AGC instruction anti-error method and system based on digital twinborn and multi-risk look-ahead

Abstract

The invention discloses an AGC instruction anti-error method and system based on digital twinning and multi-risk look-ahead, and belongs to the technical field of automation and control of power systems. The method comprises the following steps: constructing a digital twinborn body which operates synchronously with a physical power system AGC control loop; acquiring operation state data; receiving an original control instruction from a superior AGC system; performing look-ahead simulation on the execution effect of the original control instruction to obtain a system state track prediction value; calculating five-dimension risk indexes by using a risk quantitative evaluator; judging whether any risk index of the five-dimension risk indexes exceeds a preset threshold value or not; if yes, triggering the instruction safety correction module, and executing the next step; if not, directly issuing the original control instruction to a station physical execution unit; an instruction safety correction module is started, rolling optimization is carried out on the original control instruction, and a safety instruction is generated; and issuing the security instruction to a physical execution unit.

Description

Technical Field

[0001] This invention relates to a method and system for preventing errors in AGC (Automatic Generation Control) commands based on digital twins and multi-risk look-ahead, belonging to the field of power system automation and control technology. It relates to a method and system for preventing errors in commands used in automatic generation control (AGC) in power systems with a high proportion of new energy and energy storage. More specifically, it relates to a systematic method based on digital twin technology that performs multi-timescale, multi-risk-dimensional look-ahead simulation and active correction before command execution. Background Technology

[0002] With the large-scale integration of intermittent new energy sources such as wind power and photovoltaics, as well as electrochemical energy storage, into the power grid, the operating characteristics of the power system have undergone fundamental changes. As the core means of maintaining the frequency and power balance of the power grid, Automatic Generation Control (AGC) has shifted its control objects from traditional thermal / hydropower units with known response characteristics to a large number of emerging control subjects with strong uncertainty, low inertia, and limited energy characteristics.

[0003] This makes it difficult for traditional AGC instruction generation mechanisms to adapt to complex and ever-changing system dynamics, facing the following challenges:

[0004] 1. Command mismatch risk: There is a discrepancy between the AGC plan based on ultra-short-term forecasts and the actual output of wind and solar power, resulting in inaccurate initial command benchmarks;

[0005] 2. Risk of misuse of energy storage: The state of charge of energy storage units changes dynamically. If the AGC command does not fully consider its SOC boundary, it is very easy for it to be quickly charged or discharged and exit regulation, which will aggravate system fluctuations.

[0006] 3. System instability risk: New energy units are connected to the grid through inverters, which have weak anti-interference capabilities. Severe AGC commands or system fluctuations may induce large-scale grid disconnection.

[0007] 4. Controlling conflict risks: The response speeds of resources such as thermal power, hydropower, energy storage, and new energy vary greatly, and improper allocation of instructions may cause regulatory oscillations.

[0008] Current AGC control methods consider the SOC of energy storage during power allocation, but fail to perform global risk anticipation and proactive correction at the source of command generation. Existing model predictive control methods also mostly focus on economic efficiency or frequency deviation as optimization objectives, lacking specific modeling and constraints for the aforementioned systemic risks.

[0009] Therefore, there is an urgent need for a method that can conduct comprehensive, multi-risk simulation assessments and proactive corrections before control commands are issued, in order to prevent control errors from the source. Summary of the Invention

[0010] Purpose of the invention: The purpose of this invention is to overcome the shortcomings of the prior art and provide an AGC command error prevention method and system based on digital twins and multi-risk anti-error. It can perform comprehensive, multi-risk simulation assessment and proactive correction before the control command is issued. The risks include not only the impact of the control command on the emerging control subject, but also the impact of the anomaly of the emerging control subject on the power grid security, so as to realize the transformation from "passive response" to "proactive error prevention" and significantly improve the safety and reliability of AGC control in high-proportion new energy power systems.

[0011] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0012] A method for preventing errors in AGC instructions based on digital twins and multi-risk anti-error includes the following steps:

[0013] Step S1: Construct a digital twin that operates synchronously with the physical power system's AGC control loop;

[0014] The digital twin includes a refined model library, a multi-timescale simulation engine, a risk quantification assessor, and an instruction security correction module;

[0015] Step S2: Obtain the operating status data of the power grid wide-area measurement system, monitoring and data acquisition system and the station monitoring system, and drive the digital twin to keep synchronized with the physical system;

[0016] The update cycle of the operating status data is no greater than the simulation step size of the digital twin;

[0017] Step S3: Receive raw control commands from the upper-level AGC system;

[0018] Step S4: In the digital twin, using the multi-timescale simulation engine and based on the refined model library, the execution effect of the original control command in the next K control cycles is simulated in advance to obtain the predicted value of the system state trajectory.

[0019] Step S5: Using the risk quantification evaluator, calculate the risk index in five dimensions based on the system state trajectory prediction value obtained in step S4.

[0020] Step S6: Determine whether any of the five risk indices exceeds its preset threshold; if so, trigger the instruction safety correction module and execute step S7; if not, directly send the original control instruction to the site physical execution unit.

