Novel power system supply and demand balance effect evaluation method based on fuzzy logic

Abstract

The invention relates to the technical field of novel electric power, in particular to a novel electric power system supply and demand balance effect evaluation method based on fuzzy logic, which comprises the following steps: constructing a hierarchical evaluation index system including power supply reliability, system flexibility, green low-carbon level and supply and demand collaboration degree; generating an interval type-2 fuzzy set by adopting an interval type-2 fuzzy membership function aiming at the uncertainty input index; meanwhile, constructing a fuzzy cognitive map, inputting real-time data by using a long and short term memory network model, and generating a dynamic weight matrix representing a dynamic time-varying nonlinear coupling relationship among the indexes; and finally, inputting the fuzzy set and the dynamic weight matrix into an updating iteration mechanism, carrying out iteration through interval type-2 fuzzy arithmetic operation, outputting a fuzzy state vector, carrying out defuzzification processing on the fuzzy state vector, and generating a comprehensive supply and demand balance effect evaluation value. According to the invention, quantitative evaluation of the supply and demand balance effect of the novel power system under high uncertainty is realized through fuzzy logic.

Description

Technical Field

[0001] This invention relates to the field of new power technology, specifically to a novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic. Background Technology

[0002] In new power systems, with the grid connection of high proportions of renewable energy sources such as wind and solar power, the source-load characteristics of the power system have undergone fundamental changes. On the one hand, the processing of renewable energy has significant intermittency, volatility and randomness; on the other hand, the randomness and peak-valley difference of user-side loads (such as electric vehicle charging piles) are increasing. This high uncertainty on both the source and load sides poses a huge challenge to the real-time supply and demand balance of the power system. Current methods for evaluating the effectiveness of power system supply and demand balance largely rely on traditional deterministic models or probabilistic statistical analysis. These methods often encounter problems when dealing with the multiple uncertainties, fuzziness, and complex nonlinear coupling relationships unique to new power systems. For example, evaluating balance effectiveness requires not only considering whether the power supply is sufficient, but also comprehensively considering multiple fuzzy and mutually influential indicators such as the timeliness and stability of the balance and the level of renewable energy absorption. Therefore, a specific technical problem that urgently needs to be solved is how to establish an evaluation model that effectively integrates and processes multi-source, fuzzy, and uncertain information in order to scientifically and accurately evaluate the actual effectiveness of the supply and demand balance of new power systems under strong random disturbances. To address this, a novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic is proposed. Summary of the Invention

[0003] The purpose of this invention is to provide a novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic, which realizes the quantitative evaluation of the supply and demand balance effectiveness of the novel power system under high uncertainty through fuzzy logic.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic includes: Collect historical and real-time operational data of the new power system, and construct a supply and demand balance effectiveness evaluation index system based on the historical operational data; For the uncertain input indicators in the indicator system, an interval type II fuzzy membership function is used for fuzzification to generate an interval type II fuzzy set representing its uncertainty footprint; a fuzzy cognitive graph is constructed with each indicator in the indicator system as a node, and the real-time running data is input into the trained long short-term memory network model to generate the dynamic weight matrix of the fuzzy cognitive graph at the current time. The dynamic weight matrix is ​​used to represent the dynamic time-varying nonlinear coupling relationship between each indicator node. The interval type II fuzzy set and the dynamic weight matrix generated at the current time are input into the update iteration mechanism to map and output the fuzzy state vector; The fuzzy state vector is defuzzified to generate a comprehensive supply and demand balance effectiveness evaluation value.

[0005] Preferably, the construction of the supply and demand balance effectiveness evaluation index system based on the historical operating data specifically includes: Analyze the historical operational data to construct a hierarchical evaluation indicator system that includes a primary indicator layer and a secondary indicator layer; The primary indicator layer includes power supply reliability, system flexibility, green and low-carbon level, and supply and demand coordination. The indicators in the secondary indicator layer correspond to those in the primary indicator layer. The indicators corresponding to power supply reliability include average outage time, voltage qualification rate, and frequency qualification rate; the indicators corresponding to system flexibility include system net load ramp-up rate, renewable energy absorption rate, and flexibility resource adjustability margin; the indicators corresponding to green and low-carbon level include clean energy power generation penetration rate and carbon emission factor per unit of power supply; and the indicators corresponding to supply and demand coordination include load peak shaving and valley filling rate and demand-side response participation.

