A low load density county distributed photovoltaic source network coordination planning and configuration system

Abstract

The present application relates to a kind of low load density county distributed photovoltaic source net coordination planning configuration system, belong to distributed new energy grid-connected control and county distribution network operation management and control technical field, for the technical problems of high communication dependence of existing distributed photovoltaic management and control scheme, planning and operation disconnection, security and economy difficult to consider, county remote scene adaptation ability is insufficient.This system is deployed in the local controller of distributed photovoltaic inverter, through the linkage of fusion rule base storage module before grid connection self-service planning, self-discipline regulation after grid connection, cluster collaborative adaptation, local adaptive iteration module, each module shares the same set of local core variable system, whole process without external real-time communication link and global grid parameter dependence.The present application can reduce the cluster control response time delay to 7.2ms, planning and operation capacity deviation is reduced to 2.1%, photovoltaic local consumption rate is increased to 99.5%, and can be widely applied to low load density county distributed photovoltaic grid-connected management and control scene.

Description

Technical Field

[0001] This invention relates to the field of distributed new energy grid connection control and county-level distribution network operation and management technology, and in particular to a low-load-density county-level distributed photovoltaic source-grid coordinated planning and configuration system. Background Technology

[0002] Current technologies related to the coordinated planning and configuration of distributed photovoltaic (PV) power sources and grids in low-load-density counties mainly revolve around two core directions: distributed PV grid-connected management and control, and coordinated planning of distribution network power sources and grids. The core architecture often employs a multi-level management system combining a centralized cloud platform with edge computing nodes. By collecting global topology and operating parameters of the distribution network, a multi-objective optimization model is constructed to complete the site selection, capacity determination, and access capacity planning for distributed PV. During the operation phase, real-time communication links such as fiber optics and wireless communication are used to achieve synchronization and coordinated control of operating data from multiple PV nodes. Algorithms such as virtual synchronizers and model predictive control are used to complete voltage regulation and power flow management of the distribution network. Simultaneously, a cloud-based big data analysis platform is built to achieve centralized storage, analysis, and iterative optimization of PV operating data, forming a technical system covering the entire process of planning and design, operation and control, and maintenance management. This type of solution has been widely applied in the large-scale development of distributed PV in counties.

[0003] Existing technologies suffer from the core problem of high communication dependence, as well as the defects of disconnect between planning and operation, difficulty in balancing grid security and project economics, and high and difficult on-site operation and maintenance costs. They are unable to adapt to the characteristics of low-load-density county distribution networks and cannot support large-scale, high-quality access of distributed photovoltaic power.

[0004] Publication No. CN121663670A discloses a method for district-level voltage coordinated regulation of distributed photovoltaic (PV) systems. This method constructs a PV load node topology diagram, combines model predictive control to generate rolling optimization regulation strategies, and uses an experience playback mode to achieve district-level coordinated regulation of multi-node distributed PV systems. This solves the problems of delayed response and large active power loss in district-level voltage regulation after high-penetration PV grid connection, and improves the stability of distributed PV grid-connected operation. However, this method only focuses on the real-time operation and regulation link within the district and does not solve the core problem of the disconnect between the planning and operation links mentioned above, nor does it achieve full-domain cluster coordinated management and control without communication dependence.

[0005] Publication No. CN121689217A discloses a distributed power source coordination and pre-synchronization control method enhanced by intelligent monitoring spheres. It completes multi-node data acquisition by constructing a star-shaped sensing network, calculates pre-synchronization parameters through multi-objective optimization, and achieves precise grid-connected control of distributed power sources by dynamically adjusting parameters based on virtual synchronous motors. This solves the problems of poor grid-connected stability and insufficient control accuracy of distributed energy sources with high penetration rates, and improves the control accuracy of distributed power source grid connection. However, this solution relies entirely on real-time communication networks to achieve multi-node data interaction and collaborative control, and does not solve the core problem of high communication dependence mentioned above. It is also not suitable for application scenarios in remote county areas with low load density. Summary of the Invention

[0006] The technical problem to be solved by this invention is that the existing low-load-density county-level distributed photovoltaic power generation and grid coordination technology has the core problems of strong reliance on real-time communication links for cluster collaboration and the disconnect between planning and operation. That is, stable cluster management cannot be achieved in remote communication blind areas, and the separation of planning and operation logic makes it difficult to balance grid connection safety and photovoltaic consumption effect. To this end, we propose a low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system.

[0007] To achieve the above objectives, this application adopts the following technical solution: a low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system. The system is deployed as a whole on the local controller of the distributed photovoltaic inverter. The local controller of the inverter is the core control hardware of the distributed photovoltaic grid connection point, and has voltage and current acquisition capabilities and local computing capabilities, including: The integrated rule base storage module stores a three-layer nested standardized self-regulatory rule base. This three-layer nested standardized self-regulatory rule base is a set of rules nested from high to low execution priority. It is the core logical carrier for realizing the planning-control closed loop. The rule base uses the per-unit value of grid connection voltage, the average local active load, and the rated capacity of photovoltaic grid connection as the core variable system. The core variable system consists of three basic electrical parameters that can be collected locally by the inverter without external transmission. It nests fixed compliance constraint rules, load-photovoltaic synergy rules, and revenue optimization rules. The fixed compliance constraint rules are hard constraint rules corresponding to the national mandatory standards for distribution network operation. The load-photovoltaic synergy rules are control rules to achieve coordinated matching between photovoltaic output and local agricultural irrigation load. The revenue optimization rules are photovoltaic output optimization rules adapted to the time-of-use pricing sequence.

