A pumped storage power station protection setting value processing method, device, equipment and medium

Abstract

The application provides a pumped storage power station protection fixed value processing method, device, equipment and medium, relates to the relay protection technology field of power systems, and comprises the following steps: determining fixed value combination data from a fixed value feasible region, inputting the fixed value combination data into a pumped storage power station simulation model, simulating multiple fault scenes, and obtaining performance data; performing feature extraction on the performance data to obtain performance indexes; taking the fixed value combination data and the performance indexes as a training set, training an initial performance proxy model, and obtaining a performance proxy model; standardizing the fixed value feasible region, determining a protection fixed value population, inputting the protection fixed value population into the performance proxy model, obtaining individual performance indexes corresponding to each individual, performing multi-objective constraint optimization processing on the protection fixed value population based on the individual performance indexes, and obtaining fixed value combination data actually used by a protection device. The technical scheme of the embodiment of the application can improve the reliability and intelligent level of relay protection of pumped storage power stations.

Description

Technical Field

[0001] This invention relates to the field of power system relay protection technology, and in particular to a method, apparatus, equipment and medium for processing protection settings of pumped storage power stations. Background Technology

[0002] Due to the complex operating conditions (frequent switching between power generation, pumping, and phase regulation) and special equipment (including static frequency converters (SFC) and bidirectional generator motors), the reliability of the relay protection system of pumped storage power stations is of paramount importance.

[0003] Traditional protection setting methods primarily rely on short-circuit current calculations during the design phase and the accumulated experience of maintenance personnel, determining protection thresholds through offline simulation and manual verification. However, with dynamic changes in the power grid structure and the evolution of operating conditions such as equipment aging, this static setting method has gradually revealed its lag problem, failing to adapt to changes in system operating parameters in real time. Furthermore, manual verification is limited by time and cost, typically only verifying typical fault scenarios (such as three-phase short circuits and phase-to-phase short circuits), making it difficult to comprehensively cover edge conditions such as composite faults and high-resistance grounding, resulting in blind spots in the selectivity and sensitivity of the protection system. Current industry standards typically set the review cycle for protection settings at 3-5 years. This long-term static adjustment mechanism is not only inefficient but also lacks quantitative assessment of dynamic factors such as equipment degradation and changes in power grid topology. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, equipment and medium for processing protection settings of pumped storage power stations, which realizes the transformation of protection settings from empirical adjustment to data-driven adaptive optimization by constructing a "simulation-evaluation-optimization" closed loop.

[0005] In a first aspect, embodiments of the present invention provide a method for processing protection settings in a pumped storage power station, comprising: From the feasible range of preset setpoints for the protection device in the pumped storage power station, determine the setpoint combination data when the protection device performs the protection action; The set value combination data is input into the pumped storage power station simulation model to simulate multiple fault scenarios and obtain the performance data of the protection device when it performs protection actions under each fault scenario. Feature extraction is performed on each of the performance data to obtain the performance index corresponding to the fixed value combination data; The fixed-value combination data and the performance indicators are used as the training set to train the initial performance proxy model, thereby obtaining the performance proxy model. The feasible domain of the set value is standardized, and a preset number of individuals are randomly selected from it as the protection set value population; wherein, the individuals are used to describe a set of set value combination data corresponding to the protection device; The protection setpoint population is input into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population; Based on the individual performance indicators, the protection setpoint population is subjected to multi-objective constraint optimization to obtain the optimal individual corresponding to the protection setpoint population, and the optimal individual is used as the setpoint combination data actually used by the protection device.

[0006] In a preferred embodiment of the present invention, the aforementioned input of the setpoint combination data into a pumped storage power station simulation model is used to simulate multiple fault scenarios, thereby obtaining performance data of the protection device when performing protection actions under each fault scenario, including: Acquire data from multiple fault scenarios; For each fault scenario data, the set value combination data and the fault scenario data are input into the pumped storage power station simulation model to obtain the performance data corresponding to the protection device performing protection actions under the fault scenario.

[0007] In a preferred embodiment of the present invention, the above-described acquisition of multiple fault scenario data includes: Acquire multiple fault dimension information pre-set for fault scenarios; wherein, the fault dimension information includes the following four dimensions: operating condition dimension, fault location dimension, fault type dimension, and power plant system boundary dimension; Based on the fault dimension information, the fault dimension information is sampled to obtain fault scenario data.

[0008] In a preferred embodiment of the present invention, before training the initial performance proxy model using the fixed-value combination data and the performance indicators as a training set to obtain the performance proxy model, the method further includes: The fixed-value combination data and the performance index are standardized to obtain a standardized training set; The standardized training set is input into the initial performance proxy model to train the initial performance proxy model and obtain the performance proxy model.

[0009] In a preferred embodiment of the present invention, after inputting the standardized training set into the initial performance proxy model and training the initial performance proxy model to obtain the performance proxy model, the method further includes: Multiple verification value combination data are obtained in the feasible domain of the fixed value; wherein the verification value combination data is different from the fixed value combination data; Each of the aforementioned verification setpoint combinations is input into the performance proxy model, and the performance proxy model is verified to obtain the verification performance index and prediction uncertainty corresponding to each of the aforementioned verification setpoint combinations. Based on the predicted uncertainty, supplementary fixed-value combination data are selected from the verified fixed-value combination data; The supplementary fixed-value combination data and the corresponding verification performance indicators are added to the training set, and the training set is updated. Return to the step of standardizing the fixed value combination and the performance index to obtain a standardized training set, until the first iteration condition is met.

[0010] In a preferred embodiment of the present invention, the above-mentioned multi-objective constraint optimization processing of the protection setpoint population based on each individual performance index to obtain the optimal individual corresponding to the protection setpoint population, and the optimal individual as the setpoint combination data actually used by the protection device, includes: Based on the individual performance indicators, determine the target function value for optimization; Based on the objective function values, individuals in the protected fixed-value population are selected, crossovered, and mutated to generate offspring fixed-value populations; The protected population with the offspring population with the value of the parent population is merged to obtain a mixed population with the value of the parent population. Based on the non-dominated ordering and crowding distance of each individual in the mixed fixed-value population, the individuals in the mixed fixed-value population are screened to obtain an updated protected fixed-value population; Return to the step of inputting the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population, until the second iteration condition is met; According to the preset engineering screening conditions, the optimal individual is selected from the protection setpoint population, and the optimal individual is used as the setpoint combination data for actual use of the protection device.

[0011] In a preferred embodiment of the present invention, based on the individual performance indicators, the protection setpoint population is subjected to multi-objective constraint optimization processing to obtain the optimal individual corresponding to the protection setpoint population. After using the optimal individual as the setpoint combination data actually used by the protection device, the method further includes: Obtain the historical setpoint combination data and corresponding historical performance indicators of the protection device; The fixed-value combination data corresponding to the optimal individual is input into the performance proxy model to obtain the optimized performance index; The historical setpoint combination data and the historical performance indicators are compared with the setpoint combination data and the optimized performance indicators corresponding to the optimal individual, an optimization report is generated, and the report is fed back to the power plant operation and maintenance management system.

