A Visual Iterative Design Method for Islanded New Energy Grid Control and Protection Devices
By using a visual iterative design method, a source-grid-load-storage scenario model library and a control and protection algorithm component library are established. A control and protection strategy topology is formed and verified, and real-time control and protection algorithm code is generated. This solves the problems of long development cycle and poor adaptability of control and protection devices for isolated new energy power grids, and realizes control and protection devices with fast response and high reliability.
Patent Information
- Application Number
- CN202611084422.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-21
- Publication Date
- 2026-11-17
AI Technical Summary
Existing isolated renewable energy grid control and protection devices have long development cycles, low strategy reuse, poor fixed logic adaptability, and insufficient control and protection coordination verification, making it difficult to meet the real-time requirements of rapid control and protection actions.
A visual iterative design approach is adopted. By establishing a source-grid-load-storage scenario model library, calling the control and protection algorithm component library, a control and protection strategy topology is formed. Then, intermediate representation model conversion and task graph generation are performed. Combined with model-in-the-loop, software-in-the-loop, processor-in-the-loop, and hardware-in-the-loop verification, real-time control and protection algorithm code is generated, realizing a closed loop of strategy design, verification, code generation, and operation feedback.
It shortened the strategy development and adjustment cycle, improved the real-time performance, adaptability and reliability of isolated new energy grid control and protection devices, enhanced the reusability of different new energy power plants and energy storage routes, reduced on-site commissioning risks, and enhanced rapid response capability and operational reliability.
Smart Images

Figure CN123121786A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of control and protection technology of new energy power systems. Specifically, it relates to a visual iterative design method for control and protection devices of isolated new energy power grids. Background Technology
[0002] As the proportion of new energy equipment such as photovoltaic, wind power, and energy storage in regional power grids continues to increase, some new energy power stations, due to their remote locations, weak connections to the main grid, or special operating modes, are prone to forming small, isolated power grids when the grid is abnormal. The energy supply of isolated new energy power grids mainly depends on the dynamic balance between local wind, solar, energy storage, and loads. Their operating status changes rapidly, they have many control and protection targets, and their anomaly response windows are short.
[0003] Existing secondary devices at renewable energy power plants mostly use traditional measurement, control, and protection devices. These devices typically employ fixed logic or custom-designed code, and their strategy design, code writing, debugging, verification, and on-site commissioning rely on specialized developers, resulting in long development cycles. Furthermore, when the type of renewable energy, energy storage technology, communication protocols, or scheduling requirements change, they need to be redeveloped.
[0004] Meanwhile, isolated renewable energy scenarios are subject to various disturbances, including voltage drops, output fluctuations, overcharging and over-discharging of energy storage, communication link interruptions, and grid-connected / islanded switching. Traditional secondary devices struggle to perform coordinated verification of control strategies, protection criteria, and action interlocking on a single platform. Some devices exhibit significant response delays, failing to fully meet the requirements for rapid control and protection actions under extreme disturbances.
[0005] Visual programming technology can lower the barrier to strategy configuration, but general visualization platforms lack dedicated power system control and protection algorithm components, source-grid-load-storage scenario models, real-time constraint verification, and closed-loop verification mechanisms, making it difficult to directly support the engineering development of isolated renewable energy grid control and protection devices. Therefore, it is necessary to propose a technical process that can integrate visual strategy design, simulation verification, code generation, field feedback, and iterative iteration. Summary of the Invention
[0006] To address the problems of long development cycles, low strategy reuse, poor fixed logic adaptability, insufficient control and protection collaborative verification, and difficulty in guaranteeing real-time performance of isolated renewable energy grid control and protection devices in existing technologies, this invention provides a visual iterative design method for isolated renewable energy grid control and protection devices. This method can form a closed loop of control and protection strategy design, verification, code generation, and online feedback, thereby reducing customized development costs, shortening strategy adjustment cycles, and improving the real-time performance, adaptability, and reliability of isolated renewable energy grid control and protection devices.
[0007] The technical solution provided by this invention is a visual iterative design method for an islanded renewable energy grid control and protection device, comprising the following steps:
[0008] S1: Collect or import topology parameters, equipment parameters, operation waveforms and fault samples of isolated new energy power grids to establish a source-grid-load-storage scenario model library;
[0009] S2: In the visual design terminal, call the control and protection algorithm component library, and form the control and protection strategy topology by dragging and connecting. The control and protection strategy topology includes at least data acquisition, signal preprocessing, operation status identification, control decision, protection criteria, action exit and communication interaction modules.
[0010] S3: Following the order of component node parsing, connection mapping, state machine extraction, and parameter table archiving, the control and protection strategy topology is converted into an intermediate representation model, and a task graph is generated from the intermediate representation model according to the sampling period, action priority, and hardware resource constraints. Static verification is performed on the physical quantity dimension, sampling period, task priority, action interlocking relationship, and hardware resource occupation.