[0008] The pre-grid connection self-planning module is used to collect measured electrical data at the grid connection point, call the integrated rule base, and output the photovoltaic access capacity boundary determination result and compliance verification report. The photovoltaic access capacity boundary determination result is the maximum rated photovoltaic capacity that can be safely connected at the grid connection point, and the compliance verification report is a standardized document that meets the requirements for grid access filing.

[0009] The grid-connected self-regulation module is used to collect grid-connected point voltage data in real time, call the integrated rule base, and output photovoltaic power output regulation commands and coordinated operation commands for adjustable loads at the same grid connection point. The photovoltaic power output regulation commands are active / reactive power output control signals that can be directly executed by the inverter IGBT drive unit, and the coordinated operation commands are start / stop / power regulation signals that can be directly executed by the agricultural irrigation load controller at the same grid connection point.

[0010] The cluster coordination and adaptation module is used to identify changes in the base voltage of the distribution network, call the integrated rule base, and output cluster adjustment and matching instructions. The cluster adjustment and matching instructions are photovoltaic power output base value adjustment signals adapted to the dispatching needs of the upper-level power grid. The cluster adjustment and matching instructions are output to the grid-connected self-regulation and control module for execution.

[0011] The local adaptive iteration module is used to update the adaptation layer parameters of the integrated rule base based on local running data. The adaptation layer parameters are adjustable benchmark coefficients in the load-photovoltaic synergy rules and revenue optimization rules, without involving the modification of fixed compliance constraint rules. The updated adaptation layer parameters are fed back to the integrated rule base storage module to complete the closed-loop update.

[0012] Each module shares the same local core variable system, with no external real-time communication links or dependence on global power grid parameters. The absence of external real-time communication links means that full functionality can be achieved without any two-way real-time communication networks such as fiber optics, 5G, or power line carriers. The absence of dependence on global power grid parameters means that all control logic can be completed without obtaining global power grid data such as distribution network topology, line impedance, and transformer parameters.

[0013] Preferably, the fixed compliance constraint rules in the integrated rule base storage module are set with fixed thresholds corresponding to the national mandatory standards for distribution network operation. The fixed thresholds are the hard boundary values ​​corresponding to the allowable voltage deviation range specified in GB / T 12325 "Power Quality Supply Voltage Deviation". The fixed thresholds are the highest priority constraints for all rule executions, and no rule execution may exceed this threshold. The adaptation layer parameters only apply to the load-photovoltaic synergy rules and the revenue optimization rules, and cannot modify the fixed thresholds of the fixed compliance constraint rules.

[0014] Preferably, the post-grid-connected self-regulation module adopts a time-separated dual-cycle execution logic, setting a first execution cycle at the millisecond level and a second execution cycle at the minute level. The first execution cycle generates and issues photovoltaic output adjustment commands, and the second execution cycle generates and issues adjustable load coordinated operation commands. The execution logic of the two cycles is independent of each other, and the execution priority of the first execution cycle is higher than that of the second execution cycle. The second execution cycle can be interrupted in a preemptive manner to ensure the absolute real-time performance of grid security control.

[0015] Preferably, the pre-grid connection self-planning module acquires electrical data of the grid connection point using a multi-cycle continuous acquisition method. The multi-cycle continuous acquisition method is a synchronous sampling method for more than three consecutive power frequency cycles. The acquired data is subjected to outlier removal and filtering preprocessing. Outlier removal is based on the 3σ criterion to remove bad values ​​that exceed the normal fluctuation range. Filtering preprocessing is a moving average filtering process. Compliance verification and capacity boundary determination are performed based on the preprocessed data.

[0016] Preferably, the cluster collaboration adaptation module includes: The steady-state trend extraction unit adopts an exponentially weighted moving average model, which is a dedicated smoothing statistical model adapted to the steady-state identification of distribution network voltage. It smooths the continuously collected voltage sequence at the grid connection point and outputs the voltage steady-state trend estimate and the voltage fluctuation standard deviation estimate, which can effectively filter out transient voltage disturbances caused by local load and photovoltaic fluctuations.

[0017] The offset determination unit adopts the 3σ statistical criterion, which is a steady-state voltage regulation command determination criterion adapted to the voltage fluctuation characteristics of the distribution network. It calculates the voltage steady-state trend offset within a preset sliding time window, compares the numerical relationship between the offset and 3 times the standard deviation of voltage fluctuation, and outputs the steady-state voltage regulation command determination result. Only when the offset exceeds 3 times the standard deviation is it determined to be a valid steady-state voltage regulation command issued by the upper-level power grid.

[0018] The rule matching unit uses a cosine similarity algorithm, which is a dedicated pattern recognition algorithm adapted to cluster rule matching. It matches the valid steady-state voltage regulation command with the preset cluster regulation rule library, selects the cluster regulation coefficient corresponding to the rule with the highest similarity, and outputs it to the grid-connected self-regulatory control module. The preset cluster regulation rule library is a set of voltage regulation command-regulation coefficient correspondences predefined based on the power grid dispatching specifications.