[0021] Step S7: The instruction safety correction module is activated to perform rolling optimization on the original control instructions with the goal of minimizing system risks and instruction deviations, and to generate safe instructions.

[0022] Step S8: The security instruction is sent to the physical execution unit. Before the instruction is sent, a security verification process is performed to ensure that there are no illegal instruction values. After the instruction is sent, the system records all relevant data of this correction event, including the original instruction, security instruction, risk index and optimization process, for post-event analysis and model optimization.

[0023] Furthermore, the digital twin is deployed in a dedicated high-performance computing server in the dispatch control center, and interacts with the power grid wide area measurement system (WAMS), the monitoring and data acquisition system (SCADA), and the station monitoring system through a high-speed data bus to ensure that the initial simulation state is consistent with the physical system.

[0024] Furthermore, the refined model library is used to simulate actual physical systems and consists of power grid structure models, traditional generating units and new energy generating unit models, etc., which are the basic data for calculation by the multi-timescale simulation engine. The multi-timescale simulation engine realizes the simulation prediction of power grid state trajectory based on the refined model library. The risk quantification assessor calculates various risk indices during AGC control based on power grid dynamic simulation data. The command safety correction module corrects the initial AGC control command based on the risk index.

[0025] Furthermore, the operating status data mentioned in step S2 includes at least: the output and switching status of each generator set, the power flow of the main transmission lines, the voltage amplitude and phase angle of all bus nodes, the actual and predicted power of the new energy power station, and the real-time power and state of charge (SOC) of the energy storage power station.

[0026] Furthermore, the look-ahead simulation described in step S4 employs parallel computing technology to accelerate computation and ensure that all K-step simulations are completed within a single AGC control cycle; the simulation step size can be set according to accuracy requirements, typically from 0.1 seconds to 1 second.

[0027] Furthermore, the five dimensions of risk indices mentioned in step S5 include:

[0028] The energy storage exit risk index R_soc is used to quantify the risk of exiting operation due to the state of charge of an energy storage unit exceeding the limit. For each energy storage unit in the system, its SOC prediction trajectory is calculated independently and the risk is assessed. Finally, the maximum value among all energy storage units is taken as the energy storage exit risk index of the system.

[0029] The renewable energy grid disconnection risk index R_trip is used to quantify the risk of renewable energy units disconnecting from the grid due to frequency or voltage exceeding limits.

[0030] The control conflict risk index Rosc is used to quantify the risk of system oscillation caused by mismatch between fast and slow resource responses; the specific frequency band of system oscillation [f_min, f_max] is obtained through offline modal analysis or online identification and is tailored to the oscillation mode between regions of this power grid.

[0031] The system's vulnerability index R_vul is used to quantify the system's frequency stability margin under the worst-case scenario of a single grid disconnection event from a new energy source.

[0032] The AGC adjustment resource depletion risk index R_exh is used to quantify the risk of AGC increasing reserve capacity being depleted when new energy sources experience sustained abnormal fluctuations.

[0033] Furthermore, the energy storage exit risk index R_soc is calculated in the following way:

[0034] First, the risk potential function is defined as follows:

[0035] ;

[0036] Where A, B, C, and D are fitting parameters greater than 0, and SOC_high and SOC_low are the upper and lower warning limits of energy storage SOC.

[0037] Therefore, the maximum value of the energy storage exit risk index within the prediction time domain is as follows:

[0038] ;

[0039] Where k = 1 to K; SOC_simulated(k) is the predicted value of the k-th step of energy storage; K is the total number of control cycles.

[0040] Furthermore, the new energy grid disconnection risk index R_trip is calculated in the following way:

[0041] ,

[0042] Where k = 1 to K; Δf_simulated(k) is the predicted value of the system frequency deviation at the k-th step; Δf_max is the maximum allowable frequency deviation threshold for the new energy unit; Δv_simulated(k) is the predicted value of the system voltage deviation at the k-th step; Δv_max is the maximum allowable voltage deviation threshold for the new energy unit; and K is the total number of control cycles.

[0043] Furthermore, the control conflict risk index R_osc is calculated using the following formula:

[0044] Perform a Fast Fourier Transform (FFT) on the predicted trajectory P_tie(t+1:t+K) of the regional tie line power, and calculate the spectral energy within a specific frequency band [f_min, f_max].

[0045] ;

[0046] Where f is in the range [f_min, f_max].

[0047] Furthermore, the system's vulnerability index R_vul is calculated using the following formula:

[0048] ;

[0049] Where f_nom is the system's rated frequency; f_dip_simulated is the lowest frequency point of the system after the simulated worst-case scenario of the new energy power station being instantly disconnected from the grid; and f_under is the frequency of the first round of low-frequency load shearing.

[0050] Furthermore, the AGC-regulated resource depletion risk index R_exh is calculated using the following formula:

[0051] ;

[0052] Where T_exh = P_ava / (ΔP_def / ΔT), P_ava is the total available up-up reserve capacity of the system, ΔP_def is the assumed continuous power deficit of new energy sources, ΔT is the control period, and a is the time scale coefficient, for example 1 / 60, which means in minutes.