[0011] S4: Based on the source-grid-load-storage scenario model library, perform model-in-the-loop, software-in-the-loop, processor-in-the-loop, or hardware-in-the-loop verification to obtain verification indicators including voltage deviation, frequency deviation, control and protection action delay, protection malfunction and leakage risk, and processor load.
[0012] S5: When the verification index does not meet the preset threshold, calculate the parameter sensitivity based on the normalized central difference of the component parameters according to the comprehensive evaluation function, and calculate the component contribution by combining the component trigger count, execution time and output deviation, generate iterative adjustment information, and return to step S2 or step S3 to update the control and protection strategy topology or task graph.
[0013] S6: When the verification indicators meet the preset threshold, the code, protection setting table, communication mapping table, hardware abstraction layer adaptation table and version signature for operation of the control and protection device are automatically generated.
[0014] S7: Deploy the generated results to the control and protection device, collect action records, waveform data and operation logs online, and write them back to the source-grid-load-storage scenario model library to trigger the next round of visualization iteration.
[0015] Furthermore, the process of establishing the source-grid-load-storage scenario model library in step S1 includes: normalizing the topology parameters and equipment parameters into an initial state x0 and a parameter set θ. s Extract disturbance type and disturbance time t from the waveform recording or fault sample. s The duration τ and the disturbance amplitude Δ; each scene is encapsulated into a scene model M. s Then, based on the scene model, the desired action a is generated. s (t) and qualification criteria Γ s .
[0016] Where x0 represents the initial running state of the scene, θ s For equipment and control parameters, d s (t) is the perturbation function, a s (t) represents the desired action sequence, Γ s This serves as the qualification criterion. For step-type or continuous disturbances, the following unified disturbance function is adopted:
[0017] The photovoltaic output disturbance model and the wind power output disturbance model are established according to the following formulas:
[0018] The energy storage state of charge model is updated discretely according to the charging and discharging power:
[0019] The load step model and the low voltage ride-through model are established according to the following formulas:
[0020] The short-circuit fault model, communication interruption model, and grid-connected / islanded handover model are established according to the following formulas:
[0021] Where H(·) is the unit step function, P pv0 P w0 P l0 For the power before the disturbance, ΔP pv ΔP w ΔP l E represents the disturbance amplitude. n For the rated capacity of energy storage, η c and η d These represent the charge / discharge efficiency, U th and Z th These represent the equivalent source voltage and equivalent impedance, R. f +jX f For fault impedance, γ c (t) is the communication availability flag, B g (t) represents the grid connection switch status.
[0022] Furthermore, the running status of the source-grid-load-storage scenario model library in step S1 is represented as follows: ,
[0023] Among them, U abc For three-phase voltage, I abc For three-phase current, P pv To contribute to photovoltaic power, P w For wind power output, P l Where is the load power, SOC is the energy storage state of charge, B is the switching state, and Q is the communication quality.
[0024] Furthermore, in step S2, the control and protection algorithm component library is called in the visual design terminal to form the control and protection strategy topology: ,
[0025] Wherein, V is the set of control and protection components, and E is the data or event connection between components. The control and protection components include sampling components, data cleaning components, RMS value calculation components, dq coordinate transformation components, phase-locked loop components, power calculation components, transient quantity extraction components, differential protection components, adaptive overcurrent protection components, islanding detection components, source-storage coordination components, low voltage ride-through components, energy storage power limiting components, switch output components, and communication interaction components.
[0026] Each component is described using a unified metamodel:
[0027] Among them, ID i For component identification, In i For input port, Out i For output ports, P i Ts is a set of parameters. i For the sampling period, Pr i Cost is the priority of the action. i For the execution cost, C i As a constraint, this meta-model enables the user's graphical configuration in the interface to be explicitly converted into verifiable and executable computation nodes by the backend.
[0028] Furthermore, the process of converting the control and protection strategy topology into an intermediate representation model in step S3 includes: reading the component nodes and connections in the control and protection strategy topology; and parsing each component node into a data flow block b. i Each connection is resolved into a data link with physical quantity units, data types, and transmission delays. ij The logic for protecting exits, interlocking, and mode switching is decomposed into a state machine a. k Write the component parameters and constants into the parameter table Θ; thereby forming the intermediate representation model G. m . , ,
[0029] The process of generating a task graph based on the intermediate representation model includes: according to the sampling period T s Action priority P r Cluster the data stream blocks to form real-time task T r Determine the message queue q between different tasks based on the data link. rs Alternatively, shared memory can be used; based on task execution cost, hardware resource consumption, and input / output channel constraints, tasks are mapped to management cores, processing cores, or hardware acceleration units, forming a task graph G. e . ,
[0030] Where B is the set of data stream blocks, L is the set of data links, A is the set of state machines, Θ is the parameter table, T is the set of real-time tasks, Q is the set of task communication, and H is the set of data stream blocks. map This represents the hardware mapping relationship, where IO is a set of input / output channels.