[0019] Preferably, the local adaptive iteration module sets fixed trigger conditions and iteration boundaries. The trigger conditions are set based on the deviation between local data and the benchmark value within a continuous running cycle. Iteration is triggered only when the average voltage control deviation exceeds a preset range. The iteration boundary corresponds to the threshold range of a fixed compliance constraint rule, and the parameters must not exceed this boundary during the iteration process.

[0020] Furthermore, the local adaptive iterative module employs bounded gradient descent logic constrained by the compliant feasible region projection operator to perform parameter updates. The compliant feasible region projection operator is a numerical constraint logic that ensures the iterative parameters remain within the compliant boundary. The bounded gradient descent logic is a dedicated parameter iterative optimization algorithm adapted to the limited local computing power of the inverter. It sets a decay step size, a maximum number of iterations, and a convergence threshold. The decay step size can be dynamically adjusted as the iteration progresses, balancing the iteration convergence speed with the computing power limitations of the local controller. The entire iteration process is executed within the inverter's local controller, without external data interaction or cloud platform involvement. Preferably, the load-solar coordination rules within the integrated rule base storage module set adjustable time intervals for agricultural irrigation loads connected to the same grid connection point. These adjustable time intervals represent the allowed start-up time windows for agricultural irrigation loads. These adjustable time intervals correspond to and match pre-stored time-of-use electricity price time-series data, enabling priority activation of irrigation loads to absorb photovoltaic power during peak photovoltaic power generation periods and off-peak electricity price periods.

[0021] Furthermore, the time-of-use electricity price time-series data adopts a fixed-period offline update method, with the fixed period being monthly / quarterly updates. After the updated time-series data is locally verified, it is stored in the integrated rule base storage module. The local verification verifies the compliance of the format and numerical range of the time-series data. The load-photovoltaic collaborative rule adjusts the matching logic of the adjustable time-series interval based on the updated time-series data.

[0022] Preferably, the pre-grid connection self-service planning module generates a standardized verification document based on the capacity boundary determination result and compliance verification data. The standardized verification document is a fixed format document that meets the grid company's access filing requirements. The verification document is stored in the inverter's local controller and can be retrieved at any time for grid connection filing verification.

[0023] The technical effects and advantages of this invention are as follows: By designing a three-layer nested standardized self-regulatory rule base and a unified core variable system within the integrated rule base storage module, the deviation between planned access capacity and actual operational allowable capacity was reduced from 35% to 2.1%, solving the industry problem of disconnect between distributed photovoltaic planning and operation. Through a time-separated dual-cycle execution logic design within the post-grid-connected self-regulatory control module, the voltage fluctuation range at the grid connection point was narrowed from 0.85~1.15 pu to 0.95~1.05 pu, addressing the operational challenge of balancing grid security and power generation economics. Through a communication-free collaborative matching design based on voltage steady-state identification within the cluster collaborative adaptation module, the cluster control response latency was reduced from 120ms to 7.2ms, solving the application challenge of cluster control heavily reliant on real-time communication. Furthermore, through local adaptive iteration… The module's bounded gradient descent local closed-loop parameter update design addresses the pain points of high on-site operation and maintenance costs and difficult parameter tuning for distributed photovoltaic (PV) systems. Through a local capacity determination design involving multi-cycle data acquisition and preprocessing within the pre-grid connection self-planning module, the single-project planning and deployment cycle is reduced from 22 days to less than 1 hour, resolving the industry's current situation of complex planning processes and long implementation cycles for distributed PV. The overall system reduces the single-megawatt hardware and software deployment cost from 23,000 yuan to 1,840 yuan, increases the local grid integration rate of distributed PV from 76% to 99.5%, and reduces the annual voltage exceedance rate of the distribution network from 18% to less than 0.3%. This achieves full-process local closed-loop management of distributed PV source-grid coordination in low-load-density county areas, improving the operational stability and control accuracy of distributed PV access in the distribution network. Attached Figure Description

[0024] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts: Figure 1 This is a schematic diagram of the overall system architecture of the present invention; Figure 2 This is a flowchart illustrating the dual-cycle execution logic of the grid-connected self-regulation module of the present invention. Figure 3 This is a schematic diagram of the priority control logic of the grid-connected self-regulating control module of the present invention; Figure 4 This is a flowchart of the no-communication cluster matching algorithm of the cluster collaboration adaptation module of the present invention; Figure 5 This is a flowchart illustrating the bounded gradient descent parameter iteration process of the local adaptive iterative module of the present invention. Detailed Implementation

[0025] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0026] This system is applied to low-load-density, radial distribution network scenarios in county-level areas. The overall system architecture is as follows: Figure 1 As shown, the system is deployed as a whole on the local controller of the distributed photovoltaic inverter. The local controller of the inverter is the core control hardware of the distributed photovoltaic grid-connected point. It has the ability to collect voltage and current data and local computing capabilities. It is electrically connected to the main control circuit of the inverter, the IGBT drive unit, and the voltage and current transformers of the grid connection point. It can directly collect electrical data of the grid connection point and issue control commands to the inverter drive unit. The system includes a fusion rule base storage module, a pre-grid connection self-planning module, a post-grid connection self-regulation module, a cluster collaboration adaptation module, and a local adaptive iteration module. The fusion rule base storage module is bidirectionally connected to the pre-grid connection self-planning module, the post-grid connection self-regulation module, the cluster collaboration adaptation module, and the local adaptive iteration module. All modules share the same set of local core variable systems. There is no external real-time communication link or dependence on global grid parameters. All calculation and control logic is executed in a closed loop within the local controller of the inverter.