[0006] Preferably, the specific process of generating the interval type-2 fuzzy set representing its uncertainty footprint is as follows: The uncertainty input indicators include renewable energy output forecast data and load forecast data; Acquire and analyze the historical prediction error sequence of the aforementioned uncertainty input indicators in the historical operating data, wherein the historical prediction error sequence consists of the deviation between the historical predicted value and the historical actual value; Statistical analysis is performed on the historical prediction error sequence to determine its probability distribution characteristics, and the interval type II fuzzy membership function is constructed based on the distribution characteristics. The interval type II fuzzy membership function is defined by the upper membership function and the lower membership function to form the interval type II fuzzy set.

[0007] Preferably, the specific process of constructing a fuzzy cognitive graph using each indicator in the indicator system as a node is as follows: The first-level indicator layer and the second-level indicator layer are set as concept nodes of the fuzzy cognitive graph. The causal relationship topology between the concept nodes is determined based on the historical operation data, and an adjacency matrix representing the topology is constructed. The causal relationship topology includes the horizontal coupling relationship between the secondary indicator nodes and the vertical aggregation relationship between the secondary indicator nodes and their corresponding primary indicator nodes. The adjacency matrix defines whether there is a connection between the nodes.

[0008] Preferably, the specific process of generating the dynamic weight matrix of the fuzzy cognitive map at the current moment is as follows: The Long Short-Term Memory (LSTM) network model is trained offline using the historical operational data to learn and fit the nonlinear coupling relationship pattern between the indicator nodes in the indicator system over time. The time series data of the indicators corresponding to the indicator nodes of the fuzzy cognitive map in the real-time operational data are then input into the trained LSM network model. The LSM network model performs forward inference calculations based on the currently input real-time indicator time series data and outputs a dynamic weight matrix.

[0009] Preferably, the specific process of mapping and outputting the fuzzy state vector is as follows: The interval type II fuzzy set is used as the initial fuzzy state of the corresponding uncertainty input index node in the fuzzy cognitive graph, and the initial states of other nodes are set to form the initial fuzzy state vector. The fuzzy state vector is iteratively calculated using the fuzzy cognitive graph iterative reasoning rule based on the dynamic weight matrix. The iterative calculation is based on interval type II fuzzy arithmetic operations, and the fuzzy state value of each index node is updated at each time step. After each iteration, it is determined whether the fuzzy state vector meets the preset convergence condition. The convergence condition is usually that the change in the fuzzy state vector between two consecutive iterations is less than a preset threshold. When the convergence condition is met, the iteration stops, and the fuzzy state vector in the final stable state is used as the output of the update iteration mechanism.

[0010] Preferably, the specific process for generating the comprehensive supply and demand balance effectiveness evaluation value is as follows: Extract the type II fuzzy state values ​​of each interval corresponding to the first-level index layer from the fuzzy state vector; For each extracted primary index, the interval type II fuzzy state value is subjected to defuzzification calculation, which includes type simplification and centroid calculation. The type simplification adopts a type simplification algorithm to simplify the interval type II fuzzy state value into an interval type I fuzzy set, where the interval represents the uncertainty range of its centroid. The centroid calculation obtains the clear evaluation value of the primary indicators by calculating the midpoint of the centroid interval; the clear evaluation values ​​of all primary indicators are weighted and summed with their corresponding evaluation weights to calculate the final comprehensive supply and demand balance effectiveness evaluation value.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This application addresses the inherent high uncertainty of key data such as renewable energy output and load forecasting in new power systems by employing interval type II fuzzy membership functions for fuzzification. Compared to traditional methods or type I fuzzy logic, interval type II fuzzy sets can more fully characterize and quantify the uncertainty footprint caused by historical forecast errors. This allows the evaluation model to maintain strong stability and reliability even when faced with fluctuations in input data and inaccurate forecasts, resulting in more credible evaluation results.

