Centralized control automation strategy optimization method and system based on fuzzy logic

Abstract

The invention provides a centralized control automation strategy optimization method and system based on fuzzy logic, and the method comprises the steps: collecting the real-time operation state data of power grid operation equipment in a target centralized control region, covering the operation parameters of a power grid and a parameter fluctuation range, and carrying out the fuzzy membership processing of the data according to the parameter fluctuation range, and generating a fuzzy rule set containing a plurality of fuzzy rules and parameter dynamic adjustment logic corresponding to each fuzzy rule, optimizing a current control strategy by applying fuzzy logic based on the fuzzy rule set, and generating an optimized control strategy containing a control instruction and a trigger condition thereof, and then the strategy is sent to execution equipment according to a trigger condition, execution feedback data is monitored, the validity of the strategy is verified according to a matching result of the execution feedback data and a preset threshold value, the control strategy is dynamically adjusted and optimized, a centralized control automation strategy can be effectively optimized, and the power grid operation control effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and more particularly, to a method and system for optimizing a centralized control automation strategy based on fuzzy logic. Background Art

[0002] In the current era of continuous development and transformation of the power system, centralized control automation plays a crucial role in ensuring the stable and efficient operation of the power grid. With the increasing expansion of the power grid scale, the increasingly complex structure, and the diverse growth of power demand, the need to optimize the centralized control automation strategy has become extremely urgent.

[0003] When traditional centralized control automation strategies handle problems related to power grid operation, most of the control strategy formulations adopt a fixed rule approach. These fixed rules are often preset based on specific and relatively ideal operating scenarios, lacking the effective ability to cope with the complex variability of power grid operation parameters. Once the power grid operating conditions change beyond the applicable range of the preset rules, these control strategies are difficult to play an effective role, unable to flexibly adjust according to the actual situation, resulting in poor control effects and even potentially affecting the stable operation of the power grid.

[0004] In the process of optimizing control strategies, the prior art fails to fully consider the mutual correlation between various power grid operation parameters and the dynamic changes in the operating environment, making the optimized control strategies still have certain limitations and unable to meet the precise control requirements in complex power grid operating scenarios. Summary of the Invention

[0005] In view of the problems mentioned above, in combination with the first aspect of the present invention, embodiments of the present invention provide a method for optimizing a centralized control automation strategy based on fuzzy logic, the method comprising: Collect real-time operating status data of power grid operating equipment in a target centralized control area, the real-time operating status data including at least one power grid operation parameter and the parameter fluctuation range corresponding to the power grid operation parameter; According to the parameter fluctuation range, perform fuzzy membership processing on the real-time operating status data to generate a fuzzy rule set matching the power grid operation parameter, the fuzzy rule set including a plurality of fuzzy rules, and each fuzzy rule corresponding to a dynamic adjustment logic of a power grid operation parameter; Based on the fuzzy rule set, perform fuzzy logic optimization on the current control strategy of the target centralized control area to generate an optimized control strategy, the optimized control strategy including a plurality of control instructions and the triggering conditions of each control instruction; According to the triggering conditions, send the optimized control strategy to the execution device in the target centralized control area and monitor the execution feedback data of the execution device for the control instruction; Verify the effectiveness of the optimization control strategy based on the matching result between the execution feedback data and the preset feedback threshold, and dynamically adjust the optimization control strategy according to the verification result.

[0006] In another aspect, an embodiment of the present invention further provides a centralized control automation strategy optimization system based on fuzzy logic, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0007] Based on the above aspects, in the embodiment of the present application, by collecting the real-time operation status data of power grid operation equipment and performing fuzzy membership processing in combination with the parameter fluctuation range, a fuzzy rule set is generated, which can effectively quantify and analyze the uncertainty of power grid operation parameters through fuzzy logic, and generate a dynamic adjustment logic matching different operation parameters. In the control strategy optimization link, based on the fuzzy rule set, fuzzy logic optimization is performed on the current control strategy to generate an optimized control strategy including multiple control instructions and their triggering conditions, realizing the intelligent optimization of the control strategy in the centralized control area. Considering various operation parameters and their dynamic changes, the optimized control strategy can better adapt to the complex working conditions of power grid operation, improve the accuracy and timeliness of the control strategy, and effectively avoid the hysteresis and inadaptability problems that may occur in traditional control strategies. In terms of strategy execution and feedback verification, the optimized control strategy is sent to the execution device according to the triggering condition, and the execution feedback data is monitored in real time. Based on the matching result between the execution feedback data and the preset feedback threshold, the optimized control strategy is dynamically adjusted, thus forming an adaptive optimization loop, enabling the entire centralized control automation system to continuously self-adjust and optimize according to the actual execution situation. Compared with the traditional method, instead of presetting a fixed control strategy that cannot be flexibly changed according to the actual execution effect, through real-time feedback verification, it is ensured that the optimized control strategy can maintain effectiveness in different power grid operation scenarios, further enhancing the ability of the centralized control automation system to cope with the complex and changeable power grid operation environment and ensuring the stable and efficient operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 is a schematic flowchart of the execution of the centralized control automation strategy optimization method based on fuzzy logic provided by the embodiment of the present invention.

[0009] Figure 2 is a schematic diagram of the hardware architecture of the centralized control automation strategy optimization system based on fuzzy logic provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0010] The present invention will be specifically described below in conjunction with the accompanying drawings of the specification.Figure 1 FIG. Figure 1 is a schematic flow chart of an optimization method for centralized control automation strategy based on fuzzy logic provided by an embodiment of the present invention. The optimization method for centralized control automation strategy based on fuzzy logic will be introduced in detail below.

[0011] Step S110: Collect real-time operation status data of power grid operation equipment in the target centralized control area. The real-time operation status data includes at least one power grid operation parameter and the parameter fluctuation range corresponding to the power grid operation parameter.

[0012] In this embodiment, taking the target centralized control area as a medium-sized power supply area including multiple substations, transmission lines and power loads as an example, in this medium-sized power supply area, there are various types of power grid operation equipment, such as transformers, circuit breakers, capacitor banks, etc.

[0013] For transformers, the power grid operation parameters to be collected include voltage value, current value, oil temperature, etc. The normal range of the voltage value may be set as 10 kV - 10.5 kV, and the currently collected real-time voltage value is 10.3 kV. Its fluctuation range may be between 10.2 kV - 10.4 kV. This is because during the daily power transmission process, affected by factors such as load changes and interference from adjacent lines, the voltage will fluctuate within a certain range.

[0014] Looking at the transmission line again, the collected parameters are line current and power factor. The normal range of the line current is assumed to be 500 A - 600 A, the current real-time current value is 550 A, and the fluctuation range is 540 A - 560 A. The normal range of the power factor is 0.9 - 0.95, the real-time power factor is 0.93, and the fluctuation range is 0.92 - 0.94. These fluctuation ranges are generated because the nature of the loads connected to the line is different. For example, some industrial loads may be inductive loads, which will affect the power factor, and the connection and disconnection of loads at different time periods will cause current fluctuations.

[0015] For power loads, the monitored parameter is load power. The normal range is set as 10 MW - 15 MW, the currently collected real-time load power is 12 MW, and the fluctuation range is 11.5 MW - 12.5 MW. This is because during different production activity periods, such as the working and off-duty hours of factories, the peak and trough periods of business areas, and the peak (such as more electricity consumption at night) and trough (such as less electricity consumption during the day when people are at work) of residential electricity consumption, the load power will change. By collecting various operation parameters of these power grid operation equipment and their corresponding fluctuation ranges, the operation status of the power grid in the target centralized control area can be comprehensively understood.

[0016] Step S120: According to the parameter fluctuation range, perform fuzzy membership processing on the real-time operation status data to generate a fuzzy rule set that matches the grid operation parameters. The fuzzy rule set contains multiple fuzzy rules, and each fuzzy rule corresponds to a dynamic adjustment logic for a grid operation parameter.

[0017] Still taking the above medium-sized power supply area as an example, for the grid operation parameter of transformer voltage, the key parameter type is voltage deviation. Suppose the parameter fluctuation range of voltage deviation is divided into three fuzzy sub-intervals: a large negative deviation interval (below 10 kV), a normal deviation interval (10 kV - 10.5 kV), and a large positive deviation interval (above 10.5 kV). Each fuzzy sub-interval corresponds to a type of membership function. For example, the membership function type corresponding to the large negative deviation interval is a descending half-trapezoidal function.

[0018] When the collected transformer voltage is 10.3 kV, according to the membership function, the membership weight of this voltage value in the normal deviation interval is calculated to be 0.6, indicating that this voltage value has a 60% probability of being in the normal deviation interval. Based on this membership weight and a preset rule template (such as when the voltage is in the normal deviation interval, if the load has an increasing trend, maintain the voltage stable; if the load has a decreasing trend, the voltage can be appropriately reduced to reduce losses), a fuzzy rule subset for this key parameter type of transformer voltage is generated.

[0019] For the parameter of power factor of the transmission line, the fuzzy sub-intervals corresponding to its fluctuation range are divided into a low power factor interval (below 0.9), a qualified power factor interval (0.9 - 0.95), and a high power factor interval (above 0.95). Suppose the current power factor is 0.93, and the calculated membership weight in the qualified power factor interval is 0.8. According to the preset rule template (such as in the qualified power factor interval, determine whether to adjust the reactive power compensation device according to the line current magnitude), a corresponding fuzzy rule subset is generated.