[0012] Secondly, embodiments of the present invention also provide a protection setting processing device for a pumped storage power station, comprising: The setting combination determination module is used to determine the setting combination data when the protection device performs protection actions from the setting feasible domain that is preset for the protection device in the pumped storage power station. The simulation module is used to input the set value combination data into the pumped storage power station simulation model, simulate multiple fault scenarios, and obtain the performance data of the protection device when performing protection actions under each fault scenario. The feature extraction module is used to extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data. The training module is used to train the initial performance proxy model using the fixed-value combination data and the performance indicators as a training set, so as to obtain the performance proxy model. The standardization module is used to standardize the feasible domain of the set value and randomly select a preset number of individuals from it as the protection set value population; wherein, the individuals are used to describe a set of set value combination data corresponding to the protection device; The performance index determination module is used to input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population; The optimization module is used to perform multi-objective constraint optimization on the protection setpoint population based on the individual performance indicators to obtain the optimal individual corresponding to the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

[0013] Thirdly, embodiments of the present invention also provide an electronic device, including a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the pumped storage power station protection setting processing method of the first aspect described above.

[0014] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are invoked and executed by a processor, the computer-executable instructions cause the processor to implement the pumped storage power station protection setting processing method described in the first aspect.

[0015] The embodiments of the present invention bring the following beneficial effects: This invention provides a method for processing protection settings in pumped storage power stations. It utilizes a pumped storage power station simulation model to simulate multiple fault scenarios, enabling the simulation of numerous rare but dangerous operating conditions. This overcomes the limitation of traditional methods, which can only verify a limited number of common scenarios. The data obtained through high-precision simulation is closer to the actual field situation than purely theoretical calculations, ensuring the accuracy of subsequent analysis and optimization. Using a trained performance surrogate model for subsequent protection setting optimization replaces time-consuming simulation calculations with rapid model predictions, allowing the complex optimization process to be completed within an acceptable timeframe, thus improving optimization efficiency. This invention establishes a closed loop of "simulation-training-optimization," eliminating reliance on manual experience or simple calculations, and achieving a shift from "manual experience-based tuning" to "data-driven adaptive optimization."

[0016] Other features and advantages of this disclosure will be set forth in the following description, or some features and advantages may be inferred from the description or determined without doubt, or may be learned by practicing the techniques described above.

[0017] To make the above-mentioned objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 A flowchart of a protection setting processing method for a pumped storage power station provided in an embodiment of the present invention; Figure 2a A flowchart of another method for processing protection settings in a pumped storage power station provided in an embodiment of the present invention; Figure 2b This is a schematic diagram of the Latin hypercube sampling method provided in an embodiment of the present invention; Figure 3 A flowchart of another method for processing protection settings in a pumped storage power station provided in an embodiment of the present invention; Figure 4 A flowchart of another method for processing protection settings in a pumped storage power station provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a protection setting processing device for a pumped storage power station provided in an embodiment of the present invention; Figure 6This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Pumped storage power stations, due to their complex operating conditions (frequent switching between power generation, pumping, and phase regulation) and specialized equipment (including static frequency converters (SFC) and bidirectional generator motors), require highly reliable relay protection systems. Current protection setting primarily relies on short-circuit calculations from the initial design phase and the experience of maintenance personnel, which presents three major drawbacks: 1) Lag: Unable to dynamically respond to changes in power grid structure, aging of generating units, and other evolutions in operating status; 2) One-sidedness: Manual verification only covers a limited number of fault types, making it difficult to verify the selectivity and sensitivity under edge conditions such as compound faults and high-resistance grounding; 3) Inefficiency: The verification cycle for fixed values ​​is long (usually 3-5 years) and lacks quantitative basis.

[0022] Based on this, the present invention provides a method for processing protection settings in pumped storage power stations. By establishing a closed loop of "simulation-training-optimization", it no longer relies on manual experience or simple calculations. Instead, it generates data through high-fidelity simulation, trains a fast and accurate "performance proxy model", and then uses this model to efficiently search for the optimal protection settings in a broad "feasible domain". This realizes the transformation of protection settings from experience-based data-driven adaptive optimization, and improves the accuracy and efficiency of protection setting optimization.

[0023] To facilitate understanding of this embodiment, a detailed description of the protection setting processing method for a pumped storage power station disclosed in this embodiment of the invention will be provided first.

[0024] Example 1 This invention provides a method for processing protection settings in a pumped storage power station. Figure 1 This is a flowchart illustrating a protection setting processing method for a pumped storage power station, provided as an embodiment of the present invention. Figure 1 As shown, the protection setting processing method for this pumped storage power station may include the following steps: Step S101: Determine the set value combination data when the protection device performs protection action from the set value feasible domain preset for the protection device in the pumped storage power station.

[0025] Protection settings are a set of parameters used in a power system to determine whether a protection device should operate. These include time settings and current settings. Time settings refer to the time required for the protection to take action within a specified time, while current settings refer to the action taken by the protection device when the electrical quantity exceeds the set value. These settings directly relate to whether the protection device can operate correctly, quickly, and selectively when a fault occurs. The scientific calculation and reasonable setting of these settings are crucial to ensuring that relay protection devices can reliably, accurately, and promptly perform tripping or alarm functions when a fault or anomaly occurs in the power system, thereby isolating the fault and preventing the accident from escalating. The setting of protection settings is not arbitrary but requires comprehensive consideration of the various characteristics of the protected object and the actual situation of the power grid. For example, the setting of current settings needs to take into account factors such as the characteristics of the power system, the grid structure, and load conditions. The setting of time settings needs to ensure that the protection device can operate quickly and accurately when a fault occurs, in order to reduce the impact of the fault on the power system. The feasible range of settings is a pre-set range of legal values ​​for the protection settings of relay protection devices. Fixed-value combination data refers to specific values ​​selected within the fixed-value feasible region.

[0026] Specifically, a certain number of fixed-value combinations can be randomly selected from the feasible region of the fixed values. For example, within the feasible region of the fixed values ​​[Z_set∈(0.6Ω, 1.2Ω), t_delay∈(0s, 0.1s)], 80 sets of fixed-value combinations can be randomly selected, where Z_set is the impedance fixed value and t_delay is the time delay.

[0027] It is understandable that there are multiple relay protection devices in a pumped storage power station. Each relay protection device has a different protection area. The relay protection devices can be classified according to the size of their protection area. For example, the levels can be divided into three stages from bottom to top: stage 1, stage 2, and stage 3, which represent the protection area from small to large.