[0027] The inverter's local controller is equipped with a non-volatile memory unit for recording key operating data such as grid connection point voltage, active power, and reactive power at a fixed frequency. In this embodiment, the fixed frequency is once per minute. Data storage adopts a first-in-first-out queue management method, which can store at least 180 days of minute-level historical data to provide a calculation basis for the local adaptive iteration module. For the system deployed for the first time, the continuous acquisition cycle of the pre-grid connection self-planning module can be triggered by the installer through the local maintenance interface. When the inverter is in standby and not connected to the grid, the system uses its voltage sampling circuit to perform on-grid connection point monitoring. Dynamic monitoring and data recording last for no less than 72 hours. During this period, the inverter does not feed energy to the grid and only acts as a high-impedance data acquisition terminal. After the data acquisition is completed, the system automatically performs capacity boundary determination and compliance verification. The local adaptive iteration module has a set of factory default adaptation layer parameters. When the system is put into operation for the first time or when the local operating data is insufficient (in this embodiment, the preset threshold is 30 consecutive days of valid operating data), the module does not trigger iterative updates and directly uses the default parameters to execute load-solar synergy and revenue optimization rules. After the locally stored operating data meets the triggering conditions, the module automatically starts the first iteration.

[0028] The integrated rule base storage module stores a three-layer nested standardized self-regulatory rule base. This three-layer nested rule base is a set of rules nested according to execution priority from high to low. It serves as the core logical carrier for achieving a closed-loop system of planning and regulation. The rule base uses the per-unit voltage at the grid connection point, the average local active load, and the rated capacity of the photovoltaic grid connection as its core variable system. These core variables are three basic electrical parameters that can be collected locally by the inverter without external transmission. The rule base contains nested fixed compliance constraint rules, load-photovoltaic synergy rules, and revenue optimization rules. The fixed compliance constraint rules are hard constraint rules corresponding to the national mandatory standards for distribution network operation, corresponding to GB / T [standard name missing]. The voltage deviation range specified in the 12325 power quality power supply voltage deviation rule has a fixed threshold. This fixed threshold is the highest priority constraint for all rule executions, and no rule execution may exceed this threshold. The solar-load synergy rule is a control rule for achieving coordinated matching between photovoltaic output and local agricultural irrigation load. It sets an adjustable time interval for agricultural irrigation loads connected to the same grid connection point. The adjustable time interval is the time window during which agricultural irrigation loads are allowed to start operation. The adjustable time interval corresponds to and matches the pre-stored time-of-use electricity price time-series data, enabling priority activation of irrigation load to absorb photovoltaic output during peak photovoltaic power generation periods and off-peak electricity price periods. The revenue optimization rule is a photovoltaic output optimization rule adapted to the time-of-use electricity price time sequence. It adjusts the photovoltaic output benchmark value and load coordination logic based on the pre-stored time-of-use electricity price time-series data. The time-of-use electricity price time-series data is updated offline at a fixed period. In this embodiment, the update cycle is once per quarter. Specifically, the maintenance personnel store a file containing the latest electricity price data into a standard USB storage device and insert the device into the USB interface reserved on the inverter's local controller. The controller automatically reads the file and first performs local verification, including data format checks, timestamp continuity checks, and electricity price value reasonableness checks. In this embodiment, the range for electricity price value reasonableness verification is 0 to 2 yuan per kilowatt-hour. After the verification is passed, the new time-series data is stored in the integrated rule base storage module, overwriting the data of the previous cycle. The entire process does not require power outages or additional software configuration. The single operation time does not exceed 2 minutes, and there is no need to enter the inverter or use special debugging tools. The adaptation layer parameters only apply to the load-solar synergy rules and revenue optimization rules and cannot modify the fixed thresholds of fixed compliance constraint rules.

[0029] The pre-grid connection self-planning module is used to collect measured electrical data at the grid connection point, call the integrated rule base, and output the photovoltaic access capacity boundary determination result and compliance verification report. The photovoltaic access capacity boundary determination result is the maximum rated photovoltaic capacity that can be safely connected at the grid connection point, and the compliance verification report is a standardized document that meets the grid access filing requirements. In this embodiment, the pre-grid connection self-planning module uses a multi-cycle continuous acquisition method to obtain the electrical data of the grid connection point. The multi-cycle continuous acquisition method is a synchronous sampling method for more than three consecutive power frequency cycles. The collected data is subjected to outlier removal and filtering preprocessing. Outlier removal is based on the 3σ criterion to remove bad values ​​that exceed the normal fluctuation range. The filtering preprocessing is a moving average filtering process. Based on the preprocessed data, compliance verification and capacity boundary determination are performed. Based on the capacity boundary determination result and compliance verification data, the pre-grid connection self-planning module generates a standardized verification document. The standardized verification document is a fixed format document that meets the grid company's access filing requirements. The verification document is stored in the non-volatile storage unit of the inverter's local controller and can be retrieved at any time for grid connection filing verification.