[0012] 2. This application combines Long Short-Term Memory Networks with Fuzzy Cognitive Maps and uses real-time running data to drive the generation of the dynamic weight matrix of FCM. This overcomes the shortcomings of traditional evaluation methods, which have fixed weights and cannot reflect the time-varying characteristics of the system. The model can capture the dynamic and complex nonlinear coupling relationship between various evaluation indicators in real time. Therefore, the evaluation results have high immediacy and adaptability and can accurately reflect the real balance state of the power system at the current moment.

[0013] 3. This application constructs a multi-dimensional, hierarchical indicator system covering power supply reliability, system flexibility, green and low-carbon level, and supply-demand coordination. The system is clearly simulated through the vertical aggregation and horizontal coupling relationship of the fuzzy cognitive graph. This method not only comprehensively considers the multiple objectives of supply and demand balance in the new power system, but also, through iterative reasoning and final defuzzification calculation, can converge the complex multi-indicator fuzzy state and transform it into a single, clear comprehensive evaluation value, providing an intuitive and comprehensive decision-making basis for system scheduling and planning. Attached Figure Description

[0014] Figure 1 A flowchart illustrating a novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic; Figure 2 This is a schematic diagram of the supply and demand balance effectiveness evaluation index system structure of the present invention. Detailed Implementation

[0015] 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.

[0016] Please see Figure 1 and Figure 2 This invention provides a novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic, the technical solution of which is as follows: A novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic includes: Collect historical and real-time operational data of the new power system, and construct a supply and demand balance effectiveness evaluation index system based on the historical operational data; For the uncertain input indicators in the indicator system, an interval type II fuzzy membership function is used for fuzzification to generate an interval type II fuzzy set representing its uncertainty footprint; a fuzzy cognitive graph is constructed with each indicator in the indicator system as a node, and the real-time running data is input into the trained long short-term memory network model to generate the dynamic weight matrix of the fuzzy cognitive graph at the current time. The dynamic weight matrix is ​​used to represent the dynamic time-varying nonlinear coupling relationship between each indicator node. The interval type II fuzzy set and the dynamic weight matrix generated at the current time are input into the update iteration mechanism to map and output the fuzzy state vector; The fuzzy state vector is defuzzified to generate a comprehensive supply and demand balance effectiveness evaluation value.

[0017] The specific details of constructing the supply and demand balance effectiveness evaluation index system based on the historical operational data are as follows: Analyze the historical operational data to construct a hierarchical evaluation indicator system that includes a primary indicator layer and a secondary indicator layer; The primary indicator layer includes power supply reliability, system flexibility, green and low-carbon level, and supply and demand coordination. The indicators in the secondary indicator layer correspond to those in the primary indicator layer. The indicators corresponding to power supply reliability include average outage time, voltage qualification rate, and frequency qualification rate; the indicators corresponding to system flexibility include system net load ramp-up rate, renewable energy absorption rate, and flexibility resource adjustability margin; the indicators corresponding to green and low-carbon level include clean energy power generation penetration rate and carbon emission factor per unit of power supply; and the indicators corresponding to supply and demand coordination include load peak shaving and valley filling rate and demand-side response participation.

[0018] By constructing a primary indicator layer that includes four core dimensions—power supply reliability, system flexibility, green and low-carbon level, and supply-demand coordination—and supplementing it with specific secondary indicators for quantitative support, the evaluation framework comprehensively and systematically covers the multiple objectives of supply-demand balance in the new power system, while ensuring a clear and hierarchical evaluation structure, breaking down macro-level effectiveness assessments into specific and measurable indicators.