[0031] The task graph adopts a multi-timescale heterogeneous scheduling mechanism, configuring protection criteria, switch outputs, and transient quantity extraction as millisecond-level high-priority tasks, voltage and frequency control and source-storage coordination as fast control tasks, and visualization communication, data recording, and parameter management as low-priority management tasks. Data synchronization between the management core, computing core, and hardware acceleration unit is achieved through shared memory, message queues, or event flags.
[0032] The static verification includes checking whether the data source is complete, whether the sampling period meets the transmission relationship from fast to slow, whether the protection action has a higher priority than the control regulation, whether the exit action has mutual exclusion and blocking, and whether the processor, memory and communication bandwidth exceed the limits.
[0033] Furthermore, in step S4, model-in-the-loop verification is used to quickly filter algorithm logic; software-in-the-loop verification is used to check the consistency between the generated code and the model logic; processor-in-the-loop verification is used to check the computation time, memory usage, and numerical errors on the target processor; and hardware-in-the-loop verification is used to check the real-time response of sampling, output, communication, and field interfaces.
[0034] The verification index is calculated using the comprehensive evaluation function J:
[0035] Where ΔU is the maximum voltage deviation, ΔF is the maximum frequency deviation, and T a To control the delay of the protection action, T lim E is the action delay limit. m R is an indicator of the risk of accidental or missed actions. c The processor resource utilization rate is represented by w1 to w5, which are weighting coefficients. If J is less than the overall threshold and each individual indicator is less than its corresponding threshold, then the current strategy is deemed to have passed the verification.
[0036] Furthermore, in step S5, the parameter sensitivity is obtained by performing normalized central difference calculation on the comprehensive evaluation function J. For the j-th parameter p of the i-th component... ij Within its allowable range, the perturbation step size Δp is selected. ij Keeping other parameters constant, calculate J(p) respectively. ij +Δp ij ) and J(p ij -Δp ij), thus obtaining the parameter sensitivity S ij :
[0037] Component contribution C i The calculation is based on a weighted sum of the number of scene triggers, component execution time, component output deviation, and the sensitivity of the component parameters, and iterative adjustment information is generated according to the component contribution from high to low.
[0038] Where, N i t represents the number of times the component is triggered in scenarios where validation fails. i For component execution time, e i N represents the component output deviation. max t max and e max These are the normalized benchmarks for the corresponding indicators, with λ1 to λ4 being the weighting coefficients and λ1+λ2+λ3+λ4=1. C i The larger the value, the more likely the component is to be a weak component that causes the verification metric to fail to meet the threshold.
[0039] Furthermore, upon successful verification in step S6, the code generator generates real-time control and protection algorithm code, communication configuration code, protection setting table, communication mapping table, HAL driver adapter table, and test cases based on the intermediate representation model and task graph. The HAL driver adapter table maps sampling channels, switch outputs, Ethernet communication, serial communication, GOOSE messages, Modbus data points, PCS control interfaces, and local storage interfaces to the abstract ports in the generated code.
[0040] The version signature is generated by the control and protection strategy topology, intermediate representation model, protection setting table, test scenario number and compilation timestamp. After deployment, the control and protection device records the current running version, parameter version and test coverage version. When the on-site verification is abnormal or the strategy needs to be rolled back, it can be traced and rolled back based on the version signature.
[0041] Furthermore, in step S7, online write-back includes writing the field voltage and current waveforms, switch changes, protection action reasons, control output, communication quality, and processor load into the running sample according to the event number, and comparing the deviation between the running sample and the simulation sample; when the deviation exceeds the model correction threshold, a model update task or strategy re-verification task is automatically generated to form the next round of visualization iteration.
[0042] Furthermore, the method also includes a protection-control collaborative arbitration mechanism. When the protection component and the control component issue action commands to the same device within the same time window, the arbitration module reads the action priority, device lockout status, islanded operation status, available energy storage capacity, and communication quality to generate an action token. The component that obtains the action token outputs the final action command, while the component that does not obtain the action token enters a suppression, degraded control, or waiting state, thereby avoiding conflicts between protection tripping, control support, and energy storage power limiting.
[0043] Compared with the prior art, the present invention has the following beneficial effects:
[0044] (1) Form a closed loop for the demand configuration, strategy design, simulation verification, code generation and operation feedback of isolated new energy grid control and protection devices, and shorten the strategy development and adjustment cycle;
[0045] (2) Improve the reusability of different new energy power stations, different energy storage routes and different communication protocols through component libraries and scenario libraries;
[0046] (3) Real-time performance and resource constraints are checked in advance through intermediate representations and task diagrams to reduce the risk of on-site debugging;
[0047] (4) Improve the rapid response capability and action reliability in islanded operation through protection-control collaborative arbitration and multi-time scale scheduling mechanism. Attached Figure Description
[0048] Figure 1 This is a flowchart illustrating the overall process of a visual iterative design method for an islanded renewable energy grid control and protection device according to an embodiment of the present invention.