[0020] Finally, logically fuse the fuzzy rule subsets of each key parameter type such as transformer voltage and power factor of the transmission line. For example, when the transformer voltage is in the large positive deviation interval and the power factor of the transmission line is in the low power factor interval, a comprehensive fuzzy rule is generated. The dynamic adjustment logic of this rule may be to first adjust the reactive power compensation device to improve the power factor, and then consider whether to adjust the tap of the transformer to reduce the voltage according to the adjusted situation, so as to generate the entire fuzzy rule set.

[0021] Step S130: Based on the fuzzy rule set, perform fuzzy logic optimization on the current control strategy of the target centralized control area to generate an optimized control strategy. The optimized control strategy contains multiple control instructions and the triggering conditions for each control instruction.

[0022] For example, in this medium-sized power supply area, the set of historical control strategies for the target centralized control area contains multiple historical control instructions and the execution effect evaluation indicators corresponding to each historical control instruction. For example, historically, there was a control instruction that when the transformer oil temperature exceeded 80°C, the cooling device was started, and its execution effect evaluation indicators were that the oil temperature dropped below 75°C within 10 minutes, and the execution success rate was 90%.

[0023] Associate and match the previously generated set of fuzzy rules with the set of historical control strategies. Suppose there is a fuzzy rule in the set of fuzzy rules regarding the transformer oil temperature. When the oil temperature is in the higher range (75°C - 85°C), the corresponding membership weight is 0.7. This fuzzy rule has a mapping relationship with the above historical control instruction.

[0024] Screen candidate control instructions from the set of historical control strategies that meet the preset optimization conditions (the execution effect evaluation indicators are higher than the preset threshold and the triggering conditions match the current power grid operation parameters). If the preset threshold is set to an execution success rate of more than 80%, then the above control instruction regarding the transformer oil temperature meets the conditions and becomes a candidate control instruction.

[0025] Rank the priorities of these candidate control instructions based on the real-time matching degree of the execution effect evaluation indicators and triggering conditions of the candidate control instructions. Suppose there is another control instruction that when the transmission line current exceeds 600A, adjust the setting value of the line protection device, and its execution success rate is 85%. However, since the current control instruction for the transformer oil temperature has a higher real-time matching degree with the current power grid operation parameters (the transformer oil temperature is close to 80°C), the control instruction for the transformer oil temperature has a higher priority, and an initial optimized control strategy is generated.

[0026] Considering the device response delay parameters in the target centralized control area, for example, the start of the transformer cooling device may have a 2-minute delay, correct the execution timing sequence of the control instructions in the initial optimized control strategy. If according to the original strategy, the cooling device is started immediately when the oil temperature reaches 80°C, after correction, it may be that the start instruction is sent when the oil temperature reaches 78°C to make up for the 2-minute delay, thereby generating the final optimized control strategy. This optimized control strategy contains multiple control instructions (such as the control instruction for the transformer oil temperature, the control instruction for the transmission line current, etc.) and the triggering conditions for each control instruction (such as the transformer oil temperature of 78°C, the transmission line current of 600A, etc.).

[0027] Step S140, according to the triggering conditions, send the optimized control strategy to the execution devices in the target centralized control area, and monitor the execution feedback data of the execution devices for the control instructions.

[0028] In this embodiment, in a medium-sized power supply area, the triggering conditions in the parsing and optimization control strategy are analyzed. For example, for the transformer oil temperature control instruction, its triggering condition is that the oil temperature reaches 78°C, and the corresponding threshold range of grid operation parameters is 78°C and above.

[0029] The grid operation parameters of the target centralized control area are monitored in real time. When it is detected that the transformer oil temperature reaches 78°C, the corresponding control instruction is activated, that is, the cooling device of the transformer is started. Then this activated control instruction is encapsulated into the format of an executable instruction for the device and sent to the target execution device (i.e., the cooling device of the transformer) through the centralized control communication network.

[0030] The instruction response signal returned by the target execution device is received, and the timestamp and execution status identifier of the instruction response signal are recorded. Suppose the timestamp of the response signal received from the cooling device is 10:02:00 on August 1, 2023, and the execution status identifier is successfully started. Based on this timestamp and execution status identifier, execution feedback data is generated. Calculate the instruction execution delay time. The time interval from when the oil temperature reaches 78°C (assumed to be 10:00:00 on August 1, 2023) to when the response signal is received is 2 minutes, and the execution success rate is 100%. The above data constitutes the execution feedback data.

[0031] Similarly, for the transmission line current control instruction, when it is detected that the transmission line current reaches 600 A, the control instruction to adjust the setting value of the line protection device is activated, and the instruction is sent, the response signal is received, and the execution feedback data is generated according to the above process.

[0032] Step S150, based on the matching result between the execution feedback data and the preset feedback threshold, verify the effectiveness of the optimization control strategy, and dynamically adjust the optimization control strategy according to the verification result.

[0033] In this embodiment, assume that in the preset feedback threshold, the threshold of the instruction execution delay time is 3 minutes, and the threshold of the execution success rate is 90%. Extract the instruction execution delay time (such as 2 minutes for the transformer cooling device) and the execution success rate (such as 100%) from the execution feedback data, and calculate the comprehensive execution efficiency index of the optimization control strategy.

[0034] For the control instruction of the transformer cooling device, assume that the standard response delay time for this type of equipment, i.e., the transformer, in the target centralized control area is 2.5 minutes. Based on the difference between the instruction execution delay time and the standard response delay time (2 - 2.5 = -0.5 minutes), calculate the delay deviation coefficient, which is assumed to be -0.2 through a specific formula. Based on the difference between the execution success rate and the preset success rate threshold (100% - 90% = 10%), calculate the success rate compensation coefficient, which is assumed to be 0.1 through a formula. According to the delay deviation coefficient and the success rate compensation coefficient, generate the local efficiency index of this control instruction through a weighted summation formula, which is assumed to be 0.08. Normalize the local efficiency indexes of all control instructions (including the transmission line current control instruction, etc.) to obtain the comprehensive execution efficiency index, which is assumed to be 0.9.

[0035] Since the comprehensive execution efficiency index of 0.9 is higher than the preset feedback threshold, it is determined that the optimized control strategy meets the effectiveness standard and does not need to be adjusted.

[0036] However, if the execution success rate is lower than the preset success rate threshold, for example, the execution success rate of the transmission line current control instruction is 80%. Determine the fuzzy rule identifier corresponding to this control instruction, and extract the target fuzzy rule and the type of grid operation parameters associated with the target fuzzy rule (such as transmission line current) from the fuzzy rule set. Obtain the current membership weight and the fuzzy sub-interval division parameters of the target fuzzy rule. Based on the difference between the execution success rate and the preset success rate threshold (80% - 90% = -10%), generate the membership weight adjustment amount, which is assumed to be -0.1. According to the membership weight adjustment amount, update the membership weight of the target fuzzy rule or re-divide the boundary range of the fuzzy sub-interval.

[0037] Based on the adjusted fuzzy rule, re-match the candidate control instructions in the historical control strategy set for conflict detection. If conflict control instructions are detected, for example, two control instructions are for the same operation period of the transmission line but have logical contradictions, according to the execution effect evaluation index, retain the control instruction with the highest priority and eliminate the remaining conflict instructions. Generate an updated set of candidate control instructions based on the elimination result, re-perform priority sorting and timing correction, generate an updated optimized control strategy, and replace the original optimized control strategy with it.

[0038] Continuously monitor the execution feedback data of the optimized control strategy after replacement. Within a preset monitoring period (such as 1 hour), collect multiple batches of execution feedback data. Calculate the moving window average of the comprehensive execution efficiency indicators in the multiple batches of execution feedback data to obtain a dynamic efficiency evaluation value. If the dynamic efficiency evaluation value is higher than the preset feedback threshold in N consecutive (assuming N = 3) monitoring periods, it is determined that the updated optimized control strategy meets the stability standard. If not, trigger the fuzzy rule adjustment and optimized control strategy update process again based on the latest execution feedback data until the stability standard is met. Mark the current optimized control strategy as the final version and generate a unique strategy identifier for it (including the target centralized control area code, generation timestamp, and version serial number). Associate the final version of the optimized control strategy with the strategy identifier, encrypt it, and transmit it to the specified storage partition of the centralized control strategy database. Establish an index relationship between the strategy identifier and the historical control strategy set, delete the copy of the optimized control strategy temporarily stored in the centralized control communication network, release the relevant computing resources, and send a strategy update completion notification (including the strategy identifier and the effective time range) to the monitoring terminal of the target centralized control area.

[0039] Based on the above steps, the embodiment of the present application can effectively quantify and analyze the uncertainty of grid operation parameters through fuzzy logic by collecting real-time operation status data of grid operation equipment and performing fuzzy membership processing in combination with the parameter fluctuation range, and generate dynamic adjustment logics that match different operation parameters. In the control strategy optimization link, based on the fuzzy rule set, perform fuzzy logic optimization on the current control strategy to generate an optimized control strategy including multiple control instructions and their triggering conditions, realizing the intelligent optimization of the control strategy in the centralized control area. Considering various operation parameters and their dynamic changes comprehensively, the optimized control strategy can better adapt to the complex working conditions of grid operation, improve the accuracy and timeliness of the control strategy, and effectively avoid the lag and inadaptability problems that may occur in traditional control strategies. In terms of strategy execution and feedback verification, send the optimized control strategy to the execution device according to the triggering conditions, and continuously monitor the execution feedback data. Dynamically adjust the optimized control strategy based on the matching result between the execution feedback data and the preset feedback threshold, thus forming an adaptive optimization loop, enabling the entire centralized control automation system to continuously self-adjust and optimize according to the actual execution situation. Compared with traditional methods, instead of presetting a fixed control strategy that cannot be flexibly changed according to the actual execution effect, through real-time feedback verification, it is ensured that the optimized control strategy can maintain effectiveness in different grid operation scenarios, further enhancing the ability of the centralized control automation system to cope with the complex and changeable grid operation environment and ensuring the stable and efficient operation of the grid.

