Fuzzy control-based high-voltage power grid fluctuation adaptive frequency conversion control method and system

Abstract

The invention relates to the technical field of automatic control of a power system, and particularly discloses a high-voltage power grid fluctuation self-adaptive frequency conversion control method and system based on fuzzy control, and the method comprises the steps: collecting power grid operation parameters, and generating an output membership function containing a plurality of candidate control instructions through fuzzy reasoning; constructing a two-dimensional arbitration space and determining an ideal reference point by calculating a rule confidence factor and a system stability cost value of each instruction, and calculating a comprehensive arbitration index based on a spatial distance; a final control instruction is selected by adopting a dynamic fault-tolerant mechanism, and the power grid frequency is accurately adjusted; and finally, optimizing the historical knowledge base through efficiency evaluation and a nonlinear updating mechanism. According to the method, the problem of a decision-making deadlock caused by a multi-peak membership function is solved, and the control precision and the system adaptive capacity under complex working conditions are improved.

Description

Technical Field

[0001] This invention relates to the field of power system automatic control technology, specifically to a high-voltage power grid fluctuation adaptive frequency conversion control method and system based on fuzzy control. Background Technology

[0002] With the large-scale integration of renewable energy and the increasing complexity of power grid structures, the power fluctuation problem faced by high-voltage power grids is becoming increasingly prominent. Traditional frequency conversion control methods mostly employ PID control based on precise mathematical models. However, when faced with complex operating conditions such as grid nonlinearity and time-varying characteristics, the control effect often exhibits problems such as response lag and overshoot. Although fuzzy control technology improves control adaptability to some extent by simulating human experience, its core defuzzification process generally uses conventional methods such as the centroid method. When the output membership function exhibits multi-peak characteristics, these methods mathematically average conflicting control strategies, resulting in control commands that lack specificity and fail to meet the high standards of control accuracy and adaptability required by modern power grids.

[0003] Existing fuzzy control suffers from decision deadlock due to multi-peak membership functions during the defuzzification process. When the power grid exhibits multiple oscillation modes simultaneously (such as the superposition of subsynchronous oscillations and low-frequency oscillations), the traditional centroid method calculates a compromise control quantity located between multiple peaks. This control quantity cannot effectively suppress high-frequency oscillations or quell low-frequency fluctuations; instead, it may introduce resonance risks due to the injection of inappropriate harmonic components, leading to further deterioration of power grid stability. Summary of the Invention

[0004] The purpose of this invention is to provide a high-voltage power grid fluctuation adaptive frequency conversion control method and system based on fuzzy control, so as to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] The adaptive frequency conversion control method for high-voltage power grid fluctuations based on fuzzy control includes the following steps:

[0007] S1: Based on real-time collected high-voltage power grid operating parameters, including node voltage, current and system frequency;

[0008] S2: The high-voltage power grid operating parameters are processed by a fuzzy inference engine to generate an output membership function containing local peak points. Each local peak point corresponds to a candidate control command, and all candidate control commands constitute the initial control command set.

[0009] S3: Calculate the rule confidence factor and system stability cost for each candidate control command;

[0010] S4: Integrate the rule confidence factor with the system stability cost to calculate a comprehensive arbitration index for each candidate control instruction; the comprehensive arbitration index is used to characterize the overall decision risk of adopting the candidate control instruction;

[0011] S5: Compare the comprehensive arbitration index of all candidate control commands, and select the candidate control command with the best comprehensive arbitration index from the initial control command set as the final frequency converter control command for the current control cycle;

[0012] S6: Output frequency control command to adjust the grid frequency, and record the control process to the historical knowledge base for subsequent updates to the rule confidence factor.

[0013] As a further aspect of the present invention: S2 specifically includes:

[0014] S201: Real-time scanning output of the domain of the membership function, identifying all points that satisfy the preset extreme value conditions as potential peak points. The preset extreme value conditions require that the membership value of the corresponding point is greater than the membership value of all other points within its adjacent predetermined range.

[0015] S202: Filter all identified potential peak points and retain only the potential peak points whose membership values ​​exceed both the global average membership value and the preset threshold, and determine them as valid local peak points.

[0016] S203: Directly map the position coordinates of each valid local peak point on the domain of the output membership function to a candidate control instruction. All the mapped candidate control instructions constitute the initial control instruction set.