[0049] Figure 2 This is a schematic diagram of the system structure of the visual iterative design platform and control and protection device in an embodiment of the present invention;
[0050] Figure 3 This is a schematic diagram of the mapping between the visualization strategy topology, intermediate representation model, and task graph in an embodiment of the present invention;
[0051] Figure 4 This is a schematic diagram of the closed-loop verification mechanism of model-in-the-loop, software-in-the-loop, processor-in-the-loop, and hardware-in-the-loop in an embodiment of the present invention;
[0052] Figure 5 This is a schematic diagram of the protection-control collaborative arbitration mechanism in an embodiment of the present invention;
[0053] Figure 6 This is a schematic diagram of the topology of an isolated new energy power grid test scenario in an embodiment of the present invention;
[0054] Figure 7These are simulation waveforms of the effective voltage response at the grid connection point under different design strategies in embodiments of the present invention.
[0055] Figure 8 This is a comparison diagram of end-to-end delay of control and protection actions in an embodiment of the present invention;
[0056] Figure 9 This is the convergence curve of key verification indicators during the visualization iteration process in this embodiment of the invention;
[0057] Figure 10 This is a curve showing the variation of power generation, storage, and load in the isolated renewable energy grid and the state of charge (SOC) of energy storage in an embodiment of the present invention. Detailed Implementation
[0058] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention.
[0059] like Figure 1 As shown, the visual iterative design process of this invention starts with the isolated new energy power grid operation scenario and ends with the on-site operation of the control and protection device. First, a source-grid-load-storage scenario model library and a control and protection algorithm component library are established. Second, the user configures the control and protection strategy topology in the visual design terminal. Then, the strategy topology is automatically converted into an intermediate representation model and task graph, and static verification is performed on physical quantities, time scales, action priorities, and hardware resources. Quantitative indicators are then obtained through model-in-the-loop, software-in-the-loop, processor-in-the-loop, or hardware-in-the-loop verification. If the verification does not meet the threshold, weak components are located and the process returns to the visual design stage for local iteration. If the verification meets the threshold, code, a setpoint table, a communication mapping table, a HAL adaptation table, and a version signature are generated, and after execution, event recordings and logs are written back to the model library. This method is not simply a graphical programming interface, but an engineering design process that unifies graphical strategies, model verification, real-time task constraints, code generation, and runtime feedback.
[0060] In one embodiment, an islanded renewable energy grid includes photovoltaic inverters, wind power converters, energy storage batteries and PCS, critical loads, grid-connected switches, lines, voltage and current transformers, communication links, and control and protection devices. The control and protection devices include analog sampling modules, digital input / output modules, communication modules, a management core, a computational core, a hardware acceleration unit, a storage unit, and a hardware abstraction layer. The management core is used for communication, configuration, visualization strategy parsing, and version management; the computational core is used for real-time control and protection algorithm calculation; and the hardware acceleration unit is used for parallel execution of transient quantity extraction, filtering, comparison, and output logic.
[0061] A visual iterative design method for islanded renewable energy grid control and protection devices is applied to an islanded renewable energy grid that includes renewable energy power generation equipment, energy storage equipment, loads, grid-connected / islanded switching switches, control and protection devices, and a visual design terminal. The method includes the following steps:
[0062] S1: Collect or import topology parameters, equipment parameters, operation waveforms and fault samples of isolated new energy power grids to establish a source-grid-load-storage scenario model library;
[0063] S2: In the visual design terminal, call the control and protection algorithm component library, and form the control and protection strategy topology by dragging and connecting. The control and protection strategy topology includes at least data acquisition, signal preprocessing, operation status identification, control decision, protection criteria, action exit and communication interaction modules.
[0064] S3: Following the order of component node parsing, connection mapping, state machine extraction, and parameter table archiving, the control and protection strategy topology is converted into an intermediate representation model, and a task graph is generated from the intermediate representation model according to the sampling period, action priority, and hardware resource constraints. Static verification is performed on the physical quantity dimension, sampling period, task priority, action interlocking relationship, and hardware resource occupation.
[0065] S4: Based on the source-grid-load-storage scenario model library, perform model-in-the-loop, software-in-the-loop, processor-in-the-loop, or hardware-in-the-loop verification to obtain verification indicators including voltage deviation, frequency deviation, control and protection action delay, protection malfunction and leakage risk, and processor load.
[0066] S5: When the verification index does not meet the preset threshold, calculate the parameter sensitivity based on the normalized central difference of the component parameters according to the comprehensive evaluation function, and calculate the component contribution by combining the component trigger count, execution time and output deviation, generate iterative adjustment information, and return to step S2 or step S3 to update the control and protection strategy topology or task graph.
[0067] S6: When the verification indicators meet the preset threshold, the code, protection setting table, communication mapping table, hardware abstraction layer adaptation table and version signature for operation of the control and protection device are automatically generated.
[0068] S7: Deploy the generated results to the control and protection device, collect action records, waveform data and operation logs online, and write them back to the source-grid-load-storage scenario model library to trigger the next round of visualization iteration.