[0053] Furthermore, since the system operation mode and weather conditions can affect the accuracy of new energy forecasting, the preset threshold mentioned in step S6 can be dynamically adjusted according to the system operation mode and weather conditions. For example, in typhoon weather, the new energy grid disconnection risk threshold R_trip_threshold can be appropriately reduced to adopt a more conservative control strategy.

[0054] Furthermore, the rolling optimization described in step S7 is solved using the interior point method or the effective set method, and the result of the previous optimization is used as a warm start to improve computational efficiency.

[0055] Furthermore, in step S7, the rolling optimization solution process is as follows:

[0056] The objective function for establishing the mathematical model is,

[0057]

[0058] Where k = 1 to K; Weighted L2 norm for safety command error; W1 is the weighted L2 norm of the overall system risk; W2 is the instruction deviation weight matrix; W3 is the system risk weight matrix.

[0059] The constraints of the mathematical model include power balance constraints, energy storage security constraints, ramp rate constraints, stability constraints, system vulnerability constraints, and reserve depletion risk constraints.

[0060] Furthermore, the power balance constraint is as follows: for all control units i in the k-th simulation step,

[0061] ,

[0062] Where ΔP_safe_i(k) is the safety instruction of the i-th AGC unit at the k-th simulation step; ΔP_total(k) is the total adjustment requirement of the system at the k-th simulation step;

[0063] The energy storage safety constraints are as follows: For all energy storage units i in the k-th simulation step...

[0064] ,

[0065] Where SOC_i(k) is the state of charge predicted by the i-th energy storage unit at the k-th simulation step; SOC_min is the upper limit of the allowed state of charge for the i-th energy storage unit; and SOC_max is the upper limit of the allowed state of charge for the i-th energy storage unit.

[0066] The ramp rate constraint is as follows: for all elements i at the k-th simulation step size...

[0067] ,

[0068] Where ΔP_safe_i(k) is the safety instruction of the i-th AGC unit at the k-th simulation step; ΔP_safe_i(k-1) is the safety instruction of the i-th AGC unit at the (k-1)-th simulation step; Ramp_max_i is the maximum ramp rate of the i-th AGC unit; and ΔT is the control cycle.

[0069] The stability constraint is as follows: for the k-th simulation step size,

[0070] ,

[0071] Where R_osc(k) is the control conflict risk index predicted by the kth simulation step; R_osc_max is the maximum allowable value of the control conflict risk index;

[0072] The system vulnerability constraint is as follows: for the k-th simulation step,

[0073] ,

[0074] Where R_vul(k) is the predicted value of the system's vulnerability index for the kth simulation step; R_vul_max is the maximum allowable value of the system's vulnerability index, i.e., the safety threshold of the system's vulnerability index.

[0075] The reserve depletion risk constraint is as follows: for the k-th simulation step,

[0076] ,

[0077] Where R_exh(k) is the predicted value of the AGC adjustment resource depletion risk index at the kth simulation step; R_exh_max is the maximum allowable value of the AGC adjustment resource depletion risk index, i.e., the safe threshold of the AGC adjustment resource depletion risk index.

[0078] This invention also provides an AGC command error prevention system based on digital twins and multi-risk look-ahead, the system comprising:

[0079] The digital twin building module is used to construct a digital twin that operates synchronously with the physical power system's AGC control loop;

[0080] The data acquisition and interaction module is used to acquire and process raw data, and to calculate and issue commands.

[0081] Furthermore, the data acquisition and interaction module includes:

[0082] The status awareness and data fusion unit is used to acquire real-time operating status data of the power grid wide-area measurement system, monitoring and data acquisition system (SCADA) and station monitoring system, and drive the digital twin to keep synchronized with the physical system;

[0083] The instruction calculation and distribution interface is used to receive raw control instructions from the upper-level AGC system and distribute the raw control instructions to the digital twin construction module.

[0084] Furthermore, the digital twin includes a refined model library, a multi-timescale simulation engine, a risk quantification assessor, and an instruction security correction module;

[0085] The refined model library stores actual physical power system model data and actual operation data, which are the basic data for multi-timescale simulation engines;

[0086] A multi-timescale simulation engine is used in a digital twin, in conjunction with the refined model library, to perform forward simulation of the execution effect of the original AGC control command in the next K control cycles, and obtain the predicted value of the system state trajectory.

[0087] The risk quantification assessor is used in a digital twin to calculate a risk index in five dimensions, in conjunction with the state trajectory prediction value calculated by the digital twin.

[0088] The instruction safety correction module is used to perform rolling optimization on the original control instructions to generate safe instructions when the risk index exceeds its preset threshold, with the goal of minimizing system risk and instruction deviation.

[0089] Furthermore, the digital twin also includes:

[0090] The risk index threshold judgment module is used to determine whether the risk index exceeds its preset threshold.

[0091] The instruction security verification module is used to ensure that there are no illegal instructions before the instructions are issued through a security verification process.

[0092] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the AGC instruction error prevention method based on digital twin and multi-risk look-ahead.