[0028] Step S102: Input the set value combination data into the pumped storage power station simulation model, simulate multiple fault scenarios, and obtain the performance data corresponding to the protection device when performing protection actions under each fault scenario.

[0029] The pumped storage power station simulation model is used to simulate the operating status of protection devices in a pumped storage power station, simulating the protective behaviors of these devices under different fault scenarios. In this embodiment of the invention, the pumped storage power station simulation model is a semi-physical simulation platform. That is, the simulation model includes a primary equipment simulation module, which simulates equipment such as generators, main transformers, lines, disconnectors, loads, and SFCs (Static Frequency Converters) in the pumped storage power station, providing an operating environment for the protection devices. It also includes a physical control panel module, which includes protection devices (relay protection cabinets) that can execute corresponding protection actions based on the operating environment provided by the primary equipment simulation module. In the physical control panel module, all equipment is of the same model as the equipment in the pumped storage power station. The operating environment of the protection device describes whether the protection device is operating normally or experiencing certain faults. The fault scenarios describe the fault conditions that occur in the protection devices during operation of the pumped storage power station.

[0030] Performance data is used to describe the protective action behavior of the protection device under fault scenarios, and may include raw data such as whether it takes action and the action delay.

[0031] Specifically, for each setpoint combination, after inputting it into the pumped storage power station simulation model, the model will perform a simulation for each fault scenario to obtain the performance data of the protection device under each fault scenario. Therefore, each setpoint combination corresponds to the performance data under multiple fault scenarios. Alternatively, multiple fault scenarios can be simulated simultaneously to obtain the corresponding performance data under each fault scenario.

[0032] For example, an RCS-943 protection device of the same model was connected to a hardware-in-the-loop (HIL) simulation platform. One by one, 1200 fault scenarios were simulated, and the protection's activation status, activation time (accuracy ±0.1ms), and fault current waveform were collected for each scenario. The power supply of the protection device was automatically controlled by a PLC, and a reset was completed within 5 seconds after each simulation, achieving unattended continuous testing. Simultaneously, only four types of characteristic data were recorded as performance data: activation time, false trip flag (0 / 1), effective starting current value, and impedance trajectory endpoint coordinates. The specific meanings of the characteristic data are as follows: Activation flag (0 / 1): Whether the protection activated; Activation time (ms): The time from the occurrence of the fault to the activation of the output relay, accuracy ±0.1ms; Effective starting current value (A): The RMS value of the three-phase current when the protection starts; Impedance trajectory endpoint coordinates (R, X): The impedance at the fault point measured from the protection element. The original waveform recording file was only saved when the activation deviation was >5ms, reducing storage and analysis workload.

[0033] Step S103: Extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data.

[0034] Performance indicators are used to describe the protection device's ability to handle fault conditions. In embodiments of this invention, performance indicators may include selectivity compliance rate, sensitivity margin, speed deviation, and coordination margin. The selectivity compliance rate describes the probability of the protection device operating correctly, where correct operation means that it should operate for faults within the protection zone but not for faults outside the zone. The sensitivity margin describes the ability to identify minor faults such as high-resistance grounding. The speed deviation describes whether the operating speed meets the requirements. The coordination margin describes whether the time or current coordination between upstream and downstream protection systems is reasonable.

[0035] Specifically, for each setpoint combination, the performance data collected under various fault scenarios is statistically analyzed to obtain the performance indicators of the protection device. For example, when optimizing the current setpoint, the formula for calculating the performance indicators is as follows: Selective pass rate = (Number of correct actions within the zone + Number of correct but non-correct actions outside the zone) / Number of failure scenarios Sensitivity margin = minimum fault current / impedance setpoint Speed ​​deviation = Action time - Theoretical action time Matching margin = Lower-level protection action time - Upper-level protection action time The number of correct actions within the protection zone refers to the number of protection actions performed when the fault scenario is within the protection zone. The number of correct non-actions outside the protection zone refers to the number of protection actions not performed when the fault scenario is outside the protection zone. The minimum fault current is the effective value of the starting current.

[0036] It is understandable that pumped-storage power stations contain multiple protection devices, each with a different protection area. These devices can be classified according to the size of their protection areas; for example, from bottom to top, they can be divided into three levels: Level 1, Level 2, and Level 3, representing protection areas from smallest to largest. The lower-level protection's operating time refers to the operating time of the next lower-level protection device, while the upper-level protection's operating time refers to the operating time of the previous-level protection device. When a protection device is located in Level 1 or Level 3, the coordination margin can be disregarded as a performance indicator.

[0037] Based on the previous example, the protection level is segment 1. Four-dimensional index calculation: a) Selective pass rate: The total number of correct actions (1182 times) and correct non-actions (1182 times) for faults within the zone were counted out of 1200 scenarios. Pass rate = (1182 + 1182) / 1200 = 98.5%. 18 instances of false actions outside the zone were found, all occurring in the "isolated network + high impedance" scenario.

[0038] b) Sensitivity margin: Calculate the minimum fault current / starting current setting for all high-resistance grounding faults within the zone. In the scenario of "pump start + islanded grid + high resistance (300Ω)", the measured minimum fault current is 1.2 times the starting setting, so the sensitivity margin = 1.2, which is lower than the required 1.5.

[0039] c) Speed ​​deviation: The theoretical action time should be 0ms (no delay), and the actual maximum deviation is 8ms, which meets the requirement of ≤10ms.

[0040] d) Coordination margin: In this example, it is segment I, which does not require coordination with the lower level, so this indicator is not applicable.

[0041] Output: A performance evaluation report that clearly points out that the current setting has insufficient sensitivity and selectivity risks in the "island network + high impedance" scenario.

[0042] Step S104: Use the fixed value combination data and the performance index as a training set to train the initial performance proxy model to obtain the performance proxy model.

[0043] An initial performance proxy model refers to a performance proxy model that has not been trained. For a performance proxy model, it can directly predict the corresponding performance metric based on the input fixed-value combination data. For each set of fixed-value combination data, the fixed-value combination data and its corresponding performance metric are used as training data to form a training set. The training set is then input into the initial performance proxy model to train it, resulting in the final performance proxy model.

[0044] Step S105: Standardize the feasible region of the fixed value and randomly select a preset number of individuals from it as the protected fixed value population.

[0045] Specifically, standardization refers to mapping values ​​within a given feasible region to a unified interval. For example, values ​​within the feasible region can be mapped to the interval [0, 1]. That is, within the given feasible region [Z_set∈(0.6Ω, 1.2Ω), t_delay∈(0s, 0.1s)], the physical range of Z_set is [0.6Ω, 1.2Ω], which is linearly mapped to the interval [0, 1]: Z_norm = (Z_set - 0.6) / (1.2 - 0.6). The physical range of t_delay is [0s, 0.1s], which is similarly mapped: t_norm = t_delay / 0.1.