[0069] like Figure 2 As shown, the visual iterative design platform includes a visual design terminal, a source-network-load-storage scenario model library, a control and protection algorithm component library, an intermediate representation converter, a verification engine, a code generator, a HAL adapter, and a runtime feedback manager. The visual design terminal displays the control and protection logic through graphical modules, allowing users to create policy topologies by dragging and dropping and connecting lines. The scenario model library provides disturbance samples. The verification engine schedules model-in-the-loop, software-in-the-loop, processor-in-the-loop, and hardware-in-the-loop tests. The code generator converts verified policies into code and configuration files.
[0070] In step S1, topology parameters, equipment parameters, operational waveforms, and fault samples of the isolated renewable energy power grid are collected or imported to establish a source-grid-load-storage scenario model library. The operational state can be represented as:
[0071] Among them, U abc For three-phase voltage, I abc For three-phase current, P pv To contribute to photovoltaic power, P w For wind power output, P l Where is the load power, SOC is the energy storage state of charge, B is the switching state, and Q is the communication quality.
[0072] Each scenario in the source-grid-load-storage scenario model library includes a scenario number, initial operating mode, disturbance type, disturbance time, disturbance amplitude, duration, expected action, qualification threshold, and waveform recording template. During model building, the topology parameters and equipment parameters are first normalized to an initial state x0 and a parameter set θ. s Extract disturbance type and disturbance time t from the waveform recording or fault sample. s The duration τ and the disturbance amplitude Δ; each scene is encapsulated into a scene model M. s Then, based on the scene model, the desired action a is generated. s (t) and qualification criteria Γ s They also developed photovoltaic power output disturbance models, wind power output disturbance models, energy storage state of charge models, load step models, low voltage ride-through models, short circuit fault models, communication interruption models, and grid-connected / islanding switching models, respectively. ,
[0073] Where x0 represents the initial running state of the scene, θ s For equipment and control parameters, d s (t) is the perturbation function, a s (t) represents the desired action sequence, Γs This serves as the qualification criterion. For step-type or continuous disturbances, the following unified disturbance function is adopted: ,
[0074] The photovoltaic output disturbance model and the wind power output disturbance model are established according to the following formulas: ,
[0075] The energy storage state of charge model is updated discretely according to the charging and discharging power: ,
[0076] The load step model and the low voltage ride-through model are established according to the following formulas: ,
[0077] The short-circuit fault model, communication interruption model, and grid-connected / islanded handover model are established according to the following formulas: ,
[0078] Where H(·) is the unit step function, P pv0 P w0 P l0 For the power before the disturbance, ΔP pv ΔP w ΔP l E represents the disturbance amplitude. n For the rated capacity of energy storage, η c and η d These represent the charge / discharge efficiency, U th and Z th These represent the equivalent source voltage and equivalent impedance, R. f +jX f For fault impedance, γ c (t) is the communication availability flag, B g (t) represents the grid connection switch status.
[0079] For example, in the voltage drop scenario, the time, duration, and low voltage ride-through or islanding switching actions that should be initiated when the grid connection voltage drops to 0.8 times the rated voltage are recorded; in the communication interruption scenario, the main communication link interruption time and the backup link switching criteria are recorded.
[0080] In step S2, the control and protection algorithm component library is called in the visual design terminal to form the control and protection strategy topology:
[0081] Where V represents the set of control and protection components, and E represents the data or event connections between components. Control and protection components include, but are not limited to, sampling components, data cleaning components, RMS calculation components, dq coordinate transformation components, phase-locked loop components, power calculation components, transient quantity extraction components, differential protection components, adaptive overcurrent protection components, islanding detection components, source-storage coordination components, low-voltage ride-through components, energy storage power limiting components, switch output components, and communication interaction components.
[0082] Each component is described using a unified metamodel:
[0083] Among them, ID i For component identification, In i For input port, Out i For output ports, P i Ts is a set of parameters. i For the sampling period, Pr i Cost is the priority of the action. i For the execution cost, C i These serve as constraints. Through this metamodel, the graphical configurations displayed by the user in the interface can be explicitly converted by the backend into verifiable and executable computation nodes.
[0084] In step S3, the control and protection strategy topology is converted into an intermediate representation model, and then the intermediate representation model is mapped into a task graph. The specific process is as follows.
[0085] The process of converting the control and protection strategy topology into an intermediate representation model includes: reading the component nodes and connections in the control and protection strategy topology; and parsing each component node into a data flow block b. i Each connection is resolved into a data link with physical quantity units, data types, and transmission delays. ij The logic for protecting exits, interlocking, and mode switching is decomposed into a state machine a. k Write the component parameters and constants into the parameter table Θ; thereby forming the intermediate representation model G. m . , ,
[0086] The process of generating a task graph based on the intermediate representation model includes: according to the sampling period T s Action priority P r Cluster the data stream blocks to form real-time task T r Determine the message queue q between different tasks based on the data link. rs Alternatively, shared memory can be used; based on task execution cost, hardware resource consumption, and input / output channel constraints, tasks are mapped to management cores, processing cores, or hardware acceleration units, forming a task graph G. e . ,
[0087] Where B is the set of data stream blocks, L is the set of data links, A is the set of state machines, Θ is the parameter table, T is the set of real-time tasks, Q is the set of task communication, and H is the set of data stream blocks. map This represents the hardware mapping relationship, where IO is a set of input / output channels. Figure 3 The above three-layer mapping relationship is shown.