[0040] In a possible implementation manner, step S120 includes: Step S121, extract the key parameter types from the grid operation parameters, where the key parameter types include voltage deviation, load volatility, and frequency offset.

[0041] Step S122, for each key parameter type, determine a plurality of fuzzy sub-intervals corresponding to the parameter fluctuation range, and each fuzzy sub-interval corresponds to a membership function type.

[0042] Step S123, according to the membership function type, assign a corresponding membership weight to each fuzzy sub-interval, and the membership weight is used to quantify the distribution probability of the real-time operation state data within the fuzzy sub-interval.

[0043] Step S124, based on the membership weight and a preset rule template, generate a fuzzy rule subset corresponding to each key parameter type, where the preset rule template contains control logic mapping relationships under different grid operation scenarios.

[0044] Step S125, perform logical fusion on the fuzzy rule subsets of each key parameter type to generate the fuzzy rule set.

[0045] In the medium-sized power supply area scenario mentioned above, voltage deviation is a key parameter type. For example, the normal operating voltage range of the transformer is set to 10 kV - 10.5 kV, and the actual measured voltage value is 10.3 kV, so its voltage deviation needs to be focused on. Load volatility is also a key parameter type. The power load fluctuates significantly at different times, and the normal load power range is 10 MW - 15 MW. The actual load power fluctuation reflects the change of the load. Frequency offset is equally important. The standard frequency of the power grid is 50 Hz, and there may be a certain offset during actual operation.

[0046] For voltage deviation, it can be divided into a large negative deviation range (below 10 kV), a small negative deviation range (10 kV - 10.1 kV), a normal range (10.1 kV - 10.4 kV), a small positive deviation range (10.4 kV - 10.5 kV), and a large positive deviation range (above 10.5 kV). Each fuzzy sub-range corresponds to a type of membership function. For example, the membership function type corresponding to the large negative deviation range is a descending semi-trapezoidal function, the small negative deviation range corresponds to a left triangle function, the normal range corresponds to a triangle function, the small positive deviation range corresponds to a right triangle function, and the large positive deviation range corresponds to an ascending semi-trapezoidal function. According to these membership function types, the corresponding membership weights are assigned to each fuzzy sub-range. When the collected voltage is 10.3 kV, it is calculated that the membership weight of this voltage value in the normal range is 0.8, indicating that there is an 80% probability that this voltage value is in the normal range. Based on this membership weight and the preset rule template (for example, when the voltage is in the normal range and the load is stable, maintain the current voltage control strategy; if the load has an upward trend and the voltage is close to the upper limit of the normal range, the reactive power compensation device can be appropriately adjusted to stabilize the voltage), a fuzzy rule subset corresponding to the key parameter type of voltage deviation is generated.

[0047] For load volatility, determine the fuzzy sub-ranges corresponding to its parameter fluctuation range. It can be divided into a low volatility range (the fluctuation range is within ±1 MW), a medium volatility range (the fluctuation range is 1 MW - 3 MW), and a high volatility range (the fluctuation range is greater than 3 MW). The membership function types corresponding to each fuzzy sub-range are a left trapezoidal function, a triangle function, and a right trapezoidal function respectively. Assuming that the current fluctuation of the load power is 2 MW, it is calculated that the membership weight in the medium volatility range is 0.6. According to the membership weight and the preset rule template (such as when in the medium volatility range and the frequency is stable, the access of the standby power supply can be adjusted according to the load prediction; if the frequency has an offset and the load volatility is in the medium volatility range, the frequency is adjusted first and then the load adjustment is considered), a fuzzy rule subset corresponding to load volatility is generated.

[0048] For the frequency offset, it is assumed that it is divided into a large negative offset range (below 49.8 Hz), a small negative offset range (49.8 Hz - 49.9 Hz), a normal range (49.9 Hz - 50.1 Hz), a small positive offset range (50.1 Hz - 50.2 Hz), and a large positive offset range (above 50.2 Hz). Each corresponds to a different type of membership function. If the collected frequency is 50.05 Hz, the membership weight in the normal range is calculated to be 0.9. According to the membership weight and the preset rule template (for example, when in the normal range and the voltage and load are stable, no adjustment is required; if in the small positive or small negative offset range and the load has a changing trend, adjust the output power of the generator), a fuzzy rule subset corresponding to the frequency offset is generated.

[0049] Finally, logically fuse the fuzzy rule subsets of each key parameter type such as voltage deviation, load volatility, and frequency offset. For example, when the voltage is in the small positive deviation range, the load volatility is in the medium volatility range, and the frequency is in the normal range, the fused fuzzy rule may be to observe for a period of time first. If the load volatility has an increasing trend, adjust the reactive power compensation device. If the frequency has a small offset, slightly adjust the output power of the generator, thereby generating the entire fuzzy rule set. This set covers the comprehensive adjustment logic under various grid operation parameter states and provides a basis for subsequent control strategy optimization.

[0050] In a possible implementation manner, step S130 includes: Step S131, obtaining the historical control strategy set of the target centralized control area, where the historical control strategy set includes multiple historical control instructions and the execution effect evaluation index corresponding to each historical control instruction.

[0051] In the above medium-sized power supply area, the historical control strategy set includes numerous historical control instructions and the execution effect evaluation index corresponding to each instruction. For example, there is a historical control instruction for the situation of too high transformer oil temperature. When the transformer oil temperature exceeds 80 °C, start the cooling device. The execution effect evaluation index corresponding to this instruction is that the probability of the oil temperature dropping below 75 °C within 10 minutes is 90%, which reflects the effectiveness of this control instruction in historical execution. There is also a historical control instruction regarding the overload protection of the transmission line. When the current of the transmission line exceeds 600 A, adjust the setting value of the line protection device. Its execution effect evaluation index is that the probability of the overload risk of the line dropping to the safe range within 15 minutes after adjustment is 85%.

[0052] Step S132, associatively matching the fuzzy rule set with the historical control strategy set to determine the mapping relationship between each fuzzy rule in the fuzzy rule set and the historical control instruction.

[0053] Taking the transformer oil temperature as an example, there are fuzzy rules in the fuzzy rule set regarding different intervals of the transformer oil temperature (such as 75°C - 80°C, 80°C - 85°C, etc.). These fuzzy rules have a mapping relationship with the historical control instructions for starting the cooling device when the transformer oil temperature is too high. For the current of the transmission line, the rules in the fuzzy rules regarding different fluctuation ranges of the transmission line current (such as 550A - 600A, 600A - 650A, etc.) have a mapping relationship with the historical control instructions for the overload protection of the transmission line. This mapping relationship is established based on the correlation between the grid operation parameters and the control instructions.

[0054] Step S133, according to the mapping relationship, screen candidate control instructions that meet the preset optimization conditions from the historical control strategy set. The preset optimization conditions include that the execution effect evaluation index is higher than the preset threshold and the triggering condition matches the current grid operation parameters.

[0055] Suppose the preset threshold is set to more than 80% of the execution effect evaluation index. For the historical control instruction for starting the cooling device when the transformer oil temperature is too high, its execution effect evaluation index is 90%, which is higher than the preset threshold. And if the current transformer oil temperature is close to 80°C, its triggering condition matches the current grid operation parameters, so this instruction becomes a candidate control instruction. For the historical control instruction for the overload protection of the transmission line, its execution effect evaluation index is 85%, which is higher than the preset threshold. If the current transmission line current is close to 600A, its triggering condition matches the current grid operation parameters, and it also becomes a candidate control instruction.

[0056] Step S134, perform a priority ranking on the candidate control instructions to generate an initial optimized control strategy. The priority ranking is based on the real-time matching degree of the execution effect evaluation index and the triggering condition of the candidate control instructions.

[0057] For example, the execution effect evaluation index of the instruction for starting the cooling device when the transformer oil temperature is too high is 90%. If the current transformer oil temperature is 79°C, the real-time matching degree of its triggering condition is relatively high. The execution effect evaluation index of the instruction for the overload protection of the transmission line is 85%. If the current transmission line current is 590A, the real-time matching degree of its triggering condition is relatively low. Based on this, the priority of the instruction for starting the cooling device when the transformer oil temperature is too high is higher than that of the instruction for the overload protection of the transmission line, thereby generating an initial optimized control strategy.

[0058] Step S135, according to the device response delay parameters of the target centralized control area, correct the execution timing of the control instructions in the initial optimized control strategy to generate the optimized control strategy.

[0059] For example, there is a 2-minute delay in the start-up of the cooling device of the transformer. In the initial optimized control strategy, if the original setting is to start the cooling device immediately when the oil temperature of the transformer reaches 80°C, considering the 2-minute delay, the corrected execution timing is to send an instruction to start the cooling device when the oil temperature reaches 78°C. For the overload protection instruction of the transmission line, if there is a 1-minute delay in the adjustment of the setting value of the line protection device, and the original setting is to adjust immediately when the current reaches 600A, the corrected setting may be to send an adjustment instruction when the current reaches 599A. In this way, through the corrected execution timing of the control instruction, the finally generated optimized control strategy can better adapt to the characteristics of the grid operation equipment, improve the accuracy and timeliness of control, and ensure the stable operation of the grid.