[0017] As a further aspect of the present invention: the calculation process of the rule confidence factor is as follows:

[0018] S301: Obtain all fuzzy rules on which the candidate control instructions depend, form an association rule set, and query the historical knowledge base for the number of successful executions and the total number of executions of the corresponding association rule set in past execution records;

[0019] S302: Based on the number of successes and the total number of executions, calculate the initial historical performance evaluation value, and at the same time calculate the matching degree between the current power grid operating parameters and the preconditions of the corresponding associated rule set to obtain the current matching degree evaluation value;

[0020] S303: The initial historical performance evaluation value and the current matching degree evaluation value are weighted and fused, and a dynamic decay factor based on time distance is applied to the fusion result, that is, the weight of recent execution records is higher than that of distant records, and finally the rule confidence factor of the candidate control instruction is calculated.

[0021] As a further aspect of the present invention: the calculation process of the system stability cost value is as follows:

[0022] S311: Simulate the grid state in the next control cycle after the candidate control command is executed, and extract the estimated frequency deviation and voltage deviation from the simulation results;

[0023] S312: Calculate the difference between the candidate control command and the actual output control command at the previous moment to obtain the estimated value of the change in control quantity.

[0024] S313: Normalize the predicted frequency deviation, voltage deviation, and control quantity change amplitude, and then perform a weighted summation. The summation result is the system stability cost.

[0025] As a further aspect of the present invention: S4 specifically includes:

[0026] S401: Construct a two-dimensional arbitration space with the rule confidence factor as the first dimension and the system stability cost value as the second dimension, and map each candidate control instruction to a coordinate point in the arbitration space;

[0027] S402: Determine an ideal reference point in the two-dimensional arbitration space. The ideal reference point is jointly determined by the optimal value direction of the rule confidence factor among all candidate control instructions and the optimal value direction of the system stability cost value.

[0028] S403: Calculate the relative distance between the coordinate point corresponding to each candidate control command in the arbitration space and the ideal reference point, and use the value of the relative distance as the comprehensive arbitration index of the candidate control command.

[0029] As a further aspect of the present invention: S402 specifically includes:

[0030] S4021: In the two-dimensional arbitration space, select the points from the coordinate points corresponding to all candidate control commands whose rule confidence factor reaches the first predetermined optimal value, and form the confidence optimal front subset; at the same time, select the points whose system stability cost reaches the second predetermined optimal value, and form the stability optimal front subset.

[0031] S4022: Calculate the spatial overlap region between the confidence optimal front subset and the stability optimal front subset. When an overlap region exists, determine the geometric center point of the corresponding overlap region as the ideal reference point.

[0032] S4023: When the confidence-optimal front subset and the stability-optimal front subset do not have spatial overlap, select the pair of coordinate points that are closest to each other in the two subsets and determine the midpoint of this pair of points as the ideal reference point.

[0033] As a further aspect of the present invention: S5 specifically includes:

[0034] S501: Calculate the relative difference between the comprehensive arbitration index of all candidate control commands and the historical best comprehensive arbitration index, and set a dynamic fault tolerance threshold based on the current power grid fluctuation state.

[0035] S502: Select the candidate control instruction with the smallest comprehensive arbitration index value as the preliminary preferred instruction, and determine whether the relative difference of the preliminary preferred instruction exceeds the dynamic fault tolerance threshold.

[0036] S503: When the relative difference does not exceed the dynamic fault tolerance threshold, the preliminary preferred instruction is directly determined as the final frequency converter control instruction; when the relative difference exceeds the dynamic fault tolerance threshold, a second arbitration is initiated, and the candidate control instruction with the second best comprehensive arbitration index and the relative difference meets the requirements is selected as the final frequency converter control instruction.

[0037] As a further aspect of the present invention: S6 specifically includes:

[0038] S601: After outputting the final frequency conversion control command, monitor the calming process of the power grid frequency fluctuation within a predetermined time window, and obtain the actual calming time required for the fluctuation amplitude to decay to a preset threshold.

[0039] S602: Compare the actual calming time with the ideal calming time predicted based on the current power grid state, and calculate the performance evaluation value of this control process based on the comparison result;

[0040] S603: Update the historical execution records of the associated rule set for generating the final frequency converter control command based on the performance evaluation value, specifically by adjusting the success count of the rule set using a non-linear mapping method;

[0041] S604: Based on the updated historical execution records, recalculate the historical execution performance of each rule in the associated rule set, and use this as the basis for calculating the rule confidence factor in subsequent control cycles.