[0019] The specific process for generating the interval type-II fuzzy set representing its uncertainty footprint is as follows: The uncertainty input indicators include renewable energy output forecast data and load forecast data; Acquire and analyze the historical prediction error sequence of the aforementioned uncertainty input indicators in the historical operating data, wherein the historical prediction error sequence consists of the deviation between the historical predicted value and the historical actual value; Specifically, for each indicator, two sets of corresponding time series data need to be obtained: (a) historical forecast value series; (b) historical actual value series. Taking the renewable energy output indicator as an example, the historical actual value series and the historical forecast value series are compared point by point on the same time section, the deviation between the two is calculated, and all the calculated deviation values ​​are combined in time order to generate the renewable energy historical forecast error series. The same operation is performed on the load indicator to generate the load historical forecast error series. Statistical analysis is performed on the historical prediction error sequence to determine its probability distribution characteristics, and the interval type II fuzzy membership function is constructed based on the distribution characteristics. The interval type II fuzzy membership function is defined by the upper membership function and the lower membership function to define the interval type II fuzzy set. The probability distribution features are the mean and standard deviation of the historical prediction error sequence. The standard deviation is determined by statistically analyzing the fluctuation range of the standard deviation under different working conditions through sliding window analysis, thereby determining the standard deviation interval. The standard deviation interval includes a lower bound and an upper bound of the standard deviation. Two membership functions, an upper membership function and a lower membership function, are constructed based on the mean and standard deviation. Specifically, the two membership functions are constructed using a Gaussian function as the prototype of the membership function. The upper membership function is constructed based on the upper bound of the standard deviation, and the lower membership function is constructed based on the lower bound of the standard deviation. A two-dimensional coordinate system is constructed, where the horizontal axis represents all possible values ​​of the prediction error variable, and the vertical axis represents the membership degree calculated by the upper and lower membership functions. A series of continuous values ​​on the horizontal axis (error value domain) are respectively input as independent variables into the upper membership function. The calculated output value of the upper membership function (i.e., the upper membership degree) is plotted in the coordinate system to form the first curve (i.e., the upper boundary). The same series of continuous values ​​on the horizontal axis are respectively input as independent variables into the lower membership function. The calculated output value of the lower membership function (i.e., the lower membership degree) is plotted in the coordinate system to form a second curve (i.e., the lower boundary). The first curve is always located above the second curve in the coordinate system, or coincides with it at a specific point (such as the mean point). The two-dimensional plane region jointly enclosed and defined by these two curves in the coordinate system constitutes the interval type II fuzzy set. By analyzing the fluctuation range of the historical error standard deviation, the interval type II fuzzy set is constructed, which can more accurately capture the uncertainty of uncertainty itself, and provide a more robust uncertainty model that is closer to the actual working conditions for power system dispatch.

[0020] The specific process of constructing a fuzzy cognitive map using each indicator in the indicator system as a node is as follows: The first-level indicator layer and the second-level indicator layer are set as concept nodes of the fuzzy cognitive graph. The causal relationship topology between the concept nodes is determined based on the historical operation data, and an adjacency matrix representing the topology is constructed. The causal relationship topology includes the horizontal coupling relationship between the secondary indicator nodes and the vertical aggregation relationship between the secondary indicator nodes and their corresponding primary indicator nodes. The adjacency matrix defines whether there is a connection between the nodes.