[0060] In a possible implementation manner, step S140 includes: Step S141, parsing the trigger conditions in the optimized control strategy to determine the threshold range of grid operation parameters corresponding to each control instruction.

[0061] For example, in this medium-sized power supply area, for the transformer oil temperature control instruction, the trigger condition in the optimized control strategy is set to start the cooling device when the oil temperature of the transformer reaches 78°C, then the corresponding threshold range of grid operation parameters is 78°C and above. For the transmission line current control instruction, if it is set to adjust the setting value of the line protection device when the transmission line current reaches 599A, its threshold range of grid operation parameters is 599A and above.

[0062] Step S142, monitoring the grid operation parameters of the target centralized control area in real time, and activating the corresponding control instruction when it is detected that the grid operation parameters fall within the threshold range of the grid operation parameters.

[0063] For example, through monitoring devices such as sensors installed on the transformer and transmission line, continuously obtain the grid operation parameters. When it is monitored that the oil temperature of the transformer rises to 78°C, this value falls within the threshold range of the grid operation parameters corresponding to the transformer oil temperature control instruction, thereby activating the corresponding control instruction, that is, the instruction to start the transformer cooling device. Similarly, when it is monitored that the transmission line current reaches 599A, the instruction to adjust the setting value of the transmission line protection device is activated.

[0064] Step S143, encapsulating the activated control instruction into a device-executable instruction format and sending it to the target execution device through the centralized control communication network.

[0065] For example, for the control instruction of the transformer cooling device, it is encapsulated in the instruction format recognizable by the transformer cooling device. For example, the oil temperature control instruction is converted into a specific digital signal code and then sent to the transformer cooling device through the centralized control communication network. For the control instruction of the transmission line protection device, a similar operation is performed. The current adjustment instruction is converted into a format recognizable by the protection device and sent to the transmission line protection device through the communication network.

[0066] Step S144, receive the instruction response signal returned by the target execution device, and record the timestamp and execution status identifier of the instruction response signal.

[0067] For example, when the transformer cooling device receives the control instruction, it will return an instruction response signal. Suppose the timestamp when this response signal is received is 10:02:00 on August 1, 2023, and the execution status identifier is successfully started. For the transmission line protection device, when receiving the instruction response signal it returns, record the corresponding timestamp, such as 10:10:00 on August 1, 2023, and the execution status identifier is successfully adjusted.

[0068] Step S145, generate the execution feedback data based on the timestamp and the execution status identifier, where the execution feedback data includes the instruction execution delay time and the execution success rate.

[0069] For example, for the control instruction of the transformer cooling device, calculate the instruction execution delay time. The time interval from when the oil temperature reaches 78°C (assumed to be 10:00:00 on August 1, 2023) to when the response signal is received is 2 minutes, and the execution success rate is 100%. These data constitute the execution feedback data. For the control instruction of the transmission line protection device, the calculated instruction execution delay time is 1 minute (assuming the time when the current reaches 599A is 10:09:00 on August 1, 2023), and the execution success rate is 100%. It is also included in the execution feedback data.

[0070] In a possible implementation manner, step S150 includes: Step S151, extract the instruction execution delay time and the execution success rate in the execution feedback data, and calculate the comprehensive execution efficiency index of the optimal control strategy.

[0071] Among them, step S151 includes: Step S1511, obtain the standard response delay time of different device types in the target centralized control area, where the device types include transformers, circuit breakers, and reactive power compensation devices.

[0072] Step S1512: For each control instruction, calculate the delay deviation coefficient according to the difference between the instruction execution delay time and the standard response delay time corresponding to the device type.

[0073] Step S1513: Calculate the success rate compensation coefficient based on the difference between the execution success rate and the preset success rate threshold.

[0074] Step S1514: Generate the local efficiency index of the control instruction through a weighted summation formula according to the delay deviation coefficient and the success rate compensation coefficient.

[0075] Step S1515: Normalize the local efficiency indices of all control instructions to obtain the comprehensive execution efficiency index.

[0076] For example, for a transformer, its standard response delay time is assumed to be 2.5 minutes, for a circuit breaker it is assumed to be 0.5 minutes (although circuit breaker control instructions are not involved in this scenario, it is listed for a complete illustration of the calculation process), and for a reactive power compensation device it is assumed to be 1 minute (also not involved but listed for completeness).

[0077] Assume the instruction execution delay time is 2 minutes, the difference from the standard response delay time of 2.5 minutes is -0.5 minutes. The delay deviation coefficient is calculated to be -0.2 through a specific formula (assume the formula is: delay deviation coefficient = (instruction execution delay time - standard response delay time) / standard response delay time). The success rate compensation coefficient is calculated based on the difference between the execution success rate and the preset success rate threshold. Assume the preset success rate threshold is 90% and the execution success rate is 100%, the difference is 10%. The success rate compensation coefficient is calculated to be 0.1 through another formula (assume the formula is: success rate compensation coefficient = (execution success rate - preset success rate threshold) / 100). According to the delay deviation coefficient and the success rate compensation coefficient, the local efficiency index of this control instruction is generated through a weighted summation formula (assume the formula is: local efficiency index = 0.6 × success rate compensation coefficient + 0.4 × delay deviation coefficient), and the calculated value is 0.08.

[0078] For the control instruction of the transmission line protection device, the same steps are calculated. The instruction execution delay time is 1 minute, the standard response delay time is assumed to be 1 minute (if there is a more accurate standard), the difference is 0 minutes, and the delay deviation coefficient is 0. The execution success rate is 100%, the preset success rate threshold is 90%, and the success rate compensation coefficient is 0.1. The local efficiency index is calculated to be 0.06 through the weighted summation formula (assuming the weighted coefficients are the same).

[0079] Assuming there are only these two control instructions, the sum of the total local efficiency indicators is 0.08 + 0.06 = 0.14. The normalized local efficiency indicator of the transformer cooling device control instruction is 0.08 / 0.14 ≈ 0.57, and the normalized local efficiency indicator of the transmission line protection device control instruction is 0.06 / 0.14 ≈ 0.43. The comprehensive execution efficiency indicator is 0.57×100% + 0.43×100% = 100% (this is a simple example, and the actual calculation may be more complex).

[0080] Step S152, compare the comprehensive execution efficiency indicator with the preset feedback threshold. If the comprehensive execution efficiency indicator is lower than the preset feedback threshold, it is determined that the optimized control strategy does not meet the effectiveness standard.

[0081] Assume the preset feedback threshold is 90%. Since the calculated comprehensive execution efficiency indicator is 100%, which is higher than the preset feedback threshold, it is determined that the optimized control strategy meets the effectiveness standard and no adjustment is required.

[0082] Step S153, based on the control instruction with an execution success rate lower than the preset success rate threshold, trace back to the corresponding fuzzy rule, and adjust the membership degree weight or the fuzzy sub - interval division method in the fuzzy rule.

[0083] If there is a situation where the execution success rate is lower than the preset success rate threshold. For example, in another case, the execution success rate of the transmission line protection device is 80%. First, based on the control instruction with an execution success rate lower than the preset success rate threshold, trace back to the corresponding fuzzy rule. Determine the fuzzy rule identifier corresponding to the transmission line protection device control instruction, and extract the target fuzzy rule and the type of power grid operation parameters associated with the target fuzzy rule (i.e., transmission line current) from the fuzzy rule set. Obtain the current membership degree weight and fuzzy sub - interval division parameters of the target fuzzy rule. Based on the difference between the execution success rate and the preset success rate threshold (80% - 90% = - 10%), generate a membership degree weight adjustment amount. Assume that - 0.1 is obtained through a specific algorithm.

[0084] According to the membership degree weight adjustment amount, update the membership degree weight of the target fuzzy rule, or re - divide the boundary range of the fuzzy sub - interval. For example, if the original membership degree weight is 0.8, it is updated to 0.7; or if the original fuzzy sub - interval division is 550A - 600A, 600A - 650A, etc., after re - division, it becomes 540A - 590A, 590A - 640A, etc.

[0085] Step S154, based on the adjusted fuzzy rule, regenerate the updated optimized control strategy, and replace the original optimized control strategy with the updated optimized control strategy.

[0086] For example, re-match the candidate control instructions in the historical control strategy set for conflict detection. If there are conflicting control instructions, for example, two control instructions target the operation of the same transmission line during the same period but have logical contradictions, according to the execution effect evaluation index, retain the control instruction with the highest priority and eliminate the remaining conflicting instructions. Generate an updated set of candidate control instructions based on the elimination results, and re-perform priority sorting and timing correction. Package the corrected set of candidate control instructions as an updated optimized control strategy, and cover the storage location of the original optimized control strategy through the centralized control communication network.

[0087] Step S155: Continuously monitor the execution feedback data of the replaced optimized control strategy until the comprehensive execution efficiency index reaches the preset feedback threshold.

[0088] For example, within a preset monitoring period (such as 1 hour), collect multiple batches of execution feedback data. Calculate the moving window average of the comprehensive execution efficiency index in the multiple batches of execution feedback data to obtain a dynamic efficiency evaluation value. If the dynamic efficiency evaluation value is higher than the preset feedback threshold in N consecutive (assuming N = 3) monitoring periods, it is determined that the updated optimized control strategy meets the stability standard. If not, trigger the fuzzy rule adjustment and optimized control strategy update process again based on the latest execution feedback data until the stability standard is met. For example, in subsequent monitoring, if the dynamic efficiency evaluation value is 85% in the first monitoring period and does not reach the preset feedback threshold of 90%, continue to adjust the fuzzy rules and the optimized control strategy until the dynamic efficiency evaluation value is higher than 90% in 3 consecutive monitoring periods. Mark the current optimized control strategy as the final version and generate a unique strategy identifier for it (including the target centralized control area code, generation timestamp, and version serial number). Associate the final version of the optimized control strategy with the strategy identifier, encrypt it and transmit it to the specified storage partition of the centralized control strategy database, establish an index relationship between the strategy identifier and the historical control strategy set, delete the copy of the optimized control strategy temporarily stored in the centralized control communication network, release the relevant computing resources, and send a strategy update completion notification (including the strategy identifier and the effective time range) to the monitoring terminal in the target centralized control area.