[0046] A predetermined number of individuals are randomly selected from the standardized feasible region of the setpoints to form a protection setpoint population. Each individual is used to describe a set of setpoint combinations corresponding to the protection device. For example, the chromosome of each individual i is chromosome_i = [Z_norm, t_norm], which is the standardized setpoint vector.

[0047] Step S106: Input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population.

[0048] The protected population is input into the performance proxy model, which predicts the performance index for each individual in the protected population and outputs the individual performance index corresponding to that individual.

[0049] Step S107: Based on the individual performance indicators, perform multi-objective constraint optimization on the protection setpoint population to obtain the optimal individual corresponding to the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

[0050] The optimal individual refers to the setpoint combination data that best meets the engineering requirements for individual performance indicators. Based on the individual performance indicators, the objective function value corresponding to each performance indicator is determined. Using the objective function value as the optimization objective, the protection setpoint population is iteratively optimized. Under the premise of satisfying preset constraints, the optimal individual is selected from the Pareto optimal solution set, and this optimal individual is used as the setpoint combination data for the actual use of the protection device. Multi-objective constraint optimization can be performed using the NSGA-II algorithm, SPEA2 algorithm, or MOEA / D algorithm.

[0051] The protection setting processing method for pumped storage power stations provided in this invention utilizes a pumped storage power station simulation model to simulate multiple fault scenarios. This allows for the simulation of numerous rare but dangerous operating conditions, overcoming the limitation of traditional methods that can only verify a limited number of common scenarios. The data obtained through high-precision simulation is closer to the actual field situation than purely theoretical calculations, ensuring the accuracy of subsequent analysis and optimization. Using a trained performance surrogate model for subsequent protection setting optimization replaces time-consuming simulation calculations with rapid model predictions, enabling the complex optimization process to be completed within an acceptable timeframe and improving optimization efficiency. This invention establishes a closed loop of "simulation-training-optimization," eliminating reliance on manual experience or simple calculations, and achieving a shift from "manual experience-based tuning" to "data-driven adaptive optimization."

[0052] Example 2 This invention also provides another method for processing protection settings in pumped storage power stations. This method is based on the method described in the above embodiments. The method focuses on the specific implementation of inputting the set value combination data into the simulation model of the pumped storage power station, simulating multiple fault scenarios, and obtaining the performance data corresponding to the protection device when performing protection actions under each fault scenario.

[0053] Figure 2a A flowchart of another method for processing protection settings in a pumped storage power station provided by an embodiment of the present invention is shown below. Figure 2a As shown, the protection setting processing method for this pumped storage power station may include the following steps: Step S201: Determine the set value combination data when the protection device performs protection action from the set value feasible domain preset for the protection device in the pumped storage power station.

[0054] Step S202: Obtain data for multiple fault scenarios.

[0055] Fault scenario data refers to data describing the possible fault conditions that may occur in various equipment during the operation of a pumped storage power station. In this embodiment of the invention, fault scenario data can be pre-set and stored in a fault scenario database. During the simulation process, the pumped storage power station simulation model reads fault scenario data from the fault scenario database.

[0056] In another possible implementation, acquiring multiple fault scenario data includes: acquiring multiple fault dimension information pre-set for the fault scenario; wherein, the fault dimension information includes information in the following four dimensions: operating condition dimension, fault location dimension, fault type dimension, and power plant system boundary dimension; and sampling each of the fault dimension information according to each of the fault dimension information to obtain fault scenario data.

[0057] The operating condition dimension describes the typical operating states of a pumped storage power station, such as full-load generation, half-load generation, pumping start-up, pumping steady state, phase adjustment, and shutdown. The fault location dimension describes the location of the fault, such as key points within the protection zone (e.g., the beginning and end of the line) and key points outside the protection zone (e.g., the opposite busbar). The fault type dimension describes the electrical type of the fault, such as three-phase short circuit, two-phase short circuit, single-phase grounding (metallic, 100Ω high resistance, 600Ω high resistance), and open circuit. The power station system boundary dimension describes changes in the power grid operating mode, such as N-1 maintenance mode, islanded operation mode, and minimum operating mode. In this embodiment of the invention, the fault dimension information can be pre-set according to actual needs. For example, taking the optimization of the 110kV line distance protection section I setting of a 300MW pumped storage power station as an example, the operating condition dimension selects the two most severe and problematic operating conditions of the power station: {pumping start-up, full-load generation}. Fault location dimension: Covers key points inside and outside the protection zone {line start (within the zone), line end (within the zone), opposite busbar (outside the zone)}. Fault type dimension: Includes conventional and edge faults {AB phase-to-phase short circuit, A-phase high-resistance grounding (300Ω)}. System boundary dimension: Considers vulnerable grid operation modes {N-1 main transformer maintenance, islanded operation}.

[0058] Based on the pre-defined fault dimension information, the Latin hypercube sampling method can be used to sample the fault dimension information of each dimension, obtaining fault scenario data corresponding to multiple fault scenarios. Specifically, Figure 2b This is a schematic diagram of the Latin hypercube sampling method provided in an embodiment of the present invention. The Latin hypercube sampling method divides the value range of each dimension into N equal intervals (N being the number of sampling points), randomly selects one sample from each interval of each dimension, and randomly combines the samples selected from each dimension to form N sets of sampling points. Accordingly, in this embodiment of the invention, multiple sets of orthogonalized fault scenario combinations are generated within the four-dimensional space composed of the operating condition dimension, fault location dimension, fault type dimension, and power plant system boundary dimension to ensure that the values ​​of each dimension are uniformly covered and to avoid sampling bias. For example, 1200 sets of uniformly distributed fault scenario data can be generated. The generated fault scenario data can be configured using a CSV scenario configuration file containing 1200 rows, with each row defining a unique scenario ID and its four-dimensional parameters.

[0059] Furthermore, for the sampled fault scenario data, risk scoring can be performed based on a historical fault database. Specifically, based on the power plant's historical fault database, the historical frequency of protection maloperation and failure to operate under each scenario is statistically analyzed, and risk scores are assigned to the fault scenario data in conjunction with expert experience. For example, the "islanded operation + high-resistance grounding" scenario is assigned a high-risk score (>0.8) because it has previously caused protection failure to operate. In subsequent simulations, the simulations of the 300 scenarios with the highest risk scores can be prioritized to ensure that limited simulation resources are focused on high-risk, high-value scenarios. A higher risk score indicates that the fault scenario data is more likely to occur.

[0060] Step S203: For each fault scenario data, the set value combination data and the fault scenario data are input into the pumped storage power station simulation model to obtain the performance data corresponding to the protection device performing protection actions under the fault scenario.

[0061] For each set of setpoint combination data, the setpoint combination data is input into the pumped storage power station simulation model to set the protection setpoints for the protection device. For each fault scenario data, the fault scenario data is input into the pumped storage power station simulation model to simulate the fault scenario for the protection device and collect the corresponding performance data when the protection device performs the protection action under the fault scenario.