[0021] Specifically, four primary indicator nodes are identified, representing the final dimensions of the assessment: power supply reliability, system flexibility, green and low-carbon level, and supply-demand coordination. Ten secondary indicator nodes are also identified, serving as the foundational data for the primary indicators. For power supply reliability, these nodes are average outage time, voltage qualification rate, and frequency qualification rate. For system flexibility, these nodes are system net load ramp-up rate, renewable energy absorption rate, and flexibility resource adjustability margin. For green and low-carbon level, these nodes are clean energy power generation penetration rate and carbon emission factor per unit of electricity supply. For supply-demand coordination, these nodes are load peak shaving and valley filling rate and demand-side response participation. A total of fourteen clear conceptual nodes are obtained, which will constitute all the elements of the fuzzy cognitive map. Based on historical operating data and the operating mechanism of the power system, it is determined whether and how these fourteen nodes influence each other. This mainly includes two types of relationships: determining vertical aggregation relationships and determining horizontal coupling relationships. The vertical aggregation relationship is a causal connection from each secondary indicator node (such as average outage time, voltage qualification rate and frequency qualification rate) to its unique corresponding upper-level primary indicator node (in this case, power supply reliability). This relationship reflects the convergence process of basic indicator data to the top-level evaluation dimension. The aforementioned horizontal coupling relationships are the specific causal connections between various secondary indicator nodes, determined based on historical operating data and the power system's operating mechanism. These connections reflect the actual physical constraints and synergistic effects of each link in the system. For example, an increase in demand-side response participation directly promotes the improvement of load peak shaving and valley filling rates and renewable energy absorption rates; while the increase in renewable energy absorption rates increases the challenge of the system's net load ramp-up rate, while significantly reducing the carbon emission factor per unit of electricity supply; furthermore, the adequacy of the adjustability margin of flexibility resources also directly restricts the level of frequency qualification rate. Based on the aforementioned vertical aggregation and horizontal coupling relationships, an adjacency matrix is ​​constructed. This adjacency matrix is ​​a 14×14 matrix, with its rows and columns arranged in a fixed order of the 14 nodes (e.g., 10 secondary indicators first, then 4 primary indicators). This adjacency matrix is ​​used to define whether connections exist between nodes. Line number The element values ​​of the column are set to ensure that there is a slave node. Pointing to node If there is a causal connection (whether vertical aggregation or horizontal coupling), the value of this element is set to 1; if the node... For nodes If no direct causal connection exists, the value of the element is set to 0. For example, if the renewable energy absorption rate is the 5th node and the frequency qualification rate is the 3rd node, then the element in the 5th row and 3rd column of the matrix is ​​1. If the average outage time (1st node) points to the power supply reliability (11th node), then the element in the 1st row and 11th column is 1. This 14×14 adjacency matrix composed of 0s and 1s completely and formally represents the static causal relationship topology. Based on the static causal relationship topology, the skeleton of the fuzzy cognitive graph is formed, which causal connection channels exist between nodes, but the strength of these connections is not quantified. For the complete realization of the fuzzy cognitive graph, it is necessary to assign dynamically changing weight values ​​to these determined connections (i.e., the elements with a value of 1 in the adjacency matrix). These weight values ​​are used to characterize the real-time strength and direction of the mutual influence between nodes under different system operating conditions. That is, the adjacency matrix is ​​the basis for the subsequent generation of the dynamic weight matrix. This application systematically reveals and solidifies the complex physical-mechanism-based coupling relationships between different basic indicators by analyzing how they converge to support the top-level evaluation dimensions (e.g., the constraint of renewable energy consumption on frequency qualification rate). This provides a complete, structured framework for subsequent dynamic weight analysis that conforms to the actual operating logic of the power system, significantly improving the accuracy and authenticity of the evaluation model.