[0093] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the AGC instruction error prevention method based on digital twin and multi-risk look-ahead.

[0094] Compared with the prior art, the present invention has the following significant advantages:

[0095] 1. Foresight and initiative: By "rehearsing" instructions in virtual space, potential risks can be identified and corrected, thus achieving "prevention in advance".

[0096] 2. Global security: For the first time, the three major risks of energy storage withdrawal, new energy grid disconnection, and control conflict are quantified into optimization goals and constraints, achieving globally optimal decision-making to prevent errors.

[0097] 3. High reliability: Through refined models and closed-loop correction, the probability of system accidents caused by improper AGC commands is significantly reduced.

[0098] 4. Strong adaptability: The digital twin can update the model and parameters online, and can adapt to the dynamic changes in the power grid structure.

[0099] 5. Comprehensive Risk Control: While existing technologies have addressed some issues such as energy storage SOC management, renewable energy grid disconnection mechanisms, and system vulnerability, a complete risk indicator system is lacking to systematically guide AGC command error prevention. This invention constructs a two-way risk control system by introducing the predicted values ​​of the system's vulnerability to disturbances (R_vul) and the predicted value of the AGC regulation resource depletion risk (R_exh). This not only prevents damage to equipment from AGC commands but also proactively mitigates the impact of renewable energy anomalies on the overall power grid, significantly improving the grid's resilience and robustness.

[0100] 6. Strategic foresight in decision-making: The newly added indicators enable the system to provide early warning of major risks (such as system disconnection) and chronic risks (such as reserve depletion), supporting dispatchers to move from "tactical" control to "strategic" defense.

[0101] 7. Computational efficiency: Through the application of parallel simulation and optimized solvers, complex multi-step forward-looking risk assessment and instruction correction can be completed in seconds, meeting the real-time requirements of AGC control. Attached Figure Description

[0102] To make the content of this invention easier to understand, the invention will be further described below with reference to the accompanying drawings:

[0103] Figure 1 The data flow diagram of the AGC command error prevention method based on digital twin and multi-risk look-ahead in Example 1 and the overall architecture of the AGC command error prevention system based on digital twin and multi-risk look-ahead in Example 3 are shown below.

[0104] Figure 2 This is a flowchart of the AGC command error prevention method based on digital twin and risk prediction in Example 1;

[0105] Figure 3 This is a schematic diagram of the working logic of the risk quantification assessor in Example 2;

[0106] Figure 4 This is a comparison chart of the effects of the present invention and the traditional AGC in suppressing system oscillations in scenario simulation one of Example 4;

[0107] Figure 5 This is an internal structural diagram of the computer device in Example 5. Detailed Implementation

[0108] The following is in conjunction with the appendix Figures 1-5 The present invention will be further described with reference to specific embodiments, so that those skilled in the art can better understand and implement the present invention, but the embodiments are not intended to limit the present invention.

[0109] Example 1:

[0110] See attached document Figure 1 and Figure 2 This invention provides an AGC instruction error prevention method based on digital twins and multi-risk look-ahead, comprising the following steps:

[0111] Step S1: Construct a digital twin that operates synchronously with the physical power system's AGC control loop;

[0112] The digital twin includes a refined model library, a multi-timescale simulation engine, a risk quantification assessor, and an instruction security correction module;

[0113] The digital twin is deployed in a dedicated high-performance computing server in the dispatch control center. It interacts with the wide area measurement system (WAMS), the monitoring and data acquisition system (SCADA), and the station monitoring system through a high-speed data bus to ensure that the initial state of the simulation is consistent with the physical system.

[0114] Figure 1 PMU is an abbreviation for phasor measurement unit, which is used to measure actual power grid operation data; SIS is an abbreviation for supervisory information system, which is used to measure the operation data of power plants, new energy plants and other facilities.

[0115] Step S2: Obtain the operating status data of the wide area measurement system (WAMS), the monitoring and data acquisition system (SCADA), and the station monitoring system, and drive the digital twin to keep synchronized with the physical system;

[0116] The update cycle of the operating status data is no greater than the simulation step size of the digital twin;

[0117] Step S3: Receive the raw control command ΔP_cmd(t) from the upper-level AGC system;

[0118] Step S4: In the digital twin, using the multi-timescale simulation engine and based on the refined model library, the execution effect of the original control command in the next K control cycles is simulated in advance to obtain the system state trajectory prediction value X(t+1:t+K).

[0119] The look-ahead simulation employs parallel computing technology to accelerate computation and ensure that all K-step simulations are completed within a single AGC control cycle; the simulation step size can be set according to accuracy requirements, typically from 0.1 seconds to 1 second.

[0120] Step S5: Using the risk quantification evaluator, calculate the risk index in five dimensions based on the system state trajectory prediction value obtained in step S4.

[0121] The five dimensions of risk index include:

[0122] The energy storage exit risk index R_soc is used to quantify the risk of exiting operation due to the state of charge of an energy storage unit exceeding the limit. For each energy storage unit in the system, its SOC prediction trajectory is calculated independently and the risk is assessed. Finally, the maximum value among all energy storage units is taken as the energy storage exit risk index of the system.