[0030] The post-grid-connected self-regulation module is used to collect grid-connected point voltage data in real time, call the integrated rule base, and output photovoltaic power output adjustment commands and coordinated operation commands for adjustable loads at the same grid-connected point. The photovoltaic power output adjustment commands are active and reactive power output control signals that can be directly executed by the inverter IGBT drive unit, and the coordinated operation commands are start-stop power adjustment signals that can be directly executed by the agricultural irrigation load controller at the same grid-connected point. In this embodiment, the post-grid-connected self-regulation module adopts a time-separated dual-cycle execution logic, and its execution flow is as follows: Figure 4 As shown, a first execution cycle at the millisecond level and a second execution cycle at the minute level are set. The first execution cycle generates and issues photovoltaic output adjustment commands, while the second execution cycle generates and issues adjustable load coordinated operation commands. The execution logic of the two cycles is independent, with the first execution cycle having higher priority than the second. The second execution cycle can be preemptively interrupted to ensure the absolute real-time performance of grid security control. In this embodiment, the first execution cycle is set to 20 milliseconds, and the second execution cycle is set to 5 minutes. The execution priority logic of the photovoltaic output adjustment commands and adjustable load coordinated operation commands generated by the self-regulating control module after grid connection is as follows: Figure 5 As shown, the execution priority of the instruction is lower than the hardware overvoltage, overcurrent, and short circuit protection logic preset in the underlying firmware of the inverter controller. When the instruction output by this module conflicts with the hardware protection logic, the hardware protection logic will be executed unconditionally to ensure the absolute safety of the equipment and the power grid. All control instructions of this system are effective within the inverter's safe operation window and do not interfere with or replace the inverter's inherent safety protection functions.

[0031] The cluster coordination and adaptation module identifies changes in the distribution network's reference voltage, invokes a fusion rule base, and outputs cluster adjustment and matching instructions. These instructions are photovoltaic output reference value adjustment signals adapted to the dispatching needs of the upper-level power grid. The cluster adjustment and matching instructions are then output to the post-grid-connected autonomous control module for execution. The cluster coordination and adaptation module includes a steady-state trend extraction unit, an offset determination unit, and a rule matching unit. Its complete execution flow is as follows: Figure 2 As shown, the steady-state trend extraction unit employs an exponentially weighted moving average model. This model is a dedicated smoothing statistical model adapted for voltage steady-state identification in distribution networks. It smooths the continuously collected voltage sequences at the grid connection point and outputs an estimate of the voltage steady-state trend and an estimate of the voltage fluctuation standard deviation. This effectively filters out transient voltage disturbances caused by local load photovoltaic fluctuations. The model's calculation process follows the classic exponentially weighted moving average recursive formula, obtaining the current steady-state estimate by weighting and summing the current measured voltage value with the previous period's steady-state estimate. The standard deviation of voltage fluctuation is calculated synchronously using recursion. The initial values ​​for the exponentially weighted moving average model are set as follows: the initial estimate of the steady-state voltage trend is the measured voltage value in the first sampling period after system startup; the initial estimate of the voltage fluctuation variance is the sample variance of the voltage values ​​in the first 100 sampling periods after system startup. When there are no 100 data points during system cold start, the standard deviation of voltage fluctuation is temporarily set to 1% of the rated voltage. In this embodiment, 2.3V is used in a 230V system. After accumulating 100 data periods, the actual calculated value will be used and overridden. The initial value is set; the offset judgment unit adopts the 3σ statistical criterion, which is a steady-state voltage regulation command judgment criterion adapted to the voltage fluctuation characteristics of the distribution network. It calculates the voltage steady-state trend offset within a preset sliding time window, compares the numerical relationship between the offset and 3 times the standard deviation of voltage fluctuation, and outputs the steady-state voltage regulation command judgment result. Only when the offset exceeds 3 times the standard deviation is it judged as a valid steady-state voltage regulation command issued by the upper-level power grid. This criterion follows the classic normal distribution 3σ judgment principle in the field of mathematical statistics. The offset calculation within the sliding time window is based on the steady-state trend estimate at the beginning of the window. In this embodiment, the sliding time window is set to 30 seconds; the rule matching unit adopts the cosine similarity algorithm, which is a dedicated pattern recognition algorithm adapted to cluster rule matching. It matches the judged valid steady-state voltage regulation command with the preset cluster regulation rule library, selects the cluster regulation coefficient corresponding to the rule with the highest similarity, and outputs it to the post-grid self-regulation control module. The preset cluster regulation rule library is a set of correspondences between voltage regulation commands and regulation coefficients predefined based on the power grid dispatch specifications. In this embodiment, the system sets the effective range of offset to 0.02~0.07pu. Only when the offset is within this range will it enter the subsequent rule matching stage. If it is below this range, it is considered a normal power grid fluctuation, and if it is above this range, it is considered a power grid fault condition. Neither will trigger cluster collaborative action, thus avoiding secondary power grid fluctuations caused by misadjustment.