[0042] The high-voltage power grid fluctuation adaptive frequency conversion control system based on fuzzy control includes:

[0043] The power grid operation parameter acquisition module is based on real-time acquisition of high-voltage power grid operation parameters, including node voltage, system frequency, active power and reactive power of key branches;

[0044] The fuzzy inference and candidate instruction generation module processes the high-voltage power grid operating parameters through a fuzzy inference engine to generate an output membership function containing local peak points. Each local peak point corresponds to a candidate control instruction, and all candidate control instructions constitute the initial control instruction set.

[0045] The dual evaluation index calculation module is used to calculate the rule confidence factor and system stability cost corresponding to all candidate control instructions respectively;

[0046] The comprehensive arbitration index fusion module integrates the rule confidence factor with the system stability cost to calculate a comprehensive arbitration index for each candidate control instruction; the comprehensive arbitration index is used to characterize the overall decision risk of adopting the candidate control instruction;

[0047] The optimal instruction decision module is used to compare the comprehensive arbitration index of all candidate control instructions and select the candidate control instruction with the best comprehensive arbitration index from the initial control instruction set as the final frequency converter control instruction for the current control cycle.

[0048] The instruction execution and knowledge update module is used to output frequency control instructions to adjust the power grid frequency, and at the same time record the current control process to the historical knowledge base for subsequent updates to the rule confidence factor.

[0049] The beneficial effects of this invention are:

[0050] (1) By constructing a two-dimensional arbitration space and introducing a comprehensive arbitration mechanism based on confidence and stability costs, the decision deadlock problem caused by the traditional centroid method in the scenario of multi-peak membership functions is effectively solved. This method can identify and avoid high-risk, low-confidence control commands, ensuring that the system can still make decisive and accurate control decisions under complex fluctuations such as subsynchronous oscillations and low-frequency oscillations, thus avoiding the resonance risk and system stability deterioration caused by invalid or inappropriate control commands.

[0051] (2) By establishing an effectiveness evaluation system based on the actual fluctuation calming time and combining it with a nonlinear update mechanism to dynamically correct the historical knowledge base, the confidence factor of the fuzzy rule can truly reflect its control effectiveness. This closed-loop learning mechanism enables the control system to continuously accumulate operating experience and adaptively adjust decision-making strategies, which not only improves the adaptability to different power grid disturbance modes, but also reduces the dependence on external parameter settings and enhances the long-term operational robustness of the system. Attached Figure Description

[0052] The invention will now be further described with reference to the accompanying drawings.

[0053] Figure 1 This is a flowchart of the method of the present invention;

[0054] Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Please see Figure 1 As shown, this invention is a high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control, which includes the following steps:

[0057] S1: Based on real-time collected high-voltage power grid operating parameters, including node voltage, system frequency, active power and reactive power of key branches;

[0058] S2: The high-voltage power grid operating parameters are processed by a fuzzy inference engine to generate an output membership function containing local peak points. Each local peak point corresponds to a candidate control command, and all candidate control commands constitute the initial control command set.

[0059] S3: Calculate the rule confidence factor and system stability cost for each candidate control command;

[0060] S4: Integrate the rule confidence factor with the system stability cost to calculate a comprehensive arbitration index for each candidate control instruction; the comprehensive arbitration index is used to characterize the overall decision risk of adopting the candidate control instruction;

[0061] S5: Compare the comprehensive arbitration index of all candidate control commands, and select the candidate control command with the best comprehensive arbitration index from the initial control command set as the final frequency converter control command for the current control cycle;

[0062] S6: Output frequency control command to adjust the grid frequency, and record the control process to the historical knowledge base for subsequent updates to the rule confidence factor.

[0063] In S1, the acquisition of high-voltage power grid operating parameters is achieved through sensing and measurement equipment deployed at key nodes of the power grid. Specifically, high-precision voltage transformers are used to measure the node-to-ground voltage of the three-phase power grid in real time; current transformers are connected in series in the transmission lines to measure the real-time current value of the conductors; and the system frequency is measured by a dedicated frequency measurement unit or by zero-crossing detection and period calculation of the voltage waveform.

[0064] In S2, after the fuzzy inference process is completed, the output membership function is obtained. The domain of this function represents the range of values ​​for the control variable, and the range represents the membership degree corresponding to each value. The generation process of the output membership function is as follows: the fuzzified power grid operating parameters are input into the fuzzy inference engine, and inference calculations are performed according to the pre-set fuzzy rule base. Each fuzzy rule is in the form of "output the corresponding conclusion if the premise condition is true". Finally, the output conclusions of all activated rules are aggregated to form a continuous output membership function.