[0123] The renewable energy grid disconnection risk index R_trip is used to quantify the risk of renewable energy units disconnecting from the grid due to frequency or voltage exceeding limits.

[0124] The control conflict risk index Rosc is used to quantify the risk of system oscillation caused by mismatch between fast and slow resource responses; the specific frequency band of system oscillation [f_min, f_max] is obtained through offline modal analysis or online identification and is tailored to the oscillation mode between regions of this power grid.

[0125] The system's vulnerability index R_vul is used to quantify the system's frequency stability margin under the worst-case scenario of a single grid disconnection event from a new energy source.

[0126] The AGC adjustment resource depletion risk index R_exh is used to quantify the risk of AGC increasing reserve capacity being depleted when new energy sources experience sustained abnormal fluctuations.

[0127] Step S6: Determine whether any of the five risk indices exceeds its preset threshold; if so, trigger the instruction safety correction module and execute step S7; if not, directly send the original control instruction ΔP_cmd(t) to the site physical execution unit.

[0128] The preset threshold can be dynamically adjusted according to the system operation mode and weather conditions (which affect the accuracy of new energy prediction). For example, during typhoon weather, the risk threshold R_trip_threshold for new energy grid disconnection can be appropriately reduced to adopt a more conservative control strategy.

[0129] Step S7: The instruction safety correction module is activated to perform rolling optimization on the original control instruction with the goal of minimizing system risk and instruction deviation, and to generate a safety instruction ΔP_safe(t).

[0130] The rolling optimization is solved using the interior point method or the effective set method, and the result of the previous optimization is used as a warm start to improve computational efficiency.

[0131] The process of solving the rolling optimization problem is as follows:

[0132] The objective function for establishing the mathematical model is,

[0133]

[0134] Where k = 1 to K; Weighted L2 norm for safety command error; W1 is the weighted L2 norm of the overall system risk; W2 is the instruction deviation weight matrix; W3 is the system risk weight matrix.

[0135] The constraints of the mathematical model include power balance constraints, energy storage security constraints, ramp rate constraints, stability constraints, system vulnerability constraints, and reserve depletion risk constraints.

[0136] Step S8: The safety instruction ΔP_safe(t) is sent to the physical execution unit. Before the instruction is sent, a safety verification process must be carried out to ensure that there are no illegal instruction values. After the instruction is sent, the system records all relevant data of this correction event, including the original instruction, safety instruction, risk index and optimization process, for post-event analysis and model optimization.

[0137] Example 2:

[0138] See attached document Figure 3 The energy storage exit risk index R_soc is calculated in the following way:

[0139] First, the risk potential function is defined as follows:

[0140] ;

[0141] Where A, B, C, and D are fitting parameters greater than 0, and SOC_high and SOC_low are the upper and lower warning limits of energy storage SOC.

[0142] Therefore, the maximum value of the energy storage exit risk index within the prediction time domain is as follows:

[0143] ;

[0144] Where k = 1 to K; SOC_simulated(k) is the predicted value of the k-th step of energy storage; K is the total number of control cycles.

[0145] The new energy grid disconnection risk index R_trip is calculated in the following way:

[0146] ,

[0147] Where k = 1 to K; Δf_simulated(k) is the predicted value of the system frequency deviation at the k-th step; Δf_max is the maximum allowable frequency deviation threshold for the new energy unit; Δv_simulated(k) is the predicted value of the system voltage deviation at the k-th step; Δv_max is the maximum allowable voltage deviation threshold for the new energy unit; and K is the total number of control cycles.

[0148] The conflict control risk index Rosc is calculated using the following formula:

[0149] Perform a Fast Fourier Transform (FFT) on the predicted trajectory P_tie(t+1:t+K) of the regional tie line power, and calculate the spectral energy within a specific frequency band [f_min, f_max].

[0150] ;

[0151] Where f is in the range [f_min, f_max].

[0152] The system's vulnerability index R_vul is calculated using the following formula:

[0153] ;

[0154] Where f_nom is the system's rated frequency; f_dip_simulated is the lowest frequency point of the system after the simulated worst-case scenario of the new energy power station being instantly disconnected from the grid; and f_under is the frequency of the first round of low-frequency load shearing.

[0155] The AGC-regulated resource depletion risk index R_exh is calculated using the following formula:

[0156] ;

[0157] Where T_exh = P_ava / (ΔP_def / ΔT), P_ava is the total available up-up reserve capacity of the system, ΔP_def is the assumed continuous power deficit of new energy sources, ΔT is the control period, and a is the time scale coefficient, for example 1 / 60, which means in minutes.

[0158] Example 3:

[0159] See attached document Figure 1 This embodiment provides an AGC command error prevention system based on digital twin and multi-risk look-ahead, the system including:

[0160] The digital twin building module is used to construct a digital twin that operates synchronously with the physical power system's AGC control loop;

[0161] The data acquisition and interaction module is used to acquire and process raw data, and to calculate and issue commands.