[0062] Step S204: Extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data.

[0063] Step S205: Use the fixed value combination data and the performance index as a training set to train the initial performance proxy model to obtain the performance proxy model.

[0064] Step S206: Standardize the feasible region of the fixed value and randomly select a preset number of individuals from it as the protected fixed value population.

[0065] Step S207: Input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population.

[0066] Step S208: Based on the individual performance indicators, perform multi-objective constraint optimization on the protection setpoint population to obtain the optimal individual corresponding to the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

[0067] Furthermore, after performing multi-objective constraint optimization on the protection setpoint population based on the individual performance indicators to obtain the optimal individual corresponding to the protection setpoint population, and using the optimal individual as the setpoint combination data actually used by the protection device, the process further includes: obtaining the historical setpoint combination data and corresponding historical performance indicators of the protection device; inputting the setpoint combination data corresponding to the optimal individual into the performance proxy model to obtain optimized performance indicators; comparing the historical setpoint combination data and the historical performance indicators with the setpoint combination data corresponding to the optimal individual and the optimized performance indicators to generate an optimization report, which is then fed back to the power plant operation and maintenance management system.

[0068] Historical setpoint combination data refers to the data used by the protection device before optimization. After inputting the historical setpoint combination data into the pumped storage power station simulation model, simulations are performed on multiple fault scenarios to obtain historical performance data corresponding to the protection device's protective actions under each fault scenario. Feature extraction is performed on each historical performance data point to obtain historical performance indicators corresponding to the historical setpoint combination data. The setpoint combination data corresponding to the optimal individual is input into the performance proxy model to obtain the optimized performance indicators output by the performance proxy model. For each value in the historical performance indicators, a comparison is made with each value in the optimized performance indicators. The historical and optimized performance indicators can be presented directly in tabular form, or further optimization suggestions can be provided to form an optimization report, which is then sent to the power station operation and maintenance management system. Relevant personnel can use the power station operation and maintenance management system to set the protection settings of the actual equipment in the pumped storage power station based on the setpoint combination data corresponding to the optimal individual.

[0069] For example, the optimization report may include the set value combination data corresponding to the best individual (such as Z_set=0.72Ω), the expected performance improvement comparison table (i.e., historical performance indicators and optimized performance indicators), and the simulation waveform comparison diagram of key risk scenarios (such as the original waveform recording file of the protection device under the "isolated network + high impedance" scenario).

[0070] By simulating the fixed value combination data corresponding to the optimal individual and comparing it with historical performance indicators, the effect of optimization on the performance improvement of the protection device can be intuitively verified, which facilitates the understanding and decision-making of relevant personnel.

[0071] The protection setting processing method for pumped storage power stations provided in this invention uses Latin hypercube sampling in a four-dimensional space to ensure that the values ​​of each dimension are evenly covered, avoid sampling bias, fully cover the complex faults and edge conditions unique to pumped storage power stations, make up for the blind spots of traditional setting, provide rich training data for the subsequent training of the initial performance proxy model, and improve the accuracy of the performance proxy model.

[0072] Example 3 This invention also provides another method for processing protection settings in pumped storage power stations; this method is implemented based on the method in the above embodiments; the method focuses on describing the specific implementation of using the setpoint combination data and the performance index as a training set to train the initial performance proxy model and obtain the performance proxy model.

[0073] Figure 3 A flowchart of another method for processing protection settings in a pumped storage power station provided by an embodiment of the present invention is shown below. Figure 3 As shown, the protection setting processing method for this pumped storage power station may include the following steps: Step S301: Determine the set value combination data when the protection device performs protection action from the set value feasible domain preset for the protection device in the pumped storage power station.

[0074] Step S302: Input the set value combination data into the pumped storage power station simulation model, simulate multiple fault scenarios, and obtain the performance data corresponding to the protection device when performing protection actions under each fault scenario.

[0075] Step S303: Extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data.

[0076] Step S304: Standardize the fixed value combination data and the performance index to obtain a standardized training set.

[0077] The fixed-value combination data and performance metrics are mapped to a unified interval and then standardized to obtain a standardized training set. For example, for the fixed-value combination data X = [Z_set, t_delay], where the physical range of Z_set is [0.6Ω, 1.2Ω], it is linearly mapped to the interval [0, 1]: Z_norm = (Z_set - 0.6) / (1.2 - 0.6). The physical range of t_delay is [0s, 0.1s], and it is similarly mapped: t_norm = t_delay / 0.1, resulting in the standardized fixed-value combination data X_norm = [Z_norm, t_norm]. For the performance metrics, Y = [selectivity_rate, sensitivity_margin], where selectivity_rate is the speed, which is a percentage and requires no additional scaling. sensitivity_margin is the fit margin, ranging from [1.0, 2.0], and its original value is used directly.

[0078] Step S305: Input the standardized training set into the initial performance proxy model and train the initial performance proxy model to obtain the performance proxy model.

[0079] In this embodiment of the invention, a Gaussian process regression (GPR) algorithm is employed, using a fixed-value vector [Z_set, t_delay] as input, [selectivity pass rate, sensitivity margin] as the main output, and speed and fit margin as constraints. A radial basis function (RBF) kernel function is used, and hyperparameters are optimized through cross-validation to train the initial performance proxy model. Specifically, a standardized training set is input into the initial performance proxy model for training. The standardized fixed-value combination data serves as the input vector of the initial performance proxy model, and the standardized performance index serves as the output vector. The Gaussian process regression algorithm is a non-parametric probabilistic model that predicts a Gaussian distribution for each input point. In the Gaussian process regression algorithm, the mean function is a constant mean function m(X) = c, where the constant c is optimized as a hyperparameter during model training. The covariance function (kernel function) is an anisotropic radial basis function (ARDRBF) kernel. The ARD property allows the model to learn the differences in the impact of different input dimensions on the output (e.g., Z_set has a much greater impact on sensitivity than t_delay). For the noisy model, the observed noise is Gaussian white noise with a variance of σ_n², which is also optimized during training. The hyperparameter optimization objective is to maximize the Log Marginal Likelihood, and the L-BFGS-B optimizer is used to search within a reasonable range of hyperparameters. After optimization, the optimal hyperparameter set is obtained. After training, the coefficient of determination R is calculated using the test set (simulation data not used in training). 2 A value >0.95 indicates that the model can explain more than 95% of the performance changes, demonstrating high prediction accuracy.