[0065] The specific implementation process for identifying local peak points in the output membership function is as follows: Starting from the beginning of the domain of the output membership function, a scan is performed with a fixed step size. For each point to be determined in the domain, a predetermined range before and after it is taken as its neighborhood, which is set to 2% of the total length of the domain. The membership value of the point to be determined is calculated and compared with the membership values ​​of all other points in its neighborhood. When the membership value of the point is strictly greater than the membership values ​​of all other points in its neighborhood, the point is determined to meet the preset extreme value condition and is marked as a potential peak point. This process is repeated until the entire domain is scanned, obtaining the position coordinates and corresponding membership values ​​of all potential peak points.

[0066] The specific implementation process for screening the identified potential peak points is as follows: First, the global average membership degree of the output membership function is calculated. This is done by uniformly discretizing the domain into one hundred points and calculating the arithmetic mean of the membership degree values ​​for each point. Simultaneously, a preset threshold is set to 0.6. Then, each potential peak point is examined one by one, retaining only those whose membership degree values ​​are simultaneously greater than both the global average membership degree and the preset threshold. The potential peak points retained after this screening step are formally identified as valid local peak points; these points represent control instruction candidates with sufficient significance.

[0067] The specific implementation process of mapping valid local peak points to candidate control commands is as follows: For each valid local peak point, its corresponding position coordinates on the domain of the output membership function are read. This position coordinate is a specific numerical value representing the magnitude of a specific control quantity. This value is directly used as the output value of the candidate control command. For example, when the output membership function represents reactive power compensation, if the position coordinate of a certain local peak point is +35.2, then the corresponding candidate control command is to inject 35.2 Mvars of reactive power into the grid. All candidate control commands obtained through this mapping constitute the initial control command set, providing a basis for subsequent decision arbitration. It should be noted that the domain of the output membership function is pre-set according to the physical characteristics of the specific controlled object to ensure that the generated control commands are within the safe operating range of the equipment.

[0068] In S3, the calculation process of the rule confidence factor is implemented as follows: First, all fuzzy rules on which the candidate control command depends are obtained; these rules constitute the associated rule set. The execution records of this associated rule set are queried in the historical knowledge base, including the number of successes and the total number of executions. The criterion for the number of successes is: after the control command corresponding to the rule set is executed, the power grid frequency fluctuation amplitude decreases to within 0.5% of the rated value within two control cycles. The initial historical performance evaluation value is calculated by the ratio of the number of successes to the total number of executions; this ratio reflects the historical reliability of the rule set.

[0069] The current matching degree evaluation value is calculated using the following method: The matching degree between the current power grid operating parameters and the preconditions of each rule in the associated rule set is calculated, and the arithmetic mean of all matching degree values ​​is taken as the current matching degree evaluation value. The matching degree of a single rule is calculated using the following formula: ;in Indicates the first The degree of matching of the rules, This indicates the number of fuzzy variables in the preconditions. Indicates the first The current precise value of a fuzzy variable. Indicates the first Rule number 1 The central value of the membership function of a fuzzy variable This represents the radius of the membership function.

[0070] When weighting and fusing the initial historical performance evaluation value with the current matching degree evaluation value, the following calculation formula is used: ;in Indicates the fusion result. This represents the initial historical performance evaluation value. This indicates the current matching score. and These are weighting coefficients, and they satisfy... .

[0071] The dynamic decay factor is applied as follows: historical records are assigned different weights based on their time proximity. Records from the most recent 30 days have a weight of 1.0, records from 30 to 90 days ago have a weight that linearly decays to 0.5, and records older than 90 days have a fixed weight of 0.2. The final rule confidence factor is calculated using the following formula: ;in This represents the confidence factor for the rule. Indicates the total number of records. This represents the time weight of the k-th record. Indicates the first The success flag for each record (1 for success, 0 for failure).

[0072] The calculation process for the system stability cost is implemented as follows: When simulating the grid state after the execution of candidate control commands, a simplified grid model is used for single-step prediction. This model is based on the grid's equivalent inertia constant and damping coefficient, and the frequency deviation prediction is obtained by solving the following differential equation: ;in This represents the estimated frequency deviation. Indicates the power imbalance. Indicates the damping coefficient. This represents the inertia constant. The voltage deviation estimate is obtained by solving the network power flow equations, taking into account the impact of control commands on the reactive power of nodes.