[0089] In a possible implementation manner, step S153 includes: Step S1531: Determine the fuzzy rule identifier corresponding to the control instruction whose execution success rate is lower than the preset success rate threshold.

[0090] Suppose that during the grid control of a medium-sized power supply area, it is found that the execution success rate of the control command for adjusting the transmission line protection device is lower than the preset success rate threshold. In the entire grid control strategy system, each control command is associated with a specific fuzzy rule and has a unique identifier to distinguish different fuzzy rules. For the control command for adjusting the transmission line protection device, the corresponding fuzzy rule identifier is found by querying the mapping relationship table between the control command and the fuzzy rule.

[0091] Step S1532: Extract the target fuzzy rule and the types of grid operation parameters associated with the target fuzzy rule from the fuzzy rule set according to the fuzzy rule identifier.

[0092] For example, accurately locate the target fuzzy rule from the huge fuzzy rule set according to the identifier. The target fuzzy rule is associated with the type of grid operation parameter of the transmission line current, and it describes the adjustment logic corresponding to the transmission line current in different intervals. For example, it may include that when the transmission line current is in the interval of 550A - 600A, a certain protection device adjustment method is adopted; when the current is in the interval of 600A - 650A, another adjustment method is adopted, etc.

[0093] Step S1533: Obtain the current membership degree weight and the fuzzy sub-interval division parameters of the target fuzzy rule, and generate a membership degree weight adjustment amount based on the difference between the execution success rate and the preset success rate threshold.

[0094] Suppose the preset success rate threshold for adjusting the transmission line protection device is 90%, and the actual execution success rate is 80%, and the difference between the two is -10%. In the target fuzzy rule, the current membership degree weight corresponding to the transmission line current interval of 550A - 600A is 0.8, and the fuzzy sub-interval division parameters are the boundary values 550A and 600A of this interval. Through a specific algorithm (this algorithm is preset based on the historical data of grid operation, equipment characteristics, and control strategy requirements), a membership degree weight adjustment amount is generated according to the difference of -10%, which is assumed to be -0.1.

[0095] Step S1534: Update the membership degree weight of the target fuzzy rule or re-divide the boundary range of the fuzzy sub-interval according to the membership degree weight adjustment amount.

[0096] In this example, according to the calculation results, the membership weight in the range of 550A - 600A is updated from 0.8 to 0.7. Or, if the boundary range of the fuzzy sub - intervals is selected to be re - divided, the original range of 550A - 600A may be adjusted to 540A - 590A to meet the new control requirements. The above - mentioned adjustments are made based on the unsatisfactory actual control effect of the transmission line current, with the aim of making the fuzzy rules more in line with the actual situation of power grid operation.

[0097] Step S1535: Re - add the updated target fuzzy rule to the fuzzy rule set and delete the original target fuzzy rule.

[0098] For example, re - integrate the adjusted target fuzzy rule into the fuzzy rule set to ensure the integrity and accuracy of the set. At the same time, delete the original target fuzzy rule to avoid confusion during the subsequent generation of the optimized control strategy. In this way, the fuzzy rule set is updated, providing a more reasonable basis for regenerating the optimized control strategy.

[0099] In a possible implementation manner, step S154 includes: Step S1541: According to the adjusted fuzzy rule set, re - match the candidate control instructions in the historical control strategy set.

[0100] In the power grid control of a medium - sized power supply area, the historical control strategy set contains numerous control instructions for different power grid operation conditions, and each instruction has a corresponding execution effect evaluation index. The adjusted fuzzy rule set changes the judgment logic and weight distribution of power grid operation parameters. Therefore, it is necessary to re - match the candidate control instructions in the historical control strategy set. For example, for the transformer oil temperature control instruction and the transmission line protection device adjustment instruction, etc., re - evaluate their matching degree with the power grid operation parameters according to the new fuzzy rules, and determine which instructions meet the new optimization conditions to become candidate control instructions.

[0101] Step S1542: Perform conflict detection on the re - matched candidate control instructions. The conflict detection includes detecting the logical contradiction of multiple control instructions received by the same device at the same time period.

[0102] In power grid operation, the same device (such as a transmission line) may be affected by multiple control instructions at the same time period. For example, a candidate control instruction may be to increase reactive power compensation to stabilize the voltage at a specific transmission line current, while another candidate control instruction may be to adjust the setting value of the line protection device under the same current condition. However, the execution of these two instructions may interfere with each other and there is a logical contradiction. By analyzing in detail the operation object, operation condition, and operation result of each candidate control instruction, this logical contradiction situation can be detected.

[0103] Step S1543, if a conflict control instruction is detected, then according to the execution effect evaluation index of the conflict control instruction, retain the control instruction with the highest priority and eliminate the remaining conflicting instructions.

[0104] Suppose that for a transmission line in a certain current range, there are two conflicting control instructions. One is to adjust the setting value of the line protection device, and its execution effect evaluation index is that the probability of reducing the line overload risk to the safe range within 30 minutes is 85%. The other is to change the input amount of the reactive power compensation device, and its execution effect evaluation index is that the probability of controlling the line voltage fluctuation within ±5% within 30 minutes is 80%. Since the execution effect evaluation index of adjusting the setting value of the line protection device is higher, this instruction is retained and the instruction to change the input amount of the reactive power compensation device is eliminated.

[0105] Step S1544, generate an updated candidate control instruction set according to the elimination result, and re - perform priority sorting and timing correction.

[0106] After conflict detection and instruction elimination, an updated candidate control instruction set is obtained. Then, re - perform priority sorting on these candidate control instructions. The sorting basis is still the real - time matching degree between the execution effect evaluation index of the candidate control instruction and the triggering condition. For example, for a transformer oil temperature control instruction, if its execution effect evaluation index is that the probability of reducing the oil temperature to the safe range within 10 minutes is 90%, and the current transformer oil temperature is close to the triggering condition value, then its priority is relatively high. At the same time, considering the equipment response delay parameter, correct the execution timing of the control instruction. For example, the transformer cooling device has a 2 - minute start - up delay. The original setting is to start when the oil temperature reaches 80°C. After correction, the start - up instruction may be sent when the oil temperature reaches 78°C.

[0107] Step S1545, encapsulate the corrected candidate control instruction set as the updated optimized control strategy, and cover the storage location of the original optimized control strategy through the centralized control communication network.

[0108] For example, encapsulate the candidate control instruction set processed through the above steps in a specific format to form an updated optimized control strategy that can be recognized and executed by the grid execution equipment. Then, through the centralized control communication network, send this new optimized control strategy to the corresponding storage location to cover the original optimized control strategy, so as to adopt the new optimized strategy in the subsequent grid operation control to improve the efficiency and stability of the grid operation.

[0109] In a possible implementation manner, step S155 includes: Step S1551, within a preset monitoring period, collect multi - batch execution feedback data of the replaced optimized control strategy.

[0110] For example, in the power grid control of a medium-sized power supply area, the preset monitoring period is set to 1 hour. Within this 1-hour period, the execution feedback data of the optimized control strategy after replacement is collected every 10 minutes. For example, for the transformer oil temperature control instruction, the instruction execution delay time and execution success rate are recorded each time it is collected. If within a certain 10-minute period, after the transformer oil temperature reaches the trigger condition (such as 78 °C), the delay time for the cooling device to start is 2 minutes and the execution success rate is 100%, this constitutes a set of execution feedback data. Similarly, for the transmission line protection device adjustment instruction, when the transmission line current reaches the trigger condition (such as 599 A), the instruction execution delay time (assumed to be 1 minute) and execution success rate (assumed to be 100%) for the adjustment of the protection device setting value are recorded as execution feedback data. By collecting these data multiple times within the preset 1-hour monitoring period, multiple batches of execution feedback data are obtained.

[0111] Step S1552, calculate the moving window average value of the comprehensive execution efficiency indicators in multiple batches of execution feedback data to obtain a dynamic efficiency evaluation value. If the dynamic efficiency evaluation value is higher than the preset feedback threshold in N consecutive monitoring periods, it is determined that the updated optimized control strategy meets the stability standard.

[0112] Suppose the moving window size is 3 batches of data. For every 3 consecutive batches of execution feedback data, calculate the local efficiency indicators of each control instruction (such as the transformer oil temperature control instruction and the transmission line protection device adjustment instruction), and then calculate the comprehensive execution efficiency indicator of these 3 batches of data according to the previous calculation method (such as calculating the delay deviation coefficient based on the difference between the instruction execution delay time and the standard response delay time of the corresponding equipment type, calculating the success rate compensation coefficient based on the difference between the execution success rate and the preset success rate threshold, then generating the local efficiency indicator of the control instruction through a weighted summation formula, and finally normalizing the local efficiency indicators of all control instructions to obtain the comprehensive execution efficiency indicator). Take the average value of these comprehensive execution efficiency indicators as the dynamic efficiency evaluation value. If the dynamic efficiency evaluation value is higher than the preset feedback threshold in N consecutive (assume N = 3) monitoring periods, it is determined that the updated optimized control strategy meets the stability standard. For example, the preset feedback threshold is set to 90%. If in 3 consecutive 1-hour monitoring periods, the calculated dynamic efficiency evaluation values are 92%, 93%, and 95% respectively, all higher than 90%, it is determined that the updated optimized control strategy meets the stability standard.