[0080] Further, after inputting the standardized training set into the initial performance proxy model and training the initial performance proxy model to obtain the performance proxy model, the process further includes: obtaining multiple verification fixed-value combination data in the fixed-value feasible region; inputting each verification fixed-value combination data into the performance proxy model and predicting the performance index corresponding to the verification fixed-value combination data to obtain the prediction uncertainty corresponding to each verification fixed-value combination data; based on the prediction uncertainty, selecting supplementary fixed-value combination data from the verification fixed-value combination data; inputting the supplementary fixed-value combination data into the pumped storage power station simulation model and simulating multiple fault scenarios to obtain the verification performance data corresponding to the supplementary fixed-value combination data; adding the supplementary fixed-value combination data and the corresponding verification performance index to the training set and updating the training set; returning to the step of standardizing the fixed-value combination and the performance index to obtain a standardized training set, until the first iteration condition is met.

[0081] Using a trained performance proxy model, predictions are made for the remaining unsimulated fixed-value combinations within the feasible region, and their prediction uncertainty (prediction variance) is calculated. Specifically, multiple validation fixed-value combinations can be obtained from the feasible region; these validation combinations differ from the standard fixed-value combinations. The validation fixed-value combinations are input into the performance proxy model to obtain the prediction uncertainty output by the model. It is understood that the performance proxy model in this embodiment utilizes a Gaussian process regression algorithm. The Gaussian regression algorithm's prediction output for the fixed-value combinations is a Gaussian distribution, including the prediction mean and prediction variance. By sorting the validation fixed-value combinations according to their prediction uncertainty, those with higher prediction uncertainty can be selected as supplementary fixed-value combinations. For example, the validation fixed-value combinations can be sorted from highest to lowest prediction uncertainty, and the top 20 can be selected as supplementary fixed-value combinations. Supplementary setpoint combination data is input into the pumped storage power station simulation model, and simulations are performed on multiple fault scenarios to obtain the performance data of the protection device when performing protection actions under each fault scenario, i.e., verification performance data. The supplementary setpoint combination data and the corresponding verification performance indicators are added to the training set, and the training set is updated to obtain a new training set. The performance proxy model is then retrained using the new training set. Training continues until the uncertainty of the performance proxy model is below a threshold (e.g., variance < 0.01) or the total number of simulations reaches the upper limit, satisfying the first iteration condition, and training ends.

[0082] Step S306: Standardize the feasible region of the fixed value and randomly select a preset number of individuals from it as the protected fixed value population.

[0083] Step S307: Input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population.

[0084] Step S308: Based on the individual performance indicators, perform multi-objective constraint optimization on the protection setpoint population to obtain the optimal individual corresponding to the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

[0085] The pumped storage power station protection setting processing method provided in this invention trains the initial performance surrogate model using standardized setting combination data. This leverages the zero mean and unit variance of the standardized data to reduce the number of iterations during training and improve training efficiency. After the performance surrogate model is trained, an active learning iterative mechanism is further used to utilize the predictive uncertainty of the performance surrogate model to filter and supplement the training set with additional setting combination data. This eliminates the need for time-consuming simulation processes to verify the selection of setting combination data, reducing the number of simulations and further improving the accuracy and training efficiency of the trained performance surrogate model.

[0086] Example 4 This invention also provides another method for processing protection settings in pumped storage power stations. This method is based on the method described in the above embodiments. The method focuses on describing how to perform multi-objective constraint optimization on the protection setting population according to the individual performance indicators to obtain the optimal individual corresponding to the protection setting population, and how to use the optimal individual as the setting combination data actually used by the protection device.

[0087] Figure 4 A flowchart of another method for processing protection settings in a pumped storage power station provided by an embodiment of the present invention is shown below. Figure 4 As shown, the protection setting processing method for this pumped storage power station may include the following steps: Step S401: Determine the setpoint combination data when the protection device performs protection actions from the setpoint feasible domain preset for the protection device in the pumped storage power station.

[0088] Step S402: Input the set value combination data into the pumped storage power station simulation model, simulate multiple fault scenarios, and obtain the performance data corresponding to the protection device when performing protection actions under each fault scenario.

[0089] Step S403: Extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data.

[0090] Step S404: Use the fixed value combination data and the performance index as a training set to train the initial performance proxy model to obtain the performance proxy model.

[0091] Step S405: Standardize the feasible region of the fixed value and randomly select a preset number of individuals from it as the protected fixed value population.

[0092] In this embodiment of the invention, the multi-objective optimization algorithm is the NSGA-II algorithm. For example, the mechanical energy of the NSGA-II algorithm can be configured as follows: population size (N_pop): 50; maximum number of generations (N_gen): 50; encoding method: real number encoding; each individual i has a chromosome_i = [Z_norm, t_norm], which is the standardized fixed-value combination data.

[0093] Step S406: Input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population.

[0094] Step S407: Determine the objective function value for optimization based on the individual performance indicators.

[0095] The objective function value is a numerical indicator used to quantify the performance of each individual in a multi-objective optimization process. The objective function value is determined by selecting the highest selectivity pass rate and the highest sensitivity threshold from the individual performance indicators corresponding to each individual.

[0096] Step S408: Based on the objective function values, select, crossover, and mutate individuals in the protected fixed-value population to generate a progeny fixed-value population.

[0097] For example, since quickness bias is strongly correlated with t_delay, a simple rule of thumb can be established: if t_delay > 0.01s, then quickness bias > 10ms. During the selection phase, any individual that violates the hard constraint (t_delay > 0.01s) has its Domination Rank forced to be the worst, ensuring that it will not enter the next generation.

[0098] During the crossover phase, a simulated binary crossover operator (SBX) is used, with a crossover probability p_c = 0.9 and a distribution index η_c = 20 (which controls the closeness between offspring and parents; the larger the value, the closer they are).

[0099] During the mutation phase, a multinomial mutation operator is used, with a mutation probability p_m = 1 / 2 = 0.5 (mutation is independent in each dimension) and a distribution exponent η_m = 20.

[0100] In this embodiment of the invention, the standard operation of NSGA-II is performed to select, crossover, and mutate individuals in the protected value population to generate a progeny value population Q_{gen}.

[0101] Step S409: The protected fixed-value population is merged with the offspring fixed-value population to obtain a mixed fixed-value population.

[0102] The protected population is used as the parent population, and the offspring population is merged to obtain a mixed population. For example, the protected population has 50 individuals, the offspring population has 50 individuals, and the resulting mixed population has 100 individuals.

[0103] Step S410: Based on the non-dominated order and crowding distance of each individual in the mixed fixed-value population, the individuals in the mixed fixed-value population are screened to obtain the updated protected fixed-value population.

[0104] For each individual in the mixed fixed-value population, the number of times they are dominated and the set of individuals they dominate are calculated. Based on these values, the individuals are added to different frontier sets. For each individual in each frontier set, the crowding distance is calculated. The individuals in each frontier set are sorted according to the crowding distance, and a predetermined number of individuals are selected as the updated protected fixed-value population. The non-dominated sorting and crowding distance methods used in this embodiment are prior art and will not be described in detail here.