[0073] The predicted value of the control variable change amplitude is obtained by calculating the absolute difference between the current candidate control command and the actual output control command at the previous moment. Normalization is performed using the maximum-minimum method, transforming each predicted value to the [0,1] interval. During weighted summation, the weights for the frequency deviation predicted value and voltage deviation predicted value are all set at 0.4 and 0.3 respectively. The weighted summation result is the system stability cost; a larger value indicates a more detrimental potential impact of the control command on system stability.

[0074] In S4, the generation of the comprehensive arbitration index is achieved by constructing a two-dimensional arbitration space and calculating the relative distance of each candidate control command to the ideal reference point. This process begins by constructing a two-dimensional coordinate system with the rule confidence factor as the horizontal axis and the system stability cost value as the vertical axis. Each candidate control command is mapped to a specific coordinate point in this coordinate system based on its corresponding rule confidence factor and system stability cost value. The rule confidence factor ranges from [0,1], and the system stability cost value also ranges from [0,1], forming a unit square region as the complete arbitration space.

[0075] The process of determining the ideal reference point involves three specific steps. First, in the two-dimensional arbitration space, all coordinate points whose rule confidence factor reaches or exceeds a first predetermined optimal value are selected. These points constitute the confidence-optimal front subset, where the first predetermined optimal value is set to 0.9. Simultaneously, all coordinate points whose system stability cost reaches or exceeds a second predetermined optimal value are selected. These points constitute the stability-optimal front subset, where the second predetermined optimal value is set to 0.9. Next, it is checked whether these two subsets have a spatial overlap, i.e., whether there are coordinate points belonging to both subsets simultaneously. When such an overlap exists, the geometric center point of the overlap is calculated. Specifically, the arithmetic mean of the x-coordinates of all points within the overlap is taken as the x-coordinate of the ideal reference point, and the arithmetic mean of the y-coordinates of all points within the overlap is taken as the y-coordinate of the ideal reference point. When there is no spatial overlap between the two subsets, the Euclidean distance between each point in the confidence-optimal front subset and each point in the stability-optimal front subset is calculated. The pair of coordinate points that are closest to each other is selected, and the midpoint of these two points is calculated as the ideal reference point. The midpoint coordinate is calculated by averaging the x-coordinates of the two points as the x-coordinate of the ideal reference point and averaging the y-coordinates of the two points as the y-coordinate of the ideal reference point.

[0076] The process of calculating the comprehensive arbitration index involves calculating the Euclidean distance between the coordinate point corresponding to each candidate control command and the ideal reference point. Specifically, this is done by adding the square of the difference between the x-coordinate of the point and the x-coordinate of the ideal reference point, plus the square of the difference between the y-coordinate of the point and the y-coordinate of the ideal reference point, and then taking the square root of the sum. The resulting distance value is the comprehensive arbitration index of the candidate control command. The smaller the index value, the closer the candidate control command's overall performance in terms of rule confidence and system stability is to the ideal state, and the lower the corresponding decision risk. This evaluation method based on spatial relative distance enables a multi-dimensional comprehensive assessment of candidate control commands, providing a quantitative basis for the final selection of the control command.

[0077] In S5, the final determination of the frequency converter control command involves three specific steps. First, the relative difference between the comprehensive arbitration index of all candidate control commands and the historical best comprehensive arbitration index is calculated, and a dynamic fault-tolerance threshold based on the current grid fluctuation state is set. The relative difference is calculated by subtracting the historical best comprehensive arbitration index from the current candidate control command's comprehensive arbitration index, and then dividing by the historical best comprehensive arbitration index to obtain the relative rate of change. The dynamic fault-tolerance threshold is set based on the current grid frequency deviation and frequency change rate. When the frequency deviation exceeds 0.8% of the rated value and the frequency change rate exceeds 0.1 Hz / s, the dynamic fault-tolerance threshold is set to 0.25; when the frequency deviation is between 0.4% and 0.8% of the rated value and the frequency change rate is between 0.05 Hz / s and 0.1 Hz / s, the dynamic fault-tolerance threshold is set to 0.20; otherwise, the dynamic fault-tolerance threshold is set to 0.15. The historical best comprehensive arbitration index is the minimum comprehensive arbitration index value that has appeared in the most recent 100 control cycles.