[0162] The data acquisition and interaction module includes:

[0163] The status awareness and data fusion unit is used to acquire real-time operating status data of the power grid wide-area measurement system, monitoring and data acquisition system (SCADA) and station monitoring system, and drive the digital twin to keep synchronized with the physical system;

[0164] The instruction calculation and distribution interface is used to receive raw control instructions from the upper-level AGC system and distribute the raw control instructions to the digital twin construction module.

[0165] The digital twin includes a refined model library, a multi-timescale simulation engine, a risk quantification assessor, and an instruction security correction module;

[0166] The refined model library stores actual physical power system model data and actual operation data, which are the basic data for multi-timescale simulation engines;

[0167] A multi-timescale simulation engine is used in a digital twin, in conjunction with the refined model library, to perform forward simulation of the execution effect of the original AGC control command in the next K control cycles, and obtain the predicted value of the system state trajectory.

[0168] The risk quantification assessor is used in a digital twin to calculate a risk index in five dimensions, in conjunction with the state trajectory prediction value calculated by the digital twin.

[0169] The instruction safety correction module is used to perform rolling optimization on the original control instructions to generate safe instructions when the risk index exceeds its preset threshold, with the goal of minimizing system risk and instruction deviation.

[0170] The digital twin also includes:

[0171] The risk index threshold judgment module is used to determine whether the risk index exceeds its preset threshold.

[0172] The instruction security verification module is used to ensure that there are no illegal instructions before the instructions are issued through a security verification process.

[0173] The risk index threshold judgment module uses the risk quantification assessor to calculate a risk index in five dimensions based on the predicted value of the system state trajectory.

[0174] The five dimensions of risk index include:

[0175] The energy storage exit risk index R_soc is used to quantify the risk of exiting operation due to the state of charge of an energy storage unit exceeding the limit. For each energy storage unit in the system, its SOC prediction trajectory is calculated independently and the risk is assessed. Finally, the maximum value among all energy storage units is taken as the energy storage exit risk index of the system.

[0176] The renewable energy grid disconnection risk index R_trip is used to quantify the risk of renewable energy units disconnecting from the grid due to frequency or voltage exceeding limits.

[0177] The control conflict risk index Rosc is used to quantify the risk of system oscillation caused by mismatch between fast and slow resource responses; the specific frequency band of system oscillation [f_min, f_max] is obtained through offline modal analysis or online identification and is tailored to the oscillation mode between regions of this power grid.

[0178] The system's vulnerability index R_vul is used to quantify the system's frequency stability margin under the worst-case scenario of a single grid disconnection event from a new energy source.

[0179] The AGC adjustment resource depletion risk index R_exh is used to quantify the risk of AGC increasing reserve capacity being depleted when new energy sources experience sustained abnormal fluctuations.

[0180] If any of the five risk indices exceeds its preset threshold, the instruction safety correction module is triggered; if none exceed the preset threshold, the original control instruction ΔP_cmd(t) is directly sent to the site physical execution unit.

[0181] Example 4:

[0182] Taking a provincial power grid as an example, the penetration rate of wind and solar power in this grid has reached 35%, and it has been equipped with hundreds of megawatt-hours of electrochemical energy storage. The anti-misoperation system of this invention is deployed at the AGC master station of this power grid.

[0183] 1. System Construction:

[0184] Refined model library: The energy storage model adopts a second-order RC equivalent circuit model; the new energy power station model includes detailed inverter control logic and low voltage ride-through criteria; all model parameters are obtained through joint identification of equipment manufacturer data and on-site testing, and are regularly verified and updated using historical data;

[0185] Hardware platform: The system is deployed in the dispatch cloud center and uses high-performance servers for parallel computing; to ensure reliability, a dual-machine redundancy configuration is adopted. Network communication adopts a dual-plane architecture of the power dispatch data network to ensure the real-time performance and reliability of data transmission.

[0186] 2. Operation process:

[0187] Receives a regional adjustment request command ΔP_cmd from the superior AGC every 4 seconds.

[0188] The digital twin is activated, and in 1-second increments, it performs rapid simulations of the system dynamics over the next 80 seconds (K=20).

[0189] The risk quantification assessor calculates five major risk indices. The energy storage exit risk threshold R_soc_threshold is set to 0.8.

[0190] Scenario Simulation 1 (Energy Storage Risk): At a certain moment, the simulation engine predicts that if the original command is executed, a critical energy storage power station will reach a SOC of 95% after 60 seconds, with a calculated R_soc value of 0.92, exceeding the threshold. The command safety correction module is triggered. The optimization model, while meeting the overall regulation requirements, reduces the charging power command of the energy storage unit by 20% and redistributes this power difference to a hydropower unit with regulation margin. The optimized safety command ΔP_safe is issued.