[0105] Step S411: Return to the step of inputting the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population, until the second iteration condition is met.

[0106] The iteration stops when the number of iterations reaches the preset maximum number of iterations, or when the Pareto front changes less than a preset threshold for several consecutive generations, satisfying the second iteration condition. In this embodiment of the invention, the non-dominated solution set in the updated protected population is the Pareto front.

[0107] Step S412: Based on the preset engineering screening conditions, select the optimal individual from the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

[0108] The engineering screening criteria can be set according to actual engineering needs. For example, based on the power plant operation and maintenance regulations, the minimum requirements are set as follows: selectivity qualification rate ≥ 99% and sensitivity margin ≥ 1.5. Based on the individual performance indicators corresponding to each individual in the protection setpoint population, screening is performed according to the set minimum requirements to obtain at least one individual that meets the minimum requirements, which is considered the solution set. Within the solution set that meets the requirements, the individual with the highest sensitivity margin among the individual performance indicators is selected as the optimal individual, as it provides the largest safety margin.

[0109] The pumped storage power station protection setting processing method provided in this embodiment of the invention optimizes selectivity and sensitivity as target performance indicators, thereby improving the performance indicators after optimization.

[0110] Example 5 Corresponding to the above method embodiments, this invention provides a protection setting processing device for pumped storage power stations. Figure 5 This is a schematic diagram of the structure of a protection setting processing device for a pumped storage power station provided in an embodiment of the present invention, as shown below. Figure 5 As shown, the protection setting processing device for the pumped storage power station may include: The setting combination determination module 501 is used to determine the setting combination data when the protection device performs protection action from the setting feasible domain that is preset for the protection device in the pumped storage power station. Simulation module 502 is used to input the set value combination data into the pumped storage power station simulation model, simulate multiple fault scenarios, and obtain the performance data of the protection device when performing protection actions under each fault scenario. Feature extraction module 503 is used to extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data; Training module 504 is used to train the initial performance proxy model using the fixed value combination data and the performance index as a training set to obtain the performance proxy model; The standardization module 505 is used to standardize the feasible domain of the set value and randomly select a preset number of individuals from it as the protection set value population; wherein, the individuals are used to describe a set of set value combination data corresponding to the protection device; The performance index determination module 506 is used to input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population; The optimization module 507 is used to perform multi-objective constraint optimization processing on the protection setpoint population according to the individual performance indicators to obtain the optimal individual corresponding to the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

[0111] The pumped storage power station protection setting processing device provided in this invention utilizes a pumped storage power station simulation model to simulate multiple fault scenarios. This allows for the simulation of numerous rare but dangerous operating conditions, overcoming the limitation of traditional methods that can only verify a limited number of common scenarios. The data obtained through high-precision simulation is closer to the actual field situation than purely theoretical calculations, ensuring the accuracy of subsequent analysis and optimization. Using a trained performance surrogate model for subsequent protection setting optimization replaces time-consuming simulation calculations with rapid model predictions, enabling the complex optimization process to be completed within an acceptable timeframe and improving optimization efficiency. This invention establishes a closed loop of "simulation-training-optimization," eliminating reliance on manual experience or simple calculations, and achieving a shift from "manual experience-based tuning" to "data-driven adaptive optimization."

[0112] In some embodiments, the simulation module 502 is further configured to: Acquire data from multiple fault scenarios; For each fault scenario data, the set value combination data and the fault scenario data are input into the pumped storage power station simulation model to obtain the performance data corresponding to the protection device performing protection actions under the fault scenario.

[0113] In some embodiments, the simulation module 502 is further configured to: Acquire multiple fault dimension information pre-set for fault scenarios; wherein, the fault dimension information includes the following four dimensions: operating condition dimension, fault location dimension, fault type dimension, and power plant system boundary dimension; Based on the fault dimension information, the fault dimension information is sampled to obtain fault scenario data.

[0114] In some embodiments, the training module 504 is further configured to: The fixed-value combination data and the performance index are standardized to obtain a standardized training set; The standardized training set is input into the initial performance proxy model to train the initial performance proxy model and obtain the performance proxy model.

[0115] In some embodiments, the device further includes: A verification data determination module is used to obtain multiple verification value combination data in the feasible domain of the set value; wherein the verification value combination data is different from the set value combination data; The verification performance index determination module is used to input the verification setpoint combination data into the performance proxy model, verify the performance proxy model, and obtain the verification performance index and prediction uncertainty corresponding to each verification setpoint combination data. The filtering module is used to filter out supplementary fixed-value combination data from the verification fixed-value combination data based on the prediction uncertainty; The supplementary module is used to add the supplementary fixed-value combination data and the corresponding verification performance indicators to the training set and update the training set. The return module is used to return the steps of performing standardization processing on the fixed value combination and the performance index to obtain a standardized training set, until the first iteration condition is met.

[0116] In some embodiments, the optimization module 507 is further configured to: Based on the individual performance indicators, determine the target function value for optimization; Based on the objective function values, individuals in the protected fixed-value population are selected, crossovered, and mutated to generate offspring fixed-value populations; The protected population with the offspring population with the value of the parent population is merged to obtain a mixed population with the value of the parent population. Based on the non-dominated ordering and crowding distance of each individual in the mixed fixed-value population, the individuals in the mixed fixed-value population are screened to obtain an updated protected fixed-value population; Return to the step of inputting the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population, until the second iteration condition is met; According to the preset engineering screening conditions, the optimal individual is selected from the protection setpoint population, and the optimal individual is used as the setpoint combination data for actual use of the protection device.

[0117] In some embodiments, the device further includes: The historical data acquisition module is used to acquire historical setpoint combination data and corresponding historical performance indicators of the protection device; The optimized data acquisition module is used to input the fixed-value combination data corresponding to the optimal individual into the performance proxy model to obtain optimized performance indicators; The comparison module is used to compare the historical fixed value combination data and the historical performance indicators with the fixed value combination data and the optimized performance indicators corresponding to the optimal individual, generate an optimization report, and feed it back to the power plant operation and maintenance management system.

[0118] The device provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0119] Example 5 This invention also provides an electronic device for running the above-described pumped storage power station protection setting processing method; see also Figure 6 The diagram shows the structure of an electronic device, which includes a memory 600 and a processor 601. The memory 600 stores one or more computer instructions, which are executed by the processor 601 to implement the above-mentioned pumped storage power station protection setting processing method.

[0120] Furthermore, Figure 6 The electronic device shown also includes a bus 602 and a communication interface 603, with the processor 601, communication interface 603 and memory 600 connected via the bus 602.

[0121] The memory 600 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 603 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 602 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0122] Processor 601 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 601 or by instructions in software form. Processor 601 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a readily available storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 600. Processor 601 reads information from memory 600 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiments.