[0022] Specifically, the process of generating the dynamic weight matrix of the fuzzy cognitive map at the current moment is as follows: The Long Short-Term Memory (LSTM) network model is trained offline using the historical operational data to learn and fit the nonlinear coupling relationship pattern between the indicator nodes in the indicator system over time. The time series data of the indicators corresponding to the indicator nodes of the fuzzy cognitive map in the real-time operational data are then input into the trained LSM network model. The LSM network model performs forward inference calculations based on the currently input real-time indicator time series data and outputs a dynamic weight matrix. The offline training is based on all historical time series data corresponding to all 14 indicator nodes (including 10 secondary indicators and 4 primary indicators) of the fuzzy cognitive map in the historical running data. Necessary cleaning, missing value filling and normalization processing are performed on these data to eliminate the differences in the units of measurement between different indicators, and they are organized into sample pairs suitable for time series model training in chronological order. The processed historical data samples are input into the Long Short-Term Memory Network model for iterative training. During the training process, the model learns and fits the complex nonlinear coupling relationship pattern between the 14 index nodes over time through its internal memory units and gating mechanism. That is, the model learns to map the index data sequence of a specific time period to a set of specific weight values ​​that can reflect the strength of the real interaction between the indicators under the working condition. After the training converges, the model parameters are saved to obtain the trained Long Short-Term Memory Network model. Real-time running data is collected to obtain the latest indicator time series data corresponding to the 14 indicator nodes (for example, the sequence window of the current moment and several past time steps is collected, and the window length is consistent with that during training). After performing the same normalization process as in the offline training stage on these real-time data, they are used as input and fed into the trained Long Short-Term Memory Network model. After receiving the real-time indicator time series data, the trained Long Short-Term Memory Network model immediately performs a forward inference calculation (i.e., an inference process without updating the model parameters). Based on the nonlinear coupling relationship pattern that it has learned and solidified in the offline stage, the model automatically infers and calculates a set of weight values ​​that best fits the current real-time working conditions. These weight values ​​are organized into a 14×14 matrix, which is the dynamic weight matrix at the current moment, accurately quantifying the real-time influence intensity and direction of each connection channel on the fuzzy cognitive graph skeleton. By successfully learning and fitting the complex nonlinear coupling pattern between various indicators that dynamically changes with operating conditions from historical data through a long short-term memory network, it can infer and output a dynamic weight matrix that truly reflects the current operating conditions in real-time evaluation based on the latest operating data. This overcomes the limitations of traditional static weights and greatly improves the timeliness, adaptability and accuracy of the evaluation model.

[0023] Specifically, the process of mapping and outputting the fuzzy state vector is as follows: The interval type II fuzzy set is used as the initial fuzzy state of the corresponding uncertainty input index node in the fuzzy cognitive graph, and the initial states of other nodes are set to form the initial fuzzy state vector. The fuzzy state vector is iteratively calculated using the fuzzy cognitive graph iterative reasoning rule based on the dynamic weight matrix. The iterative calculation is based on interval type II fuzzy arithmetic operations, and the fuzzy state value of each index node is updated at each time step. After each iteration, it is determined whether the fuzzy state vector meets the preset convergence condition. The convergence condition is usually that the change in the fuzzy state vector between two consecutive iterations is less than a preset threshold. When the convergence condition is met, the iteration stops, and the fuzzy state vector in the final stable state is used as the output of the update iteration mechanism.

[0024] The initial state is achieved by assigning the interval type-II fuzzy set (containing upper and lower membership functions) generated from the uncertain input indicators to the corresponding concept nodes in the fuzzy cognitive graph (such as nodes related to renewable energy absorption rate or load peak shaving and valley filling rate), serving as the initial fuzzy state for these nodes. Simultaneously, initial states are set for other non-uncertainty indicator nodes in the indicator system (for example, they can be set to interval-degraded fuzzy numbers based on their normalized real-time observations, or set to a zero state). The initial states of all 14 nodes are combined to form a 14-dimensional initial fuzzy state vector. Using the dynamic weight matrix at the current moment and the preset fuzzy cognitive graph iterative reasoning rules, the initial fuzzy state vector is iteratively calculated. This iterative calculation is strictly based on interval type II fuzzy arithmetic operations to ensure that uncertainty is fully preserved and transmitted during the calculation process. In each iteration, the fuzzy state vector at the previous moment is subjected to fuzzy matrix operations with the dynamic weight matrix, and the result is usually processed through a nonlinear activation function to calculate the new fuzzy state values ​​of all 14 nodes at the current moment, thereby obtaining a new fuzzy state vector. After each iteration, a convergence check is immediately performed. The newly generated fuzzy state vector is compared with the fuzzy state vector obtained in the previous iteration, and the change between them is calculated (e.g., the norm of a certain distance or difference between two interval type II fuzzy vectors). It is then determined whether this change is less than a preset convergence threshold. The convergence condition (i.e., the change in the fuzzy state vector between two consecutive iterations is less than the preset threshold) is used to ensure that the calculation process has reached a stable equilibrium state, that is, the fuzzy state values ​​of each node no longer change significantly. If the convergence condition is not met, the newly generated fuzzy state vector is used as the input for the next iteration, and the iterative reasoning steps described in the second paragraph are returned to be executed. If the convergence condition is met, the iteration is stopped immediately. At this time, the fuzzy state vector obtained from the last iteration, which has reached a stable state, is used as the final output result of the update iteration mechanism. By strictly executing interval type II fuzzy arithmetic operations, it is ensured that the uncertainty footprint at the input end can be completely preserved and propagated level by level throughout the entire evaluation network. This process cleverly combines the propagation of this uncertainty with the dynamic time-varying nonlinear coupling relationship. Through iterative calculation until convergence, the final comprehensive impact of the initial uncertainty on all evaluation indicators is clearly presented.