[0191] Scenario Simulation 2 (System Vulnerability Risk): One afternoon, the system load is relatively light, with photovoltaic output accounting for 50% of the total load. The digital twin calculates R_vul, simulating the instantaneous grid disconnection of the largest photovoltaic power plant (300MW). The simulation shows that the system frequency will drop to 49.1Hz, lower than the first round of low-frequency load shedding setpoint (49.2Hz), and R_vul is calculated to be 1.1, exceeding the threshold of 1.0. The instruction safety correction module is triggered. The optimization result is: based on the original AGC instruction, an additional preventative instruction is generated, ordering a 300MW gas turbine unit to maintain 150MW of spinning reserve instead of shutting down according to economic dispatch. Although this increases operating costs slightly, it reduces R_vul to 0.7, avoiding a potential grid accident.

[0192] Scenario Simulation 3 (Risk of Reserve Exhaustion): At night, wind power forecast output remains consistently high, but R_exh calculations show that if actual wind power output is 1000MW lower than the forecast for 10 minutes, the system's AGC-adjusted reserve will be exhausted within 5 minutes. The R_exh index exceeds the threshold. The instruction safety correction module adjusts the instructions, prioritizing the use of slow-responding but large-capacity thermal power units for deep adjustments, and reserving emergency reserves for hydropower units, rather than relying entirely on fast but limited energy storage. Simultaneously, a "Reserve Capacity Critical" warning is issued to the dispatcher.

[0193] Example 5:

[0194] This embodiment provides a computer device, which may be a server, and its internal structure diagram may be as follows. Figure 5As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The database stores data required for this invention. The network interface is used for communication with external terminals via a network connection.

[0195] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0196] Example 6:

[0197] This embodiment provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of an AGC instruction error prevention method based on digital twins and multi-risk look-ahead.

[0198] Example 7:

[0199] This embodiment provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of an AGC instruction error prevention method based on digital twins and multi-risk look-ahead.

[0200] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0201] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for preventing errors in AGC instructions based on digital twins and multi-risk look-ahead, characterized in that, Includes the following steps: Step S1: Construct a digital twin that operates synchronously with the physical power system's AGC control loop; The digital twin includes a refined model library, a multi-timescale simulation engine, a risk quantification assessor, and an instruction security correction module; Step S2: Obtain the operating status data of the power grid wide-area measurement system, monitoring and data acquisition system and the station monitoring system, and drive the digital twin to keep synchronized with the physical system; The update cycle of the operating status data is no greater than the simulation step size of the digital twin; Step S3: Receive raw control commands from the upper-level AGC system; Step S4: In the digital twin, using the multi-timescale simulation engine and based on the refined model library, the execution effect of the original control command in the next K control cycles is simulated in advance to obtain the predicted value of the system state trajectory. Step S5: Using the risk quantification evaluator, calculate the risk index in five dimensions based on the system state trajectory prediction value obtained in step S4. Step S6: Determine whether any of the five risk indices exceeds its preset threshold. If yes, the instruction safety correction module is triggered, and step S7 is executed; if no, the original control instruction is directly sent to the site physical execution unit. Step S7: The instruction safety correction module is activated to perform rolling optimization on the original control instructions with the goal of minimizing system risks and instruction deviations, and to generate safe instructions. Step S8: Send the security command to the physical execution unit; Before issuing the instruction, a security verification process is conducted to ensure that there are no illegal instruction values. After issuance, the system records all relevant data of this correction event, including the original instruction, security instruction, risk index and optimization process, for post-event analysis and model optimization.

2. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 1, characterized in that, The refined model library is used to simulate actual physical systems and consists of power grid structure models, traditional generating unit models, and new energy generating unit models. The multi-timescale simulation engine calculates the basic data; the multi-timescale simulation engine is based on a refined model library to realize the simulation and prediction of power grid state trajectory; The risk quantification assessor calculates various risk indices during AGC control based on power grid dynamic simulation data. The instruction safety correction module corrects the initial AGC control instructions based on the risk index.

3. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 2, characterized in that, The operating status data mentioned in step S2 includes at least: the output and switching status of each generator set, the power flow of the main transmission lines, the voltage amplitude and phase angle of all bus nodes, the actual and predicted power of the new energy power station, and the real-time power and state of charge of the energy storage power station.

4. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 2, characterized in that, The look-ahead simulation described in step S4 uses parallel computing technology to accelerate the calculation and ensure that all K-step simulations are completed within a single AGC control cycle; the simulation step size can be set according to the accuracy requirements, ranging from 0.1 seconds to 1 second.

5. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 2, characterized in that, The five dimensions of risk indices mentioned in step S5 include: The energy storage exit risk index R_soc is used to quantify the risk of exiting operation due to the state of charge of an energy storage unit exceeding the limit. For each energy storage unit in the system, its SOC prediction trajectory is calculated independently and the risk is assessed. Finally, the maximum value among all energy storage units is taken as the energy storage exit risk index of the system. The renewable energy grid disconnection risk index R_trip is used to quantify the risk of renewable energy units disconnecting from the grid due to frequency or voltage exceeding limits. The control conflict risk index Rosc is used to quantify the risk of system oscillation caused by mismatch between fast and slow resource responses; the specific frequency band of system oscillation [f_min, f_max] is obtained through offline modal analysis or online identification and is tailored to the oscillation mode between regions of this power grid. The system's vulnerability index R_vul is used to quantify the system's frequency stability margin under the worst-case scenario of a single grid disconnection event from a new energy source. The AGC adjustment resource depletion risk index R_exh is used to quantify the risk of AGC increasing reserve capacity being depleted when new energy sources experience sustained abnormal fluctuations.

6. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 5, characterized in that, The energy storage exit risk index R_soc is calculated in the following way: First, the risk potential function is defined as follows: ； Where A, B, C, and D are fitting parameters greater than 0, and SOC_high and SOC_low are the upper and lower warning limits of energy storage SOC. Therefore, the maximum value of the energy storage exit risk index within the prediction time domain is as follows: ； Where k = 1 to K; SOC_simulated(k) is the predicted value of the k-th step of energy storage; K is the total number of control cycles.

7. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 5, characterized in that, The new energy grid disconnection risk index R_trip is calculated in the following way: ， Where k = 1 to K; Δf_simulated(k) is the predicted value of the system frequency deviation at the k-th step; Δf_max is the maximum allowable frequency deviation threshold for the new energy unit; Δv_simulated(k) is the predicted value of the system voltage deviation at the k-th step; Δv_max is the maximum allowable voltage deviation threshold for the new energy unit; and K is the total number of control cycles.

8. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 5, characterized in that, The conflict control risk index Rosc is calculated using the following formula: Perform a Fast Fourier Transform (FFT) on the predicted trajectory P_tie(t+1:t+K) of the regional tie line power, and calculate the spectral energy within a specific frequency band [f_min, f_max]. ； Where f is in the range [f_min, f_max].

9. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 5, characterized in that, The system's vulnerability index R_vul is calculated using the following formula: ； Where f_nom is the system's rated frequency; f_dip_simulated is the lowest frequency point of the system after the simulated worst-case scenario of the new energy power station being instantly disconnected from the grid; and f_under is the frequency of the first round of low-frequency load shearing.

10. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 5, characterized in that, The AGC-regulated resource depletion risk index R_exh is calculated using the following formula: ； Where T_exh = P_ava / (ΔP_def / ΔT), P_ava is the total available up-up reserve capacity of the system, ΔP_def is the assumed continuous power deficit of new energy sources, ΔT is the control period, and a is the time scale coefficient, for example 1 / 60, which means in minutes.

11. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 2, characterized in that, The preset threshold mentioned in step S6 is dynamically adjusted according to the system operation mode and weather conditions.

12. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 2, characterized in that, The rolling optimization described in step S7 is solved using the interior point method or the effective set method, and the result of the previous optimization is used as a warm start to improve computational efficiency.

13. The AGC instruction error prevention method based on digital twin and multi-risk look-ahead as described in claim 12, characterized in that, In step S7, the rolling optimization solution process is as follows: The objective function for establishing the mathematical model is, Where k = 1 to K; Weighted L2 norm for safety command error; W1 is the weighted L2 norm of the overall system risk; W2 is the instruction deviation weight matrix; W3 is the system risk weight matrix. The constraints of the mathematical model include power balance constraints, energy storage security constraints, ramp rate constraints, stability constraints, system vulnerability constraints, and reserve depletion risk constraints.

14. An AGC command error prevention system based on digital twin and multi-risk look-ahead, characterized in that, The system includes: The digital twin building module is used to construct a digital twin that operates synchronously with the physical power system's AGC control loop; The data acquisition and interaction module is used to acquire and process raw data, and to calculate and issue commands.

15. The AGC command error prevention system based on digital twin and multi-risk look-ahead as described in claim 14, characterized in that, The data acquisition and interaction module includes: The status awareness and data fusion unit is used to acquire real-time operating status data of the power grid wide-area measurement system, the monitoring and data acquisition system SCADA and the station monitoring system, and drive the digital twin to keep synchronized with the physical system; The instruction calculation and distribution interface is used to receive raw control instructions from the upper-level AGC system and distribute the raw control instructions to the digital twin construction module.

16. The AGC command error prevention system based on digital twin and multi-risk look-ahead as described in claim 14, characterized in that, The digital twin includes a refined model library, a multi-timescale simulation engine, a risk quantification assessor, and an instruction security correction module; The refined model library stores actual physical power system model data and actual operation data, which are the basic data for multi-timescale simulation engines; A multi-timescale simulation engine is used in a digital twin, in conjunction with the refined model library, to perform forward simulation of the execution effect of the original AGC control command in the next K control cycles, and obtain the predicted value of the system state trajectory. The risk quantification assessor is used in a digital twin to calculate a risk index in five dimensions, in conjunction with the state trajectory prediction value calculated by the digital twin. The instruction safety correction module is used to perform rolling optimization on the original control instructions to generate safe instructions when the risk index exceeds its preset threshold, with the goal of minimizing system risk and instruction deviation.

17. The AGC command error prevention system based on digital twin and multi-risk look-ahead as described in claim 16, characterized in that, The digital twin also includes: The risk index threshold judgment module is used to determine whether the risk index exceeds its preset threshold. The instruction security verification module is used to ensure that there are no illegal instructions before the instructions are issued through a security verification process.

Images