[0123] This invention also provides a computer-readable storage medium storing computer-executable instructions. When these computer-executable instructions are called and executed by a processor, they cause the processor to implement the above-described pumped storage power station protection setting processing method. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0124] The computer program product for processing protection settings of pumped storage power stations provided in this embodiment of the invention includes a computer-readable storage medium storing non-volatile program code executable by a processor. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0125] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0126] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0127] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0128] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0129] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0130] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method of protection setting value processing for a pumped storage power station, characterized by, include: From the feasible range of preset setpoints for the protection device in the pumped storage power station, determine the setpoint combination data when the protection device performs the protection action; Acquire data from multiple fault scenarios; For each fault scenario data, the set value combination data and the fault scenario data are input into the pumped storage power station simulation model to obtain the performance data corresponding to the protection device performing protection actions under the fault scenario; wherein, the pumped storage power station simulation model is used to simulate the operating status of the protection device in the pumped storage power station and simulate the protection behavior taken by the protection device under different fault scenarios. Feature extraction is performed on each of the performance data to obtain the performance index corresponding to the fixed value combination data; wherein, the performance index is used to describe the protection device's ability to handle fault conditions; Using the fixed-value combination data and the performance indicators as the training set, the initial performance proxy model is trained by using the fixed-value combination data as input and the performance indicators as output and constraints to obtain the performance proxy model; wherein, the performance indicators include selective pass rate, sensitivity margin, speed deviation and fit margin. The feasible domain of the set value is standardized, and a preset number of individuals are randomly selected from it as the protection set value population; wherein, the individuals are used to describe a set of set value combination data corresponding to the protection device; The protection setpoint population is input into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population; Based on the individual performance indicators, the protection setpoint population is subjected to multi-objective constraint optimization to obtain the optimal individual corresponding to the protection setpoint population, and the optimal individual is used as the setpoint combination data actually used by the protection device.

2. The method of claim 1, wherein, The acquisition of multiple fault scenario data includes: Acquire multiple fault dimension information pre-set for fault scenarios; wherein, the fault dimension information includes the following four dimensions: operating condition dimension, fault location dimension, fault type dimension, and power plant system boundary dimension; Based on the fault dimension information, the fault dimension information is sampled to obtain fault scenario data.

3. The method of claim 1, wherein, The step of using the fixed-value combination data and the performance indicators as a training set to train the initial performance proxy model to obtain the performance proxy model includes: The fixed-value combination data and the performance index are standardized to obtain a standardized training set; The standardized training set is input into the initial performance proxy model to train the initial performance proxy model and obtain the performance proxy model.

4. The method of claim 3, wherein, After inputting the standardized training set into the initial performance proxy model and training the initial performance proxy model to obtain the performance proxy model, the process further includes: Multiple verification value combination data are obtained in the feasible domain of the fixed value; wherein the verification value combination data is different from the fixed value combination data; Each of the verification setpoint combination data is input into the performance proxy model, and the performance index corresponding to the verification setpoint combination data is predicted to obtain the prediction uncertainty corresponding to each of the verification setpoint combination data. Based on the predicted uncertainty, supplementary fixed-value combination data are selected from the verified fixed-value combination data; The supplementary setpoint combination data is input into the pumped storage power station simulation model to simulate multiple fault scenarios and obtain the supplementary performance data corresponding to the supplementary setpoint combination data. The supplementary fixed-value combination data and the corresponding supplementary performance indicators are added to the training set, and the training set is updated. Return to the step of standardizing the fixed value combination and the performance index to obtain a standardized training set, until the first iteration condition is met.

5. The method of claim 1, wherein, The step of performing multi-objective constraint optimization on the protection setpoint population based on the individual performance indicators to obtain the optimal individual corresponding to the protection setpoint population, and using the optimal individual as the setpoint combination data actually used by the protection device, includes: Based on the individual performance indicators, determine the target function value for optimization; Based on the objective function values, individuals in the protected fixed-value population are selected, crossovered, and mutated to generate offspring fixed-value populations; The protected population with the offspring population with the value of the parent population is merged to obtain a mixed population with the value of the parent population. Based on the non-dominated ordering and crowding distance of each individual in the mixed fixed-value population, the individuals in the mixed fixed-value population are screened to obtain an updated protected fixed-value population; Return to the step of inputting the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population, until the second iteration condition is met; According to the preset engineering screening conditions, the optimal individual is selected from the protection setpoint population, and the optimal individual is used as the setpoint combination data for actual use of the protection device.

6. The method of claim 1, wherein, The step of performing multi-objective constraint optimization on the protection setpoint population based on the individual performance indicators to obtain the optimal individual corresponding to the protection setpoint population, and then using the optimal individual as the setpoint combination data actually used by the protection device, further includes: Obtain the historical setpoint combination data and corresponding historical performance indicators of the protection device; The fixed-value combination data corresponding to the optimal individual is input into the performance proxy model to obtain the optimized performance index; The historical setpoint combination data and the historical performance indicators are compared with the setpoint combination data and the optimized performance indicators corresponding to the optimal individual, an optimization report is generated, and the report is fed back to the power plant operation and maintenance management system.

7. A protection setting processing device for a pumped storage power station, characterized in that, include: The setting combination determination module is used to determine the setting combination data when the protection device performs protection actions from the setting feasible domain that is preset for the protection device in the pumped storage power station. The simulation module is used to acquire data from multiple fault scenarios; For each fault scenario data, the set value combination data and the fault scenario data are input into the pumped storage power station simulation model to obtain the performance data corresponding to the protection device performing protection actions under the fault scenario; wherein, the pumped storage power station simulation model is used to simulate the operating status of the protection device in the pumped storage power station and simulate the protection behavior taken by the protection device under different fault scenarios. The feature extraction module is used to extract features from each of the performance data to obtain the performance index corresponding to the fixed value combination data; wherein, the performance index is used to describe the protection device's ability to handle fault conditions; The training module is used to train the initial performance proxy model by taking the fixed value combination data and the performance indicators as the training set, using the fixed value combination data as the input, and constructing outputs and constraints with the performance indicators to obtain the performance proxy model; wherein, the performance indicators include selective pass rate, sensitivity margin, speed deviation and fit margin. The standardization module is used to standardize the feasible domain of the set value and randomly select a preset number of individuals from it as the protection set value population; wherein, the individuals are used to describe a set of set value combination data corresponding to the protection device; The performance index determination module is used to input the protection setpoint population into the performance proxy model to obtain the individual performance index corresponding to each individual in the protection setpoint population; The optimization module is used to perform multi-objective constraint optimization on the protection setpoint population based on the individual performance indicators to obtain the optimal individual corresponding to the protection setpoint population, and use the optimal individual as the setpoint combination data actually used by the protection device.

8. An electronic device, characterized in that, The device includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the pumped storage power station protection setting processing method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the pumped storage power station protection setting processing method according to any one of claims 1 to 6.

Images