[0025] The specific process for generating the comprehensive supply and demand balance effectiveness assessment value is as follows: Extract the type II fuzzy state values ​​of each interval corresponding to the first-level index layer from the fuzzy state vector; For each extracted primary index, the interval type II fuzzy state value is subjected to defuzzification calculation, which includes type simplification and centroid calculation. The type simplification adopts a type simplification algorithm to simplify the interval type II fuzzy state value into an interval type I fuzzy set, where the interval represents the uncertainty range of its centroid. The centroid calculation involves obtaining the clear evaluation value of the primary indicators by calculating the midpoint of the centroid interval; then, the clear evaluation values ​​of all primary indicators are weighted and summed with their corresponding evaluation weights to calculate the final comprehensive supply and demand balance effectiveness evaluation value. After the update iteration mechanism outputs the stable fuzzy state vector, it first extracts four interval type II fuzzy state values ​​corresponding to the first-level index layer from the 14-dimensional vector.

[0026] For the extracted type II fuzzy state values ​​of the four primary indicators, the type simplification step in the defuzzification calculation is performed one by one. This process adopts a predetermined type simplification algorithm, the Karnik-Mendel algorithm, to simplify each two-dimensional type II fuzzy state value (the region enveloped by the upper and lower membership functions) and calculate a one-dimensional centroid interval (a type I fuzzy set of interval type). This interval (containing a left endpoint and a right endpoint) clearly represents the uncertainty range of the final clear centroid of the indicator. After type simplification, the centroid calculation step continues. For each centroid interval obtained in the previous step (for example, the centroid range of a certain indicator is calculated as [0.7, 0.8]), the uncertainty interval of the indicator is transformed into a unique and clear scalar value by calculating the midpoint value of the interval (i.e., [left endpoint + right endpoint] / 2, such as [0.7 + 0.8] / 2 = 0.75). Thus, a clear evaluation value is obtained for each of the four primary indicators: power supply reliability, system flexibility, green and low-carbon level, and supply-demand coordination. The importance percentage of each indicator is determined using the analytic hierarchy process (AHP). The clear evaluation value is then multiplied by its corresponding evaluation weight. Finally, all four products are weighted and summed to obtain the final total value, which yields the comprehensive supply and demand balance effectiveness evaluation value.

[0027] This application captures the uncertainty footprint of renewable energy and load forecasting through interval type II fuzzy sets. At the same time, it accurately fits the nonlinear coupling relationship between various evaluation indicators as the operating conditions change in real time through fuzzy cognitive graphs empowered by long short-term memory networks. Through interval type II fuzzy operations, it ensures the complete transmission of deep uncertainty in the dynamic evaluation network. The final evaluation results have extremely high robustness, adaptability and accuracy.