[0113] Step S1553, if the dynamic efficiency evaluation value does not reach the preset feedback threshold, trigger the fuzzy rule adjustment and optimization control strategy update process again according to the latest execution feedback data. When the stability standard is reached, mark the current optimized control strategy as the final version, and generate a unique strategy identifier for the optimized control strategy of the final version. The unique strategy identifier includes the target centralized control area code, generation timestamp, and version serial number.

[0114] Suppose that in a certain monitoring period, the dynamic efficiency evaluation value is 85%, which is lower than the preset feedback threshold of 90%. At this time, analyze according to the latest collected execution feedback data. For example, it is found that the execution success rate of the transmission line protection device adjustment instruction has decreased in the data of the recent several batches. Then, adjust the fuzzy rule corresponding to this control instruction. According to the method described above, determine the fuzzy rule identifier corresponding to the control instruction whose execution success rate is lower than the preset success rate threshold, extract the target fuzzy rule and the associated power grid operation parameter type (transmission line current) from the fuzzy rule set, obtain the current membership degree weight and fuzzy sub - interval division parameters of the target fuzzy rule, generate a membership degree weight adjustment amount based on the difference between the execution success rate and the preset success rate threshold, and then update the membership degree weight or re - divide the boundary range of the fuzzy sub - interval according to the adjustment amount. After that, regenerate the updated optimized control strategy based on the adjusted fuzzy rule, such as re - matching the candidate control instructions in the historical control strategy set, performing conflict detection (such as detecting the logical contradiction of multiple control instructions received by the same device at the same time period. If there is a conflict, retain the control instruction with the highest priority according to the execution effect evaluation index and eliminate the remaining conflicting instructions), generating an updated candidate control instruction set according to the elimination result, re - performing priority sorting and timing correction, and finally encapsulating the corrected candidate control instruction set as a new optimized control strategy and covering the original optimized control strategy through the centralized control communication network.

[0115] When the stability standard is reached after multiple adjustments, mark the current optimized control strategy as the final version, and generate a unique strategy identifier for the optimized control strategy of the final version. The unique strategy identifier includes the target centralized control area code (for example, the specific code for a medium - sized power supply area is "MG - 001"), the generation timestamp (such as 12:00:00 on August 1, 2023, indicating the time when the optimized control strategy is finally determined), and the version serial number (assumed to be "V3.0", indicating that this is the third version after multiple adjustments).

[0116] Step S1554, associate the optimized control strategy of the final version with the strategy identifier, encrypt it and transmit it to the specified storage partition of the centralized control strategy database, and establish an index relationship between the strategy identifier and the historical control strategy set in the centralized control strategy database for version comparison during subsequent strategy retrieval.

[0117] For example, encrypt the optimized control strategy and the strategy identifier through a specific encryption algorithm (such as the AES encryption algorithm) to ensure data security. Transmit the encrypted data to the storage partition specifically set for the optimized control strategy of the medium-sized power supply area in the centralized control strategy database. Establish an index relationship between the strategy identifier and the set of historical control strategies in the centralized control strategy database for version comparison during subsequent strategy retrieval. In this way, when it is necessary to query the historical control strategy or compare different versions of the optimized control strategy, relevant information can be quickly and accurately obtained according to the index relationship.

[0118] Step S1555: Delete the copy of the optimized control strategy temporarily stored in the centralized control communication network and release the relevant computing resources.

[0119] During the previous update process of the optimized control strategy, a copy of the optimized control strategy may be stored in the temporary storage area of the centralized control communication network to facilitate data transmission and processing. When the final version of the optimized control strategy has been successfully stored in the centralized control strategy database and the index relationship has been established, these temporarily stored copies can be deleted to release the computing resources (such as network bandwidth, storage space, etc.) they occupy, improving the resource utilization rate of the entire power grid control system.

[0120] Step S1556: Send a policy update completion notification to the monitoring terminal in the target centralized control area, where the policy update completion notification includes the strategy identifier and the effective time range.

[0121] For example, send a notification to the monitoring terminal in the medium-sized power supply area, and the content of the notification is "The update of the optimized control strategy is completed, the strategy identifier is MG - 001 - 20230801120000 - V3.0, and the effective time range is from 12:00:00 on August 1, 2023". After receiving the notification, relevant staff can understand the update situation of the optimized control strategy and can query the detailed strategy content in the centralized control strategy database according to the strategy identifier for monitoring and managing the operation of the power grid.

[0122] In a possible implementation manner, after the step S150, the method further includes: Step S160: Perform an AI model training operation based on the final version of the optimized control strategy and the execution feedback data. The specific steps include: Step S161: Generate a training data set corresponding to the optimized control strategy of the final version. The training data set includes the historical sequence of grid operation parameters in the target centralized control area, the execution sequence of control instructions of the optimized control strategy, and the comprehensive execution efficiency index sequence of the execution feedback data, where the historical sequence of grid operation parameters is aligned with the execution sequence of control instructions by time stamp.

[0123] In this embodiment, in a medium-sized power supply area, the historical sequence of grid operation parameters in the target centralized control area includes the variation of relevant parameters of various devices over time. For example, for a transformer, the historical sequence records voltage values, oil temperature values, current values, etc. at different time points; for a transmission line, parameters such as current values and power factors are recorded. The time span of the above historical sequence of grid operation parameters may be several weeks or months in the past, long enough to reflect the grid operation state under different working conditions.

[0124] The execution sequence of control instructions of the optimized control strategy records the execution of each control instruction at the corresponding time point. For example, for the control instruction to start the cooling device when the oil temperature of the transformer reaches a specific value, record the execution time point; for the control instruction to adjust the setting value of the protection device when the current of the transmission line reaches the set value, also record the execution time. Moreover, the historical sequence of grid operation parameters is aligned with the execution sequence of control instructions by time stamp, which means that the sampling time of each grid operation parameter can be accurately corresponded to the execution time of the corresponding control instruction.

[0125] The comprehensive execution efficiency index sequence of the execution feedback data is the sequence of efficiency evaluation indexes over time after the execution of each control instruction calculated previously. For example, the comprehensive execution efficiency index after the execution of the control instruction for transformer oil temperature control at a certain moment is 90%, and the comprehensive execution efficiency index after the execution of the adjustment instruction for the transmission line protection device is 85%, etc. These indexes are arranged in chronological order to form the comprehensive execution efficiency index sequence. By combining these three parts, a training data set corresponding to the optimized control strategy of the final version is generated.

[0126] Step S162: Based on the training data set, extract the key parameter fluctuation patterns in the historical sequence of grid operation parameters, and perform spatio-temporal correlation matching between the key parameter fluctuation patterns and the instruction trigger conditions in the execution sequence of control instructions to generate a parameter-instruction mapping feature vector. The parameter-instruction mapping feature vector includes the trigger moment of each control instruction, the fluctuation amplitude and duration of the associated grid operation parameters.

[0127] For example, for a transformer, the key parameter fluctuation pattern may be the rising trend of the oil temperature over a period of time or the change in the voltage fluctuation amplitude during a specific period. Taking the oil temperature as an example, if the oil temperature gradually rises from 70°C to 80°C within a certain period of time, this is a key parameter fluctuation pattern. When the oil temperature reaches 78°C, a control instruction to start the cooling device is triggered, and this fluctuation pattern is associated and matched with the instruction trigger condition. Specifically, the parameter-instruction mapping feature vector includes the trigger moment of each control instruction (such as starting the transformer cooling device at 10:00:00 on August 1, 2023), the fluctuation amplitude of the associated power grid operating parameters (the oil temperature rises from 70°C to 78°C, a total of 8°C fluctuation amplitude), and the duration (it takes a total of 2 hours from the start of the rise to the trigger of the control instruction). The same operation is also performed on the current parameters of the transmission line and its corresponding protection device adjustment instructions, thereby constructing a complete parameter-instruction mapping feature vector.

[0128] Step S163: Bind labels to the parameter-instruction mapping feature vector and the comprehensive execution efficiency index sequence to generate a supervised model training sample set, where the parameter-instruction mapping feature vector serves as the input feature and the comprehensive execution efficiency index serves as the prediction label.

[0129] In this process, the parameter-instruction mapping feature vector is used as the input feature, and the comprehensive execution efficiency index is used as the prediction label. For example, an input feature is the parameter-instruction mapping feature vector related to the transformer oil temperature triggering the cooling device under a specific fluctuation pattern, and its corresponding prediction label is the comprehensive execution efficiency index of 90% at that time. In this way, for each set of input features and the corresponding prediction labels, a sample in the supervised model training sample set is formed, and many such samples combined together form a complete supervised model training sample set.

[0130] Step S164: Train a prediction model based on the supervised model training sample set.

[0131] In a possible implementation manner, step S164 includes: Step S1641: Configure the initial training parameters of the prediction model. The prediction model includes a combined structure of a multi-layer temporal attention network and a fully connected decision layer. The temporal attention network is used to capture the temporal dependence relationship in the parameter-instruction mapping feature vector, and the fully connected decision layer is used to output the association matrix between the trigger condition of the predicted control instruction and the predicted execution efficiency value.