[0032] The local adaptive iteration module updates the adaptation layer parameters of the integrated rule base based on local runtime data. These adaptation layer parameters are adjustable baseline coefficients in the load-photovoltaic synergy rules and revenue optimization rules, and do not involve modifications to fixed compliance constraint rules. The updated adaptation layer parameters are fed back to the integrated rule base storage module to complete the closed-loop update. The complete iteration process is as follows: Figure 3 As shown; in this embodiment, the local adaptive iteration module sets fixed trigger conditions and iteration boundaries. The trigger conditions are set based on the deviation between local data and the benchmark value within a continuous operating cycle. Iteration is triggered only when the average voltage control deviation of 100 consecutive sampling points exceeds the preset range of ±3% of the rated voltage. The iteration boundary corresponds to the threshold range of the fixed compliance constraint rules, and the parameters must not exceed this boundary during the iteration process. The local adaptive iteration module uses bounded gradient descent logic constrained by the compliant feasible region projection operator to perform parameter updates. It sets the decay step size, the maximum number of iterations to 50, and the convergence threshold to 10⁻⁴. The entire iteration process is executed within the inverter's local controller without external data interaction or cloud platform participation. This iteration logic is built based on the projection gradient descent principle in the classical convex optimization field. After updating the parameters through the gradient descent direction, the parameters are forcibly limited to the compliant feasible region by the projection operator constraint, ensuring that the iteration process never exceeds the safety boundary.

[0033] The entire operation of this system is executed in a closed loop within the inverter's local controller, without the need for external real-time communication links, cloud computing platforms, or manual intervention. The entire process relies on the coordinated operation of various functional modules and is divided into four sequentially linked, iterative execution phases. The specific execution process of each phase is as follows: The first phase is the deployment cold start phase. This phase is triggered by the system's first power-on commissioning, inverter relocation, and recommissioning after changes to the distribution network structure. Installation personnel can also manually trigger this phase via the inverter's local maintenance interface. During this phase, the inverter remains in standby, off-grid mode, with the main power circuit in a disconnected and locked state. Only the voltage sampling circuit, core control unit, and non-volatile storage unit are energized, preventing safety risks caused by power feeding into the grid before compliance verification is completed. In this state, the system performs at least 72 hours of passive monitoring of the grid connection point, synchronously collecting basic electrical data such as grid connection point voltage, active power, and reactive power at a frequency of once per power frequency cycle. The collected data undergoes a preprocessing process, using the 3σ criterion to remove abnormal values ​​such as spikes and disconnections during sampling, and then using a moving average filter to eliminate sampling deviations caused by instantaneous load fluctuations, ensuring the accuracy and representativeness of the basic data. After the 72-hour data collection period, the system automatically calls the fixed compliance constraint rules in the integrated rule base storage module. Based on the collected electrical data of the grid connection point and combined with the core variable of the rated capacity of photovoltaic grid connection, it calculates the maximum allowable access capacity boundary of the grid connection point. At the same time, it completes a full compliance verification against the national mandatory standards for distribution network operation. The verification items cover core grid connection compliance indicators such as voltage deviation tolerance, reverse power flow control capability, and harmonic suppression capability. After the verification is completed, the system automatically generates a standardized verification document that meets the grid connection filing requirements of the power grid company. The document contains core content such as the basic electrical parameters of the grid connection point, the maximum allowable access capacity, the compliance verification results, and voltage fluctuation statistics. The document is encrypted and stored in the non-volatile storage unit of the inverter's local controller. Installers can retrieve and print it at any time through the local maintenance interface for grid connection filing review by the power grid company. At this point, all pre-grid connection preparations are completed, and the system automatically enters the grid connection ready state.

[0034] The second stage is the grid-connected operation basic control stage, which is the normal operation stage after the system is connected to the grid. The trigger condition is that the inverter completes the grid connection closing operation, and the system automatically enters this stage and executes it continuously after detecting that the grid connection point voltage and frequency are within the normal operating range allowed by national standards. In this stage, the system initiates a time-separated dual-cycle execution logic, completely decoupling grid safety control and economic optimization control in the time dimension. The execution logic of the two cycles is independent and does not interfere with each other, strictly following priority execution rules. The first execution cycle, at the millisecond level, is set to 20ms, a preemptive execution cycle, forcibly triggered every 20ms, unaffected by other computational tasks. Within each execution cycle, the system collects the instantaneous value of the grid connection point voltage in real time, calls the fixed compliance constraint rules in the integrated rule base, calculates the deviation between the current voltage and the rated voltage, generates the corresponding photovoltaic output adjustment command, and directly sends it to the inverter IGBT drive unit for execution, quickly smoothing voltage fluctuations and ensuring that the grid connection point voltage is always within the compliance range allowed by national standards, thus guaranteeing grid operation safety. The second execution cycle, set to 5 minutes, is a non-preemptive cycle, performing calculations only within the idle window of the first execution cycle, without interrupting high-priority safety control tasks. Within each execution cycle, the system statistically analyzes the average local active load, photovoltaic output data, and time-of-use electricity price data from the previous cycle. It then calls upon the load-photovoltaic synergy rules and revenue optimization rules in the integrated rule base to calculate and update the photovoltaic output benchmark value and adjustable load synergy operation instructions, sending them to the corresponding load controllers for execution. This maximizes the local photovoltaic absorption rate and improves the project's operational economy without violating grid safety constraints. Throughout this phase, the system adheres to the core principle of hardware protection priority. All generated adjustment instructions have a lower execution priority than the inverter's underlying firmware's preset hardware overvoltage, overcurrent, and short-circuit protection logic. When an instruction conflicts with the hardware protection logic, the hardware protection logic unconditionally takes precedence, ensuring the absolute safety of the equipment and the grid.