[0088] During the conversion process, the system performs static verification. This static verification includes: checking whether the input signal has a unique source; checking whether physical quantities such as voltage, current, power, frequency, and SOC are consistent; checking whether the output of the fast sampling component is incorrectly consumed by the slow sampling task; checking whether the priority of protection actions is higher than that of conventional control regulation; checking whether the same switch or the same energy storage PCS receives mutual exclusion commands; and checking whether the management core, computing core, hardware acceleration unit, memory, and communication bandwidth exceed limits.
[0089] In step S4, the verification engine performs verification one by one according to the scenario model library. Model-in-the-loop verification is used to quickly filter algorithm logic; software-in-the-loop verification is used to check the consistency between the generated code and the model logic; processor-in-the-loop verification is used to check the computing time, memory usage, and numerical errors on the target processor; hardware-in-the-loop verification is used to check the real-time response of sampling, output, communication, and field interfaces. Figure 4 The closed-loop verification mechanism is shown.
[0090] The verification index is calculated using the comprehensive evaluation function J:
[0091] Where ΔU is the maximum voltage deviation, ΔF is the maximum frequency deviation, and T a To control the delay of the protection action, T lim E is the action delay limit. m R is an indicator of the risk of accidental or missed actions. c The processor resource utilization rate is represented by w1 to w5, which are weighting coefficients. If J is less than the overall threshold and each individual indicator is less than its corresponding threshold, the current strategy is considered to have passed the validation.
[0092] In step S5, when the verification fails, the system locates the weak component based on the component contribution and parameter sensitivity. The specific process is as follows.
[0093] The parameter sensitivity is obtained by normalizing the central difference of the comprehensive evaluation function J. For the j-th parameter p of the i-th component... ij Within its allowable range, the perturbation step size Δp is selected. ij Keeping other parameters constant, calculate J(p) respectively. ij +Δp ij ) and J(pij -Δp ij ), thus obtaining the parameter sensitivity S ij :
[0094] Component contribution C i The calculation is based on a weighted sum of the number of scene triggers, component execution time, component output deviation, and the sensitivity of the component parameters:
[0095] Where, N i t represents the number of times the component is triggered in scenarios where validation fails. i For component execution time, e i N represents the component output deviation. max t max and e max These are the normalized benchmarks for the corresponding indicators, with λ1 to λ4 being the weighting coefficients and λ1+λ2+λ3+λ4=1. C i The larger the value, the more likely the component is to be a weak component that causes the verification metric to fail to meet the threshold.
[0096] like Figure 5 As shown, this invention also includes a protection-control collaborative arbitration mechanism. When the protection component and the control component issue action commands to the same device within the same time window, the arbitration module reads the action priority, device lockout status, islanded operation status, available energy storage capacity, and communication quality to generate an action token. The component that obtains the action token outputs the final action command, while the component that does not obtain the action token enters a suppression, degraded control, or waiting state. This avoids conflicts between protection tripping, control support, and energy storage power limiting.
[0097] In step S6, when verification is successful, the code generator generates real-time control and protection algorithm code, communication configuration code, protection setting table, communication mapping table, HAL driver adapter table, test cases, and version signature based on the intermediate representation model and task graph. The HAL driver adapter table maps sampling channels, switch outputs, Ethernet communication, serial communication, GOOSE messages, Modbus data points, PCS control interfaces, and local storage interfaces to the abstract ports in the generated code.
[0098] The version signature is generated jointly by the protection strategy topology, intermediate representation model, protection setting table, test scenario number, and compilation timestamp. After deployment, the protection device records the current running version, parameter version, and test coverage version; when on-site verification fails or the strategy needs to be rolled back, it can be traced and rolled back based on the version signature.
[0099] In step S7, the control and protection device collects action records, waveform data, and operation logs online, including voltage and current waveforms, switch changes, reasons for protection actions, control outputs, communication quality, processor load, and storage usage. The operation feedback manager writes the above data into the operation sample according to the event number and compares the deviation with the simulation sample. When the deviation exceeds the model correction threshold, a model update task or strategy re-verification task is automatically generated, forming the next round of visualization iteration.
[0100] In a specific example, the test scenario is as follows: Figure 6 As shown, the system includes photovoltaics, wind turbines, energy storage, loads, and grid-connected switches. The scenario applies a load step at 0.40s, switches to islanded operation at 0.90s, applies a fault disturbance at 1.25s, and clears the fault at 1.50s. Figure 7 The effective value response of the grid connection point voltage of the fixed logic device and the visualization iterative device of the present invention is shown in this scenario. Figure 7 It is understood that the embodiments of the present invention can reduce the voltage drop and enable faster recovery after the fault is cleared.