[0132] For example, a temporal attention network is used to capture the temporal dependencies in the parameter-instruction mapping feature vector. For instance, when dealing with multiple fluctuations in the transformer oil temperature and the execution of corresponding control instructions, the temporal attention network can focus on the sequence of oil temperature fluctuations and the time interval relationship with the execution of control instructions. The fully connected decision layer is used to output the correlation matrix between the trigger conditions of the predicted control instructions and the estimated execution efficiency values, such as predicting the likelihood of control instruction triggering under specific power grid operating parameters and the estimated execution efficiency values that may be achieved after execution.

[0133] Step S1642: Divide the supervised model training sample set into a training subset and a validation subset according to a preset ratio, and perform multiple rounds of iterative training on the prediction model using the training subset. In each round of iteration, update the weight parameters of the temporal attention network and the bias coefficients of the fully connected decision layer using the backpropagation algorithm.

[0134] For example, divide according to a ratio of 80%:20%. 80% of the samples are used as the training subset, and 20% of the samples are used as the validation subset. Perform multiple rounds of iterative training on the prediction model using the training subset. In each round of iteration, update the weight parameters of the temporal attention network and the bias coefficients of the fully connected decision layer using the backpropagation algorithm. In each round of iteration, the model adjusts its own parameters according to the samples in the training subset to make the prediction results closer to the true comprehensive execution efficiency index.

[0135] Step S1643: After each round of iterative training is completed, use the validation subset to calculate the current loss function value and the prediction accuracy index of the prediction model. If the current loss function value does not converge or the prediction accuracy index is lower than the preset model threshold, adjust the number of layers of the temporal attention network or the activation function type of the fully connected decision layer, and restart the iterative training.

[0136] The loss function value reflects the degree of difference between the prediction result of the prediction model and the true label. The prediction accuracy index represents the proportion of correctly predicted samples. If the current loss function value does not converge (i.e., the loss function value does not stabilize at a small value as the number of iterations increases) or the prediction accuracy index is lower than the preset model threshold (for example, the preset model threshold is set to 85%, and the actually calculated prediction accuracy index is 80%), then adjust the number of layers of the temporal attention network or the activation function type of the fully connected decision layer, and restart the iterative training process. For example, if the current temporal attention network has 3 layers, it may be increased to 4 layers; if the activation function type of the fully connected decision layer is the Sigmoid function, it may be adjusted to the ReLU function, and then the iterative training process is restarted.

[0137] Step S1644, when the prediction model reaches the convergence state and the prediction accuracy rate index is higher than the preset model threshold, freeze the network parameters of the prediction model, and deploy the prediction model to the real-time inference interface of the centralized control strategy optimization engine, where the real-time inference interface is used to receive the current power grid operation parameter sequence and output a set of prediction control instructions.

[0138] When the prediction model reaches the convergence state (the loss function value stabilizes at a relatively small value) and the prediction accuracy rate index is higher than the preset model threshold (such as reaching 90%), freeze the network parameters of the prediction model, and deploy the prediction model to the real-time inference interface of the centralized control strategy optimization engine. The real-time inference interface is used to receive the current power grid operation parameter sequence and output a set of prediction control instructions. For example, when the current oil temperature, voltage, current and other power grid operation parameter sequences of the transformer are received in real time, the prediction model can output a set of prediction control instructions such as whether to start the cooling device and whether to adjust the voltage according to the rules learned before.

[0139] Step S1645, based on the set of prediction control instructions, pre-screen the fuzzy sub-interval partitioning method and membership weight in the fuzzy rule set, eliminate the fuzzy rules associated with the inefficient instructions in the set of prediction control instructions, and generate a refined fuzzy rule candidate pool.

[0140] Eliminate the fuzzy rules in the fuzzy rule set that are associated with the inefficient instructions (such as instructions with low execution efficiency or rarely triggered) in the set of prediction control instructions, and generate a refined fuzzy rule candidate pool. For example, if a certain fuzzy rule is always associated with the transformer oil temperature control instruction with low execution efficiency, then this fuzzy rule will be eliminated.

[0141] Step S1646, perform cross-cycle correlation analysis on the refined fuzzy rule candidate pool and the historical control strategy set, identify the target fuzzy rules that have a strong correlation with the efficient control instructions in multiple historical cycles, and add dynamic priority coefficients to the target fuzzy rules.

[0142] In multiple historical cycles (such as different periods in the past few months), identify the target fuzzy rules that have a strong correlation with the efficient control instructions. For example, if a certain fuzzy rule is closely related to the adjustment instruction of the transmission line protection device with high efficiency in multiple historical cycles, then this fuzzy rule is the target fuzzy rule. Add dynamic priority coefficients to these target fuzzy rules, for example, increase their priority coefficient from 1 to 1.5.

[0143] Step S1647: According to the dynamic priority coefficient, reorder the triggering order of the rules in the fuzzy rule set, and integrate the reordered fuzzy rule set and the prediction model into the generation process of the next round of optimized control strategy, so that the set of predicted control instructions output by the prediction model preferentially participates in the matching and screening of candidate control instructions.

[0144] For example, when generating the next round of optimized control strategy, make the set of predicted control instructions output by the prediction model preferentially participate in the matching and screening of candidate control instructions. In this way, through the AI model training operation, continuously optimize the fuzzy rule set and the generation process of the optimized control strategy, and improve the accuracy and efficiency of power grid control.

[0145] Figure 2 FIG. shows a schematic diagram of exemplary hardware and software components of a fuzzy logic-based centralized control automation strategy optimization system 100 that can implement the idea of the present application. For example, the processor 120 can be used on the fuzzy logic-based centralized control automation strategy optimization system 100 and is used to execute the functions in the present application.

[0146] The fuzzy logic-based centralized control automation strategy optimization system 100 can be a general-purpose server or a special-purpose server, both of which can be used to implement the fuzzy logic-based centralized control automation strategy optimization method of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0147] For example, the fuzzy logic-based centralized control automation strategy optimization system 100 can include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as disks, ROM, or RAM, or any combination thereof. Exemplarily, the fuzzy logic-based centralized control automation strategy optimization system 100 can also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The fuzzy logic-based centralized control automation strategy optimization system 100 also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0148] For ease of explanation, only one processor is described in the centralized control automation policy optimization system 100 based on fuzzy logic. However, it should be noted that the centralized control automation policy optimization system 100 in the present application may also include multiple processors. Therefore, the steps executed by one processor described in the present application may also be jointly executed or separately executed by multiple processors. For example, if the processor of the centralized control automation policy optimization system 100 based on fuzzy logic executes step A and step B, it should be understood that step A and step B may also be jointly executed by two different processors or separately executed in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0149] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned centralized control automation policy optimization method based on fuzzy logic is implemented.

[0150] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the present invention, in the previous description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing or description thereof.

Claims

1. A method for optimizing the centralized control automation strategy based on fuzzy logic, characterized in that The method includes: Collecting real-time operation status data of power grid operation equipment in a target centralized control area, where the real-time operation status data includes at least one power grid operation parameter and a parameter fluctuation range corresponding to the power grid operation parameter; Performing fuzzy membership processing on the real-time operation status data according to the parameter fluctuation range to generate a fuzzy rule set matching the power grid operation parameter, where the fuzzy rule set includes multiple fuzzy rules, and each fuzzy rule corresponds to a dynamic adjustment logic of a power grid operation parameter; Based on the fuzzy rule set, performing fuzzy logic optimization on the current control strategy of the target centralized control area to generate an optimized control strategy, where the optimized control strategy includes multiple control instructions and a trigger condition for each control instruction; According to the trigger condition, sending the optimized control strategy to an execution device in the target centralized control area and monitoring execution feedback data of the execution device for the control instruction; Based on a matching result between the execution feedback data and a preset feedback threshold, verifying the effectiveness of the optimized control strategy and dynamically adjusting the optimized control strategy according to the verification result.

2. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 1, characterized in that, The performing fuzzy membership processing on the real-time operation status data according to the parameter fluctuation range to generate a fuzzy rule set matching the power grid operation parameter includes: Extracting key parameter types in the power grid operation parameter, where the key parameter types include voltage deviation, load volatility, and frequency offset; For each key parameter type, determining multiple fuzzy sub-intervals corresponding to the parameter fluctuation range, and each fuzzy sub-interval corresponds to a membership function type; According to the membership function type, assigning a corresponding membership weight to each fuzzy sub-interval, where the membership weight is used to quantify the distribution probability of the real-time operation status data in the fuzzy sub-interval; Based on the membership weight and a preset rule template, generating a fuzzy rule subset corresponding to each key parameter type, where the preset rule template includes a control logic mapping relationship under different power grid operation scenarios; Logically fusing the fuzzy rule subsets of each key parameter type to generate the fuzzy rule set.

3. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 1, wherein The performing fuzzy logic optimization on the current control strategy of the target centralized control area based on the fuzzy rule set to generate an optimized control strategy includes: Obtaining a historical control strategy set of the target centralized control area, where the historical control strategy set includes multiple historical control instructions and an execution effect evaluation index corresponding to each historical control instruction; Associating and matching the fuzzy rule set with the historical control strategy set to determine a mapping relationship between each fuzzy rule in the fuzzy rule set and the historical control instruction; According to the mapping relationship, screening candidate control instructions that meet a preset optimization condition from the historical control strategy set, where the preset optimization condition includes that the execution effect evaluation index is higher than a preset threshold and the trigger condition matches the current power grid operation parameter; Perform priority sorting on the candidate control instructions to generate an initial optimized control strategy, where the priority sorting is based on the real-time matching degree between the execution effect evaluation index of the candidate control instructions and the triggering conditions; According to the device response delay parameter of the target centralized control area, correct the execution timing of the control instructions in the initial optimized control strategy to generate the optimized control strategy.

4. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 3, characterized in that According to the triggering conditions, send the optimized control strategy to the execution devices in the target centralized control area, and monitor the execution feedback data of the execution devices for the control instructions, including: Analyze the triggering conditions in the optimized control strategy to determine the threshold range of the grid operation parameters corresponding to each control instruction; Real-time monitor the grid operation parameters of the target centralized control area, and when it is detected that the grid operation parameters fall within the threshold range of the grid operation parameters, activate the corresponding control instruction; Encapsulate the activated control instruction into a device-executable instruction format, and send it to the target execution device through the centralized control communication network; Receive the instruction response signal returned by the target execution device, and record the timestamp and execution status identifier of the instruction response signal; Based on the timestamp and the execution status identifier, generate the execution feedback data, where the execution feedback data includes the instruction execution delay time and the execution success rate.

5. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 4, characterized in that Based on the matching result between the execution feedback data and the preset feedback threshold, verify the effectiveness of the optimized control strategy, and dynamically adjust the optimized control strategy according to the verification result, including: Extract the instruction execution delay time and the execution success rate from the execution feedback data, and calculate the comprehensive execution efficiency index of the optimized control strategy; Compare the comprehensive execution efficiency index with the preset feedback threshold. If the comprehensive execution efficiency index is lower than the preset feedback threshold, it is determined that the optimized control strategy does not meet the effectiveness standard; According to the control instructions with the execution success rate lower than the preset success rate threshold, trace back to the corresponding fuzzy rules, and adjust the membership weight or the fuzzy sub-interval division method in the fuzzy rules; Based on the adjusted fuzzy rules, regenerate the updated optimized control strategy, and replace the original optimized control strategy with the updated optimized control strategy; Continuously monitor the execution feedback data of the replaced optimized control strategy until the comprehensive execution efficiency index reaches the preset feedback threshold; Among them, the extracting the instruction execution delay time and the execution success rate from the execution feedback data, and calculating the comprehensive execution efficiency index of the optimized control strategy, includes: Obtain the standard response delay time of different device types in the target centralized control area, where the device types include transformers, circuit breakers, and reactive power compensation devices; For each control instruction, calculate the delay deviation coefficient according to the difference between the instruction execution delay time and the standard response delay time of the corresponding device type; Based on the difference between the execution success rate and the preset success rate threshold, calculate the success rate compensation coefficient; According to the delay deviation coefficient and the success rate compensation coefficient, generate the local efficiency index of the control instruction through the weighted summation formula; Normalize the local efficiency indicators of all control instructions to obtain the comprehensive execution efficiency indicator.

6. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 5, wherein, Based on the control instructions with an execution success rate lower than the preset success rate threshold, trace back to the corresponding fuzzy rules and adjust the membership weight or fuzzy sub-interval division method in the fuzzy rules, including: Determine the fuzzy rule identifier corresponding to the control instruction with an execution success rate lower than the preset success rate threshold; According to the fuzzy rule identifier, extract the target fuzzy rule and the types of grid operation parameters associated with the target fuzzy rule from the fuzzy rule set; Obtain the current membership weight and fuzzy sub-interval division parameters of the target fuzzy rule, and generate a membership weight adjustment amount based on the difference between the execution success rate and the preset success rate threshold; According to the membership weight adjustment amount, update the membership weight of the target fuzzy rule or re-divide the boundary range of the fuzzy sub-interval; Re-add the updated target fuzzy rule to the fuzzy rule set and delete the original target fuzzy rule.

7. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 5, characterized in that, Based on the adjusted fuzzy rules, regenerate the updated optimized control strategy and replace the original optimized control strategy with the updated optimized control strategy, including: According to the adjusted fuzzy rule set, re-match the candidate control instructions in the historical control strategy set; Perform conflict detection on the re-matched candidate control instructions. The conflict detection includes detecting the logical contradiction of multiple control instructions received by the same device at the same time; If conflict control instructions are detected, retain the control instruction with the highest priority and eliminate the remaining conflict instructions according to the execution effect evaluation index of the conflict control instructions; Generate an updated set of candidate control instructions according to the elimination result, and re-perform priority sorting and timing correction; Package the corrected set of candidate control instructions as the updated optimized control strategy and overwrite the storage location of the original optimized control strategy through the centralized control communication network.

8. The method for optimizing the centralized control automation strategy based on fuzzy logic according to claim 7, characterized in that, Continuously monitor the execution feedback data of the replaced optimized control strategy until the comprehensive execution efficiency indicator reaches the preset feedback threshold, including: During the preset monitoring period, collect multiple batches of execution feedback data of the replaced optimized control strategy; Calculate the moving window average value of the comprehensive execution efficiency indicator in multiple batches of execution feedback data to obtain a dynamic efficiency evaluation value. If the dynamic efficiency evaluation value is higher than the preset feedback threshold in N consecutive monitoring periods, it is determined that the updated optimized control strategy meets the stability standard; If the dynamic efficiency evaluation value does not reach the preset feedback threshold, trigger the fuzzy rule adjustment and optimized control strategy update process again according to the latest execution feedback data. When the stability standard is met, mark the current optimized control strategy as the final version, and generate a unique strategy identifier for the final version of the optimized control strategy. The unique strategy identifier includes the target centralized control area code, generation timestamp, and version serial number; Associate the optimized control strategy of the final version with the strategy identifier, encrypt it, and transmit it to the designated storage partition of the centralized control strategy database. Establish an index relationship between the strategy identifier and the historical control strategy set in the centralized control strategy database for version comparison during subsequent strategy retrieval; Delete the copy of the optimized control strategy temporarily stored in the centralized control communication network and release the relevant computing resources; Send a policy update completion notification to the monitoring terminals in the target centralized control area, where the policy update completion notification includes the strategy identifier and the effective time range.

9. The method for optimizing the centralized control automation strategy based on fuzzy logic according to any one of claims 1-8, characterized in that, After the step of sending the policy update completion notification to the monitoring terminals in the target centralized control area, the method further includes: Perform an AI model training operation based on the optimized control strategy of the final version and the execution feedback data. The specific steps include: Generate a training data set corresponding to the optimized control strategy of the final version. The training data set includes the historical sequence of power grid operation parameters in the target centralized control area, the execution sequence of control instructions of the optimized control strategy, and the comprehensive execution efficiency index sequence of the execution feedback data, where the historical sequence of power grid operation parameters and the execution sequence of control instructions are aligned by timestamp; Based on the training data set, extract the key parameter fluctuation patterns in the historical sequence of power grid operation parameters, and perform spatio-temporal correlation matching between the key parameter fluctuation patterns and the instruction trigger conditions in the execution sequence of control instructions to generate a parameter-instruction mapping feature vector. The parameter-instruction mapping feature vector includes the trigger moment of each control instruction, the fluctuation amplitude and duration of the associated power grid operation parameters; Bind labels to the parameter-instruction mapping feature vector and the comprehensive execution efficiency index sequence to generate a supervised model training sample set, where the parameter-instruction mapping feature vector is used as the input feature and the comprehensive execution efficiency index is used as the prediction label; Train a prediction model based on the supervised model training sample set; Among them, the step of training a prediction model based on the supervised model training sample set includes: Configure the initial training parameters of the prediction model. The prediction model includes a combined structure of a multi-layer time series attention network and a fully connected decision layer. The time series attention network is used to capture the time series dependence relationship in the parameter-instruction mapping feature vector, and the fully connected decision layer is used to output the association matrix between the trigger condition of the predicted control instruction and the predicted execution efficiency value; Divide the supervised model training sample set into a training subset and a validation subset according to a preset ratio, and perform multiple rounds of iterative training on the prediction model through the training subset. In each round of iteration, use the backpropagation algorithm to update the weight parameters of the time series attention network and the bias coefficients of the fully connected decision layer; After each round of iterative training is completed, the current loss function value and prediction accuracy metric of the prediction model are calculated using the validation subset. If the current loss function value does not converge or the prediction accuracy metric is lower than the preset model threshold, the number of layers of the temporal attention network or the activation function type of the fully connected decision layer is adjusted, and iterative training is restarted; When the prediction model reaches a converged state and the prediction accuracy metric is higher than the preset model threshold, the network parameters of the prediction model are frozen, and the prediction model is deployed to the real-time inference interface of the centralized control strategy optimization engine, which is used to receive the current power grid operation parameter sequence and output a set of prediction control instructions; Based on the set of prediction control instructions, the fuzzy sub-interval partitioning method and membership weight in the fuzzy rule set are pre-screened, and the fuzzy rules associated with the inefficient instructions in the set of prediction control instructions are eliminated to generate a refined fuzzy rule candidate pool; The refined fuzzy rule candidate pool is subjected to cross-cycle correlation analysis with the historical control strategy set to identify the target fuzzy rules that have a strong correlation with efficient control instructions in multiple historical cycles, and a dynamic priority coefficient is added to the target fuzzy rules; According to the dynamic priority coefficient, the rule triggering order in the fuzzy rule set is re-sorted, and the re-sorted fuzzy rule set and the prediction model are integrated into the generation process of the next round of optimized control strategy, so that the set of prediction control instructions output by the prediction model preferentially participates in the matching and screening of candidate control instructions.

10. A centralized control automation strategy optimization system based on fuzzy logic, characterized in that, The centralized control automation strategy optimization system based on fuzzy logic includes a processor and a memory. The memory is connected to the processor. The memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the centralized control automation strategy optimization method according to any one of claims 1-9 above.

Images