[0078] From all candidate control instructions, the instruction with the lowest comprehensive arbitration index value is selected as the preliminary preferred instruction, and it is then determined whether the relative difference of this preliminary preferred instruction exceeds the dynamic fault tolerance threshold. The selection process involves comparing the comprehensive arbitration index values ​​of all candidate control instructions and choosing the instruction with the lowest value. The determination process compares the relative difference of the preliminary preferred instruction with the dynamic fault tolerance threshold. If the relative difference is less than or equal to the dynamic fault tolerance threshold, the process proceeds to the next step; if the relative difference is greater than the dynamic fault tolerance threshold, a secondary arbitration procedure is initiated.

[0079] The final inverter control command is determined based on the relative difference judgment result. If the relative difference of the initially selected command does not exceed the dynamic fault tolerance threshold, the initially selected command is directly determined as the final inverter control command. If the relative difference of the initially selected command exceeds the dynamic fault tolerance threshold, a secondary arbitration procedure is initiated. This procedure involves selecting the command with the second smallest comprehensive arbitration index from the remaining candidate control commands as a new candidate command, calculating its relative difference, and determining it as the final inverter control command if the relative difference does not exceed the dynamic fault tolerance threshold. If it still exceeds the dynamic fault tolerance threshold, the command with the third smallest comprehensive arbitration index is selected for judgment, until the first candidate control command with a relative difference not exceeding the dynamic fault tolerance threshold is found and determined as the final inverter control command. If the relative difference of all candidate control commands exceeds the dynamic fault tolerance threshold, the command with the smallest comprehensive arbitration index is selected as the final inverter control command, and a warning signal is generated simultaneously.

[0080] In S6, the specific implementation process of control command execution and knowledge update includes the following four steps. After outputting the final frequency converter control command, monitoring of the grid frequency fluctuation calming process is initiated, with a predetermined monitoring time window set at 10 seconds. Within this time window, the grid frequency deviation value is continuously measured. When the absolute value of the frequency deviation remains within 0.1% of the rated frequency for 0.5 seconds consecutively, the fluctuation is considered to have calmed down. The time interval from the control command output time to the fluctuation calming time is recorded as the actual calming time. If the fluctuation fails to calm down within this time window, the actual calming time is recorded as 10 seconds.

[0081] The actual calming time is compared with the ideal calming time predicted based on the current power grid state. The ideal calming time is calculated as follows: First, the disturbance intensity coefficient is determined based on the initial frequency change rate of the power grid disturbance. Then, combined with the equivalent inertia constant and damping coefficient of the power grid, the theoretical minimum calming time is calculated using a formula. Specifically, the ideal calming time equals the product of the disturbance intensity coefficient and the equivalent inertia constant of the power grid, divided by the square root of the damping coefficient. The effectiveness evaluation value of this control process is calculated as follows: when the actual calming time is less than or equal to the ideal calming time, the effectiveness evaluation value is 1; when the actual calming time is greater than the ideal calming time, the effectiveness evaluation value is the ratio of the ideal calming time to the actual calming time.

[0082] Based on the calculated performance evaluation value, the historical execution records of the associated rule set that generated the final frequency converter control command are updated. The update process uses a non-linear mapping method to adjust the success count of the rule set: when the performance evaluation value is greater than or equal to 0.8, the success count increases by 2; when the performance evaluation value is between 0.6 and 0.8, the success count increases by 1; when the performance evaluation value is below 0.6, the success count remains unchanged. Regardless of the performance evaluation value, the total number of executions increases by 1. This non-linear update method ensures that rules with excellent control performance accumulate more success counts.

[0083] Based on the updated historical execution records, the historical execution performance of each rule in the associated rule set is recalculated. The historical execution performance is calculated by dividing the sum of the success counts of all rules in the rule set by the sum of the total execution counts. Simultaneously, the individual execution performance of each rule in the rule set is also updated by dividing the success count of that rule by its total execution count. The updated historical execution performance data will be stored in a historical knowledge base as the basis for calculating rule confidence factors in subsequent control cycles, thereby enabling the self-learning and continuous optimization of the control strategy.

[0084] Please see Figure 2 As shown, the high-voltage power grid fluctuation adaptive frequency conversion control system based on fuzzy control includes:

[0085] The power grid operation parameter acquisition module is based on real-time acquisition of high-voltage power grid operation parameters, including node voltage, system frequency, active power and reactive power of key branches;

[0086] The fuzzy inference and candidate instruction generation module processes the high-voltage power grid operating parameters through a fuzzy inference engine to generate an output membership function containing local peak points. Each local peak point corresponds to a candidate control instruction, and all candidate control instructions constitute the initial control instruction set.