[0101] Figure 8 The invention demonstrates a comparison of end-to-end latency for control and protection actions. The median and 95th percentile latency of fixed logic devices are relatively high, while this invention reduces end-to-end latency to a preset fast action range through strategy preloading, heterogeneous scheduling, and static resource verification. Figure 9 The process of the maximum voltage deviation, malfunction risk index, and action delay gradually converging as the number of iterations increases is shown, demonstrating that the present invention can drive strategy optimization with quantitative indicators. Figure 10 The changes in photovoltaic, wind power, energy storage, load, and SOC during islanded operation are shown, illustrating that the present invention can simultaneously consider source-storage-load balance and energy storage constraints in the design of control and protection strategies.
[0102] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific embodiments of the present invention, and any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.
Claims
1. A visual iterative design method for an islanded renewable energy grid control and protection device, characterized in that, Includes the following steps: S1: Collect or import topology parameters, equipment parameters, operation waveforms and fault samples of isolated new energy power grids to establish a source-grid-load-storage scenario model library; S2: In the visual design terminal, call the control and protection algorithm component library, and form the control and protection strategy topology by dragging and connecting. The control and protection strategy topology includes at least data acquisition, signal preprocessing, operation status identification, control decision, protection criteria, action exit and communication interaction modules. S3: Following the order of component node parsing, connection mapping, state machine extraction, and parameter table archiving, the control and protection strategy topology is converted into an intermediate representation model, and a task graph is generated from the intermediate representation model according to the sampling period, action priority, and hardware resource constraints. Static verification is performed on the physical quantity dimension, sampling period, task priority, action interlocking relationship, and hardware resource occupation. S4: Based on the source-grid-load-storage scenario model library, perform model-in-the-loop, software-in-the-loop, processor-in-the-loop, or hardware-in-the-loop verification to obtain verification indicators including voltage deviation, frequency deviation, control and protection action delay, protection malfunction and leakage risk, and processor load. S5: When the verification index does not meet the preset threshold, calculate the parameter sensitivity based on the normalized central difference of the component parameters according to the comprehensive evaluation function, and calculate the component contribution by combining the component trigger count, execution time and output deviation, generate iterative adjustment information, and return to step S2 or step S3 to update the control and protection strategy topology or task graph. S6: When the verification indicators meet the preset threshold, the code, protection setting table, communication mapping table, hardware abstraction layer adaptation table and version signature for operation of the control and protection device are automatically generated. S7: Deploy the generated results to the control and protection device, collect action records, waveform data and operation logs online, and write them back to the source-grid-load-storage scenario model library to trigger the next round of visualization iteration.
2. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, Step S1 involves establishing the source-grid-load-storage scenario model library, which includes normalizing the topology parameters and equipment parameters into an initial state x0 and a parameter set θ. s Extract disturbance type and disturbance time t from the waveform recording or fault sample. s The duration τ and the disturbance amplitude Δ; each scene is encapsulated into a scene model M. s Then, based on the scene model, the desired action a is generated. s (t) and qualification criteria Γ s They also developed photovoltaic power output disturbance models, wind power output disturbance models, energy storage state of charge models, load step models, low voltage ride-through models, short circuit fault models, communication interruption models, and grid-connected / islanding switching models, respectively. , Where x0 represents the initial running state of the scene, θ s For equipment and control parameters, d s (t) is the perturbation function, a s (t) represents the desired action sequence, Γ s This serves as the qualification criterion. For step-type or continuous disturbances, the following unified disturbance function is adopted: , The photovoltaic output disturbance model and the wind power output disturbance model are established according to the following formulas: , The energy storage state of charge model is updated discretely according to the charging and discharging power: , The load step model and the low voltage ride-through model are established according to the following formulas: , The short-circuit fault model, communication interruption model, and grid-connected / islanded handover model are established according to the following formulas: , Where H(·) is the unit step function, P pv0 P w0 P l0 For the power before the disturbance, ΔP pv ΔP w ΔP l E represents the disturbance amplitude. n For the rated capacity of energy storage, η c and η d These represent the charge / discharge efficiency, U th and Z th These represent the equivalent source voltage and equivalent impedance, R. f +jX f For fault impedance, γ c (t) is the communication availability flag, B g (t) represents the grid connection switch status.
3. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, The running status of the source-grid-load-storage scenario model library in step S1 is represented as follows: , Among them, U abc For three-phase voltage, I abc For three-phase current, P pv To contribute to photovoltaic power, P w For wind power output, P l Where is the load power, SOC is the energy storage state of charge, B is the switching state, and Q is the communication quality.
4. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, In step S2, the control and protection algorithm component library is called in the visual design terminal to form the control and protection strategy topology: , Wherein, V is the set of control and protection components, and E is the data or event connection between components. The control and protection components include sampling components, data cleaning components, RMS value calculation components, dq coordinate transformation components, phase-locked loop components, power calculation components, transient quantity extraction components, differential protection components, adaptive overcurrent protection components, islanding detection components, source-storage coordination components, low voltage ride-through components, energy storage power limiting components, switch output components, and communication interaction components. Each component is described using a unified metamodel: , Among them, ID i For component identification, In i For input port, Out i For output ports, P i Ts is a set of parameters. i For the sampling period, Pr i Cost is the priority of the action. i For the execution cost, C i As a constraint, this meta-model enables the user's graphical configuration in the interface to be explicitly converted into verifiable and executable computation nodes by the backend.
5. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, Step S3, the process of converting the control and protection strategy topology into an intermediate representation model, includes: reading the component nodes and connections in the control and protection strategy topology; and parsing each component node into a data flow block b. i Each connection is resolved into a data link with physical quantity units, data types, and transmission delays. ij The logic for protecting exits, interlocking, and mode switching is decomposed into a state machine a. k Write the component parameters and constants into the parameter table Θ; thereby forming the intermediate representation model G. m ; , , , The process of generating a task graph based on the intermediate representation model includes: according to the sampling period T s Action priority P r Cluster the data stream blocks to form real-time task T r Determine the message queue q between different tasks based on the data link. rs Alternatively, shared memory can be used; based on task execution cost, hardware resource consumption, and input / output channel constraints, tasks are mapped to management cores, processing cores, or hardware acceleration units, forming a task graph G. e ; , , Where B is the set of data stream blocks, L is the set of data links, A is the set of state machines, Θ is the parameter table, T is the set of real-time tasks, Q is the set of task communication, and H is the set of data stream blocks. map This refers to the hardware mapping relationship; IO is a set of input / output channels. The static verification includes checking whether the data source is complete, whether the sampling period meets the transmission relationship from fast to slow, whether the protection action has a higher priority than the control regulation, whether the exit action has mutual exclusion and blocking, and whether the processor, memory and communication bandwidth exceed the limits.
6. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, In step S4, model-in-the-loop verification is used to quickly filter algorithm logic; software-in-the-loop verification is used to check the consistency between the generated code and the model logic; processor-in-the-loop verification is used to check the computation time, memory usage, and numerical errors on the target processor; and hardware-in-the-loop verification is used to check the real-time response of sampling, output, communication, and field interfaces. The verification index is calculated using the comprehensive evaluation function J: , Where ΔU is the maximum voltage deviation, ΔF is the maximum frequency deviation, and T a To control the delay of the protection action, T lim E is the action delay limit. m R is an indicator of the risk of accidental or missed actions. c The processor resource utilization rate is represented by w1 to w5, which are weighting coefficients. If J is less than the overall threshold and each individual indicator is less than its corresponding threshold, then the current strategy is deemed to have passed the verification.
7. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, In step S5, the parameter sensitivity is obtained by performing normalized central difference calculation on the comprehensive evaluation function J. For the j-th parameter p of the i-th component... ij Within its allowable range, the perturbation step size Δp is selected. ij Keeping other parameters constant, calculate J(p) respectively. ij +Δp ij ) and J(p ij -Δp ij ), thus obtaining the parameter sensitivity S ij : , Component contribution C i The calculation is based on a weighted sum of the number of scene triggers, component execution time, component output deviation, and the sensitivity of the component parameters, and iterative adjustment information is generated according to the component contribution from high to low. , Where, N i t represents the number of times the component is triggered in scenarios where validation fails. i For component execution time, e i N represents the component output deviation. max t max and e max These are the normalized benchmarks for the corresponding indicators, λ1 to λ4 are the weighting coefficients and λ1+λ2+λ3+λ4=1, C i The larger the value, the more likely the component is to be a weak component that causes the verification metric to fail to meet the threshold.
8. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, When the verification is successful in step S6, the code generator generates real-time control and protection algorithm code, communication configuration code, protection setting table, communication mapping table, HAL driver adapter table and test cases based on the intermediate representation model and task graph. The HAL driver adapter table maps sampling channels, digital outputs, Ethernet communication, serial communication, GOOSE messages, Modbus data points, PCS control interfaces, and local storage interfaces to the abstract ports in the generated code. The version signature is generated by the control and protection strategy topology, intermediate representation model, protection setting table, test scenario number and compilation timestamp. After deployment, the control and protection device records the current running version, parameter version and test coverage version. When the on-site verification is abnormal or the strategy needs to be rolled back, it can be traced and rolled back based on the version signature.
9. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, In step S7, online write-back includes writing the field voltage and current waveforms, switch position changes, protection action reasons, control outputs, communication quality, and processor load into the running sample according to the event number, and comparing the deviation between the running sample and the simulation sample; when the deviation exceeds the model correction threshold, a model update task or strategy re-verification task is automatically generated to form the next round of visualization iteration.
10. The visual iterative design method for an islanded new energy grid control and protection device according to claim 1, characterized in that, The method also includes a protection-control collaborative arbitration mechanism. When the protection component and the control component issue action commands to the same device within the same time window, the arbitration module reads the action priority, device lockout status, islanded operation status, available energy storage capacity, and communication quality to generate an action token. The component that obtains the action token outputs the final action command, while the component that does not obtain the action token enters a suppression, degraded control, or waiting state.