[0035] The third stage is the cluster collaborative response stage, which runs in parallel with the grid-connected operation basic control stage in the background. It does not affect the normal operation of the basic safety control tasks. When the system is in the grid-connected operation basic control stage, the cluster collaborative adaptation module synchronously collects grid-connected point voltage data every power frequency cycle. When a continuous change in the voltage steady-state trend is detected, it automatically enters this stage to execute the corresponding logic. First, the steady-state trend extraction unit uses an exponentially weighted moving average model to smooth the continuously collected grid-connected point voltage sequence. By weighted summing the current measured voltage value with the previous steady-state estimate, it outputs the real-time voltage steady-state trend estimate. Simultaneously, it recursively calculates the voltage fluctuation standard deviation estimate, effectively filtering out transient voltage disturbances caused by local load start-up and shutdown and photovoltaic instantaneous output fluctuations, and accurately extracting the global voltage steady-state change trend of the distribution network. Subsequently, the offset determination unit uses the 3σ statistical criterion, with a 30s sliding time window, to calculate the voltage steady-state trend offset within the window. Simultaneously, it calculates the mean of the standard deviation of voltage fluctuations within the window. When the offset is greater than or equal to 3 times the standard deviation, the voltage change is determined to be a valid steady-state voltage regulation command issued by the upstream power grid, rather than a local random disturbance. If the offset does not reach the determination threshold, it is determined to be a normal voltage fluctuation, and the system exits this stage and returns to continuous monitoring. After completing the valid command determination, the system synchronously verifies the valid range of the offset. In this embodiment, only when the offset is within the preset range of 0.02~0.07pu will it enter the subsequent rule matching stage. Offsets below this range are considered normal power grid fluctuations, and offsets above this range are considered power grid fault conditions. Neither triggers cluster coordinated action, avoiding secondary power grid fluctuations caused by erroneous adjustments. Finally, the rule matching unit uses a cosine similarity algorithm to match the valid steady-state voltage regulation command with the standard voltage regulation command in the preset cluster regulation rule library, calculates the cosine similarity between the two, selects the cluster regulation coefficient corresponding to the rule with the highest similarity, and adds this coefficient to the current photovoltaic output benchmark value to generate the final cluster regulation matching command. This command is then sent to the grid-connected self-regulatory control module for execution, realizing synchronous and coordinated response with other photovoltaic nodes in the county. No data interaction or real-time communication link between any nodes is required, thus achieving communication-free full-domain cluster coordination.

[0036] The fourth stage is the adaptive iterative optimization stage. This stage is triggered when the system is in a grid-connected steady-state operation, the effective operating data accumulated in the local non-volatile storage unit meets a preset threshold for more than 30 consecutive days, and the average voltage control deviation of 100 consecutive sampling points exceeds the preset range of ±3% of the rated voltage. This stage is only executed under grid steady-state conditions. When the system detects a grid fault, voltage exceeding limits, or transient disturbance, it immediately terminates the iteration, restores the default parameters, and prioritizes grid safety. Upon entering the iteration stage, the system first initializes the iteration parameters, using the currently effective adaptation layer parameters in the fusion rule base as the initial iteration values. It sets the initial attenuation step size to 0.02, the fixed attenuation coefficient to 0.01, the maximum number of iterations to 50, and the convergence threshold to 10⁻⁴. Simultaneously, it delineates the compliant and feasible domain of parameter iteration using the threshold of the fixed compliance constraint rules as the boundary, ensuring that the iteration process does not violate the hard constraints of grid safety. After initialization, the system retrieves historical operating data from local storage, aiming to maximize the local photovoltaic grid integration rate. Voltage control deviation is used as a hard constraint, meaning the iterated parameters must not cause the number of voltage exceedances to exceed a preset threshold. The gradient of the objective function is calculated, and gradient truncation is performed simultaneously to avoid gradient explosion. After completing parameter update calculations along the gradient descent direction, the updated parameters are constrained using a feasible region projection operator. If a parameter exceeds the feasible region, it is forcibly pulled back to the feasible region boundary, ensuring that the iterated parameters always meet safety requirements. After a single parameter update, the system checks whether the parameter change is less than the convergence threshold. If the convergence condition is met, the iteration terminates; otherwise, the iteration step size is updated using a fixed attenuation coefficient, and the next iteration cycle begins until the parameters converge or the maximum number of iterations is reached, avoiding unnecessary iterations that consume controller computing power. After the iteration is completed, the system locks the optimized adaptation layer parameters and writes them into the integrated rule base storage module, overwriting the original adaptation layer parameters and completing the local closed-loop update of the rule base. The updated parameters take effect immediately and are applied to subsequent load-solar synergy and revenue optimization control, continuously optimizing the photovoltaic absorption effect and operating economy of the system. The entire process does not require cloud data backhaul, manual intervention, or professional operation and maintenance, adapting to the characteristics of low operation and maintenance capabilities in county-level scenarios.

