based on the current operating condition and the key performance indicator dataset, it determines the inverter's control parameter set, enabling the anti-reverse current control parameters to be adjusted according to the operating environment and actual control effect. Through the coordination of operating condition perception, indicator monitoring, parameter optimization, safety verification, and the basic anti-reverse current control loop, the anti-reverse current response speed, control stability, and renewable energy utilization can be coordinated under different operating conditions.

[0092] It is understood that the specific examples in this document are merely to help those skilled in the art better understand the embodiments of this application, and are not intended to limit the scope of the invention. In the various embodiments of this application, the sequence number of each process does not imply the order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The various embodiments described in this application can be implemented individually; they can also be implemented in combination where there is no technical conflict, and the embodiments of this application are not limited in this respect.

[0093] Unless otherwise stated, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this application includes any and all combinations of one or more of the associated listed items. The singular forms "a," "the," and "the" as used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0094] The above description is merely a specific embodiment of this application, but the scope of protection of this invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. A control method for an inverter, characterized in that, The control method is used to adjust the output of the inverter, and the control method includes: Collect power information from the power grid side; the power information includes current data and voltage data from the power grid side; The current operating condition type is identified based on the power information collected from the grid side; the current operating condition type is used to characterize the current operating environment status of the inverter. Collect a dataset of key performance indicators for the inverter; the dataset includes quantitative indicators used to evaluate the current anti-backflow control effect of the inverter and the deviation from the target value. The control parameter set of the inverter is determined based on the current operating condition type and the key performance indicator dataset.

2. The control method according to claim 1, characterized in that, The key performance indicator dataset includes stability indicators, dynamic performance indicators, and economic indicators, among which, The stability indicators include one or more of the following: phase margin of the anti-reverse current control loop, gain margin of the anti-reverse current control loop, oscillation amplitude and attenuation rate of inverter output current, and oscillation amplitude and attenuation rate of inverter output power. The dynamic performance indicators include one or more of the following: reverse current regulation time, reverse current overshoot, and steady-state error; the reverse current regulation time is the time from the occurrence of reverse current to the recovery to normal; the reverse current overshoot refers to the percentage deviation between the maximum reverse current power and the target regulation value during the reverse current regulation process; the steady-state error is the average deviation between the grid-side power and the anti-reverse current target power within a preset statistical time after the anti-reverse current regulation process enters a stable state. The economic indicators include curtailment rate and / or inverter life loss; the curtailment rate is the loss of renewable energy output due to anti-reverse flow restrictions; the inverter life loss is the life loss of the inverter calculated by the junction temperature estimation model.

3. The control method according to claim 1, characterized in that, A lightweight machine learning model is used to identify the current operating condition type, which includes: Steady-state operating condition; the steady-state operating condition is that the power and voltage on the grid side remain within a preset fluctuation range within a preset time. And / or, transient operating conditions; the transient operating conditions are when the grid-side power and voltage exceed a preset fluctuation range; And / or, extreme operating conditions; the extreme operating conditions are when the grid-side power and voltage exceed permissible physical limits or safety specification boundaries.

4. The control method according to claim 1, characterized in that, The control parameters of the inverter include at least one of proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd. The proportional coefficient Kp is used to characterize the adjustment gain coefficient that is proportional to the deviation between the current grid-side power and the anti-reverse current target power. The integral coefficient Ki is used to characterize the adjustment gain coefficient that is proportional to the integral of the deviation value over time. The derivative coefficient Kd is used to characterize the adjustment gain coefficient that is proportional to the rate of change of the deviation value over time.

5. The control method according to claim 1, characterized in that, The inverter's control parameter set includes one or more of the following: control gain parameter, upper limit of power command output, lower limit of power command output, control parameter change rate threshold, and power command change rate limit. The steps for determining the control parameter set of the inverter based on the current operating condition type and the key performance indicator dataset include: Based on the current operating condition type, determine one or more of the following: the upper limit of the power command output, the lower limit of the power command output, the control parameter change rate threshold, and the power command change rate limit; Based on the aforementioned key performance index dataset, construct the objective function and constraints for control parameter optimization; Solve for the optimal values ​​of the control parameters to determine the control parameter set of the inverter.

6. The control method according to claim 5, characterized in that, An improved particle swarm optimization algorithm is used to solve for the optimized values ​​of the control parameters to determine the control parameter set of the inverter.

7. The control method according to claim 5, characterized in that, The objective function is obtained by weighted optimization of multiple objective functions including reverse current regulation time, reverse current overshoot, and renewable energy output loss; the reverse current regulation time is the time from the occurrence of reverse current to recovery to normal; the reverse current overshoot refers to the percentage deviation between the maximum reverse current power and the target regulation value during the reverse current regulation process; the renewable energy output loss is the renewable energy power sacrificed due to reverse current prevention; the control method further includes linear interpolation of the control parameters before and after optimization to determine the control parameter set of the inverter.

8. The control method according to claim 1, characterized in that, The control method also includes security verification, specifically including: Perform physical limit hard constraint checks on the updated control parameters; The updated control parameters are subjected to stability prediction; the stability prediction is determined by obtaining the predicted quantitative value of the index in the key performance index dataset through the Nyquist criterion or Lyapunov function and the predicted quantitative value of the index exceeds the preset range, in which case the updated control parameters are determined to lack stability. And when it is detected that the updated control parameters do not meet the physical limit hard constraint check or lack stability, a rollback to the previous control parameters is triggered.

9. A control device for an inverter, characterized in that, The control device includes an adaptive body, the adaptive body comprising: The acquisition submodule is used to acquire power information from the power grid side, including current data and voltage data from the power grid side. The operating condition sensing submodule is used to identify the current operating condition type based on the power information; the current operating condition type is used to characterize the current operating environment state of the inverter. The indicator monitoring submodule is used to acquire the operating data of the inverter during the anti-reverse current control process, and determine the key performance indicator dataset of the inverter based on the operating data; the key performance indicator dataset includes quantitative indicators used to evaluate the current anti-reverse current control effect of the inverter and the deviation from the target value. The parameter optimization submodule is used to determine the control parameter set of the inverter based on the current operating condition type and the key performance index dataset.

10. The control device according to claim 9, characterized in that, The control device further includes an anti-reverse current control loop, which is configured to generate power commands for each controllable port based on grid-side power feedback to adjust the output of the inverter; the adaptive body provides the updated set of control parameters to the anti-reverse current control loop.

11. A non-transient computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the control method as described in any one of claims 1-8.

12. An inverter, characterized in that, It includes a power conversion unit and a control unit, the control unit including the control device as described in claim 9 or 10, the control device being used to adjust the output of the power conversion unit, and the control device being used to adjust the output of the power conversion unit based on a determined set of control parameters.

Images