[0087] The dual evaluation index calculation module is used to calculate the rule confidence factor and system stability cost corresponding to all candidate control instructions respectively;

[0088] The comprehensive arbitration index fusion module integrates the rule confidence factor with the system stability cost to calculate a comprehensive arbitration index for each candidate control instruction; the comprehensive arbitration index is used to characterize the overall decision risk of adopting the candidate control instruction;

[0089] The optimal instruction decision module is used to compare the comprehensive arbitration index of all candidate control instructions and select the candidate control instruction with the best comprehensive arbitration index from the initial control instruction set as the final frequency converter control instruction for the current control cycle.

[0090] The instruction execution and knowledge update module is used to output frequency control instructions to adjust the power grid frequency, and at the same time record the current control process to the historical knowledge base for subsequent updates to the rule confidence factor.

[0091] The working principle of this invention is as follows: By real-time acquisition of node voltage, system frequency, and key branch power parameters, an output membership function containing multiple local peak points is generated through fuzzy inference, and each peak point is mapped to a candidate control command. By calculating the rule confidence factor and system stability cost corresponding to each command, a two-dimensional arbitration space is constructed and an ideal reference point is determined. A comprehensive arbitration index is calculated based on the spatial relative distance. A dynamic fault-tolerant mechanism is used to select the final control command from the candidate commands to achieve power grid frequency regulation. Finally, the control effectiveness is evaluated by monitoring the fluctuation calming process, and the historical knowledge base is optimized in a nonlinear update manner to form a closed-loop control system with self-learning ability, effectively solving the multi-peak decision deadlock problem and improving control accuracy and system stability.

[0092] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control, characterized in that, Includes the following steps: S1: Based on real-time collected high-voltage power grid operating parameters, including node voltage, current and system frequency; S2: The high-voltage power grid operating parameters are processed by a fuzzy inference engine to generate an output membership function containing local peak points. Each local peak point corresponds to a candidate control command, and all candidate control commands constitute the initial control command set. S3: Calculate the rule confidence factor and system stability cost for each candidate control command; S4: Integrate the rule confidence factor with the system stability cost to calculate a comprehensive arbitration index for each candidate control instruction; The comprehensive arbitration index is used to characterize the overall decision risk of adopting candidate control instructions; S5: Compare the comprehensive arbitration index of all candidate control commands, and select the candidate control command with the best comprehensive arbitration index from the initial control command set as the final frequency converter control command for the current control cycle; S6: Output frequency control command to adjust the grid frequency, and record the control process to the historical knowledge base for subsequent updates to the rule confidence factor.

2. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 1, characterized in that, S2 specifically includes: S201: Real-time scanning output of the domain of the membership function, identifying all points that satisfy the preset extreme value conditions as potential peak points. The preset extreme value conditions require that the membership value of the corresponding point is greater than the membership value of all other points within its adjacent predetermined range. S202: Filter all identified potential peak points and retain only the potential peak points whose membership values ​​exceed both the global average membership value and the preset threshold, and determine them as valid local peak points. S203: Directly map the position coordinates of each valid local peak point on the domain of the output membership function to a candidate control instruction. All the mapped candidate control instructions constitute the initial control instruction set.

3. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 1, characterized in that, The calculation process for the rule confidence factor is as follows: S301: Obtain all fuzzy rules on which the candidate control instructions depend, form an association rule set, and query the historical knowledge base for the number of successful executions and the total number of executions of the corresponding association rule set in past execution records; S302: Based on the number of successes and the total number of executions, calculate the initial historical performance evaluation value, and at the same time calculate the matching degree between the current power grid operating parameters and the preconditions of the corresponding associated rule set to obtain the current matching degree evaluation value; S303: The initial historical performance evaluation value and the current matching degree evaluation value are weighted and fused, and a dynamic decay factor based on time distance is applied to the fusion result, that is, the weight of recent execution records is higher than that of distant records, and finally the rule confidence factor of the candidate control instruction is calculated.

4. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 1, characterized in that, The calculation process for the system stability cost is as follows: S311: Simulate the grid state in the next control cycle after the candidate control command is executed, and extract the estimated frequency deviation and voltage deviation from the simulation results; S312: Calculate the difference between the candidate control command and the actual output control command at the previous moment to obtain the estimated value of the change in control quantity. S313: Normalize the predicted frequency deviation, voltage deviation, and control quantity change amplitude, and then perform a weighted summation. The summation result is the system stability cost.

5. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 1, characterized in that, S4 specifically includes: S401: Construct a two-dimensional arbitration space with the rule confidence factor as the first dimension and the system stability cost value as the second dimension, and map each candidate control instruction to a coordinate point in the arbitration space; S402: Determine an ideal reference point in the two-dimensional arbitration space. The ideal reference point is jointly determined by the optimal value direction of the rule confidence factor among all candidate control instructions and the optimal value direction of the system stability cost value. S403: Calculate the relative distance between the coordinate point corresponding to each candidate control command in the arbitration space and the ideal reference point, and use the value of the relative distance as the comprehensive arbitration index of the candidate control command.

6. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 5, characterized in that, S402 specifically includes: S4021: In the two-dimensional arbitration space, select the points from the coordinate points corresponding to all candidate control commands whose rule confidence factor reaches the first predetermined optimal value, and form the confidence optimal front subset; at the same time, select the points whose system stability cost reaches the second predetermined optimal value, and form the stability optimal front subset. S4022: Calculate the spatial overlap region between the confidence optimal front subset and the stability optimal front subset. When an overlap region exists, determine the geometric center point of the corresponding overlap region as the ideal reference point. S4023: When the confidence-optimal front subset and the stability-optimal front subset do not have spatial overlap, select the pair of coordinate points that are closest to each other in the two subsets and determine the midpoint of this pair of points as the ideal reference point.

7. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 1, characterized in that, S5 specifically includes: S501: Calculate the relative difference between the comprehensive arbitration index of all candidate control commands and the historical best comprehensive arbitration index, and set a dynamic fault tolerance threshold based on the current power grid fluctuation state. S502: Select the candidate control instruction with the smallest comprehensive arbitration index value as the preliminary preferred instruction, and determine whether the relative difference of the preliminary preferred instruction exceeds the dynamic fault tolerance threshold. S503: When the relative difference does not exceed the dynamic fault tolerance threshold, the preliminary preferred instruction is directly determined as the final frequency converter control instruction; when the relative difference exceeds the dynamic fault tolerance threshold, a second arbitration is initiated, and the candidate control instruction with the second best comprehensive arbitration index and the relative difference meets the requirements is selected as the final frequency converter control instruction.

8. The high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control according to claim 1, characterized in that, S6 specifically includes: S601: After outputting the final frequency conversion control command, monitor the calming process of the power grid frequency fluctuation within a predetermined time window, and obtain the actual calming time required for the fluctuation amplitude to decay to a preset threshold. S602: Compare the actual calming time with the ideal calming time predicted based on the current power grid state, and calculate the performance evaluation value of this control process based on the comparison result; S603: Update the historical execution records of the associated rule set for generating the final frequency converter control command based on the performance evaluation value, specifically by adjusting the success count of the rule set using a non-linear mapping method; S604: Based on the updated historical execution records, recalculate the historical execution performance of each rule in the associated rule set, and use this as the basis for calculating the rule confidence factor in subsequent control cycles.

9. A high-voltage power grid fluctuation adaptive frequency conversion control system based on fuzzy control, characterized in that, The method for implementing the high-voltage power grid fluctuation adaptive frequency conversion control method based on fuzzy control as described in any one of claims 1-8 includes: The power grid operation parameter acquisition module is based on real-time acquisition of high-voltage power grid operation parameters, including node voltage, system frequency, active power and reactive power of key branches; The fuzzy inference and candidate instruction generation module processes the high-voltage power grid operating parameters through a fuzzy inference engine to generate an output membership function containing local peak points. Each local peak point corresponds to a candidate control instruction, and all candidate control instructions constitute the initial control instruction set. The dual evaluation index calculation module is used to calculate the rule confidence factor and system stability cost corresponding to all candidate control instructions respectively; The comprehensive arbitration index fusion module integrates the rule confidence factor with the system stability cost to calculate a comprehensive arbitration index for each candidate control instruction; the comprehensive arbitration index is used to characterize the overall decision risk of adopting the candidate control instruction; The optimal instruction decision module is used to compare the comprehensive arbitration index of all candidate control instructions and select the candidate control instruction with the best comprehensive arbitration index from the initial control instruction set as the final frequency converter control instruction for the current control cycle. The instruction execution and knowledge update module is used to output frequency conversion control instructions to adjust the grid frequency, and at the same time record the current control process to the historical knowledge base for subsequent updates of the rule confidence factor.

Images