[0037] The four stages form a complete closed-loop operation system. The cold start stage completes the compliance preparation before grid connection. The grid-connected operation basic control stage realizes normalized safety management and economic optimization. The cluster collaborative response stage responds to the global grid dispatching needs. The adaptive iterative optimization stage continuously optimizes the system operation performance. Each stage is closely linked and logically connected. The entire process is completed within the inverter's local controller, realizing local closed-loop management of the entire process of distributed photovoltaic power generation and grid coordination.

[0038] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system, characterized in that, include: The integrated rule base storage module stores a three-layer nested standardized self-regulatory rule base. This three-layer nested rule base is a set of rules nested from high to low execution priority. The rule base uses the per-unit value of grid connection voltage, the average local active load, and the rated capacity of grid-connected photovoltaic power as core variables, and nests fixed compliance constraint rules, load-solar synergy rules, and revenue optimization rules. The fixed compliance constraint rules are hard constraint rules corresponding to the national mandatory standards for distribution network operation, and set fixed thresholds corresponding to the allowable voltage deviation range specified in GB / T 12325 Power Quality Power Supply Voltage Deviation. The pre-grid connection self-service planning module is used to collect measured electrical data at the grid connection point, call the integrated rule base, and output the photovoltaic access capacity boundary determination result and compliance verification report; The grid-connected self-regulation module is used to collect grid connection point voltage data in real time, call the integrated rule base, and output photovoltaic power output adjustment commands and coordinated operation commands for adjustable loads at the same grid connection point. The cluster collaboration and adaptation module is used to identify changes in the base voltage of the distribution network, call the integrated rule base, and output cluster adjustment and matching instructions. The local adaptive iteration module is used to update the adaptation layer parameters of the integrated rule base based on local runtime data. Each module shares the same local core variable system, with no external real-time communication links or global power grid parameter dependencies.

2. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The fixed compliance constraint rules in the integrated rule base storage module have fixed thresholds set according to the national mandatory standards for distribution network operation. The fixed thresholds are the highest priority constraints for all rule execution. The adaptation layer parameters only apply to load-solar synergy rules and revenue optimization rules.

3. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The grid-connected self-regulation module adopts a time-separated dual-cycle execution logic, setting a first execution cycle at the millisecond level and a second execution cycle at the minute level. The first execution cycle generates and issues photovoltaic output adjustment commands, while the second execution cycle generates and issues adjustable load coordinated operation commands. The execution logic of the two cycles is independent of each other, and the execution priority of the first execution cycle is higher than that of the second execution cycle.

4. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The pre-grid connection self-planning module acquires electrical data of the grid connection point through a multi-cycle continuous acquisition method, performs outlier removal and filtering preprocessing on the acquired data, and performs compliance verification and capacity boundary determination based on the preprocessed data.

5. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The cluster collaboration adaptation module includes: The steady-state trend extraction unit uses an exponentially weighted moving average model to smooth the continuously collected grid-connected point voltage series and outputs the voltage steady-state trend estimate and the voltage fluctuation standard deviation estimate. The offset determination unit uses the 3σ statistical criterion to calculate the voltage steady-state trend offset within a preset sliding time window, compares the offset with the numerical relationship of 3 times the standard deviation of voltage fluctuation, and outputs the steady-state voltage regulation command determination result. The rule matching unit uses a cosine similarity algorithm to match the valid steady-state voltage regulation commands with the preset cluster regulation rule library, selects the cluster regulation coefficient corresponding to the rule with the highest similarity, and outputs it to the grid-connected self-regulation module.

6. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The local adaptive iteration module sets fixed trigger conditions and iteration boundaries. The trigger conditions are set based on the deviation between local data and the benchmark value within a continuous running cycle, and the iteration boundaries correspond to the threshold range of fixed compliance constraint rules.

7. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 6, characterized in that, The local adaptive iterative module uses bounded gradient descent logic constrained by the feasible region projection operator to perform parameter updates, sets the decay step size, maximum number of iterations and convergence threshold, and the entire iteration process is executed within the inverter's local controller without external data interaction.

8. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The load-light coordination rules in the integrated rule base storage module are configured with adjustable time intervals for agricultural irrigation loads connected to the same grid connection point. The adjustable time intervals are matched with the pre-stored time-of-use electricity price time-series data.

9. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 8, characterized in that, The time-of-use electricity price time-series data is updated offline at fixed intervals. After local verification, the updated time-series data is stored in the integrated rule base storage module. The load-solar coordination rule adjusts the matching logic of the adjustable time-series interval based on the updated time-series data.

10. The low-load-density county-level distributed photovoltaic power generation and grid coordination planning and configuration system according to claim 1, characterized in that, The pre-grid connection self-planning module generates standardized verification documents based on capacity boundary determination results and compliance verification data. These verification documents are stored in the inverter's local controller.

Images