[0028] Example 2: This embodiment takes a typical day (e.g., time T on a certain day) during the peak summer season as an example to specifically illustrate the novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic described in this invention. Firstly, based on its historical operating data, a hierarchical evaluation index system as described in this invention was constructed. The first-level index layer includes: power supply reliability, system flexibility, green and low-carbon level, and supply-demand coordination. The second-level index layer includes: average outage time, voltage qualification rate, and frequency qualification rate (corresponding to power supply reliability); system net load ramp-up rate, renewable energy absorption rate, and flexibility resource adjustability margin (corresponding to system flexibility); clean energy power generation penetration rate and carbon emission factor per unit of power supply (corresponding to green and low-carbon level); load peak shaving and valley filling rate, and demand-side response participation (corresponding to supply-demand coordination). At time T, the weather forecast indicates severe convective weather in the afternoon, leading to a significant increase in uncertainty in the renewable energy output forecast data (wind power) and load forecast data (air conditioning load). The assessment process is as follows: Historical prediction error sequences (i.e., the deviation between historical predicted values ​​and historical actual values) of these two indicators were retrieved. Statistical analysis of the historical prediction errors of renewable energy output revealed that the mean error was close to 0, but the standard deviation fluctuated significantly under different weather conditions (e.g., stable weather and drastic weather changes). Through sliding window analysis, the standard deviation fluctuation range (i.e., the lower and upper bounds of the standard deviation) was determined to be [3.0%, 9.5%]. Based on this, a Gaussian function was used as a prototype to construct its interval type II fuzzy set: the lower membership function was constructed based on the lower bound of the standard deviation (3.0%), and the upper membership function was constructed based on the upper bound of the standard deviation (9.5%). The same operation was performed on the load prediction indicator, resulting in a standard deviation fluctuation range of [2.0%, 6.0%], and the corresponding interval type II fuzzy set was constructed. These two fuzzy sets constitute the uncertainty footprint of the uncertain input indicator at time T. Based on the above 14 indicators (4 primary and 10 secondary), a concept node of the fuzzy cognitive graph was constructed. By analyzing historical data and the operation mechanism of the power grid, the vertical aggregation relationship (such as the frequency qualification rate pointing to the power supply reliability) and the horizontal coupling relationship (such as the increase in the renewable energy absorption rate will increase the challenge of the system net load ramping rate) between nodes were determined. These relationships were solidified into a 14×14 static adjacency matrix. Using historical operational data from the past year (including time series data for all 14 indicators), a Long Short-Term Memory (LSTM) network model was thoroughly trained offline, enabling it to learn and fit the nonlinear coupling patterns between the indicators under different operating conditions. Real-time time series data of the 14 indicators up to time T (e.g., the past 30 minutes) (such as a slight decrease in frequency compliance rate or an increase in ramp rate) were input into this trained LSM network model. Since the input real-time data reflected the high uncertainty of an impending severe convective weather event, the LSM network model immediately performed forward inference, outputting a dynamic weight matrix at time T. In this dynamic matrix at time T, the negative impact weight of the renewable energy absorption rate on the frequency compliance rate was significantly enhanced (compared to when the weather was fine); simultaneously, the positive impact weight of the flexibility resource adjustability margin on system flexibility was also significantly increased, which fully aligns with the actual operating logic of the system under this specific condition. A 14-dimensional initial fuzzy state vector is constructed: the two previously generated interval type-2 fuzzy sets representing renewable energy and load uncertainty are assigned to the corresponding secondary indicator nodes such as renewable energy absorption rate, while the states of the other 12 nodes are set as interval-degraded fuzzy numbers based on their normalized real-time observations at time T. Then, the initial fuzzy state vector is iteratively calculated using the dynamic weight matrix generated at time T. This iteration is strictly based on interval type-2 fuzzy arithmetic to ensure the complete transmission of uncertainty in the calculation. After each iteration, it is determined whether the change in the fuzzy state vector is less than a preset threshold (e.g., 0.001). Assuming that this convergence condition is met after the k-th iteration, the iteration is immediately stopped, and a 14-dimensional fuzzy state vector in a stable state is output. Each element in this vector (all are interval type-2 fuzzy sets) reflects the final comprehensive impact of the initial uncertainty on all indicators. From the final fuzzy state vector, the interval type II fuzzy state values ​​corresponding to the four primary indicators—power supply reliability, system flexibility, green and low-carbon level, and supply-demand coordination—are extracted, and defuzzification calculations are performed on each of these four values. Taking system flexibility as an example, the Karnik-Mendel algorithm is used for type simplification. Due to the large input uncertainty at time T and the significant influence of dynamic weights, its centroid interval is calculated as [0.68, 0.80]. Centroid calculation is performed on this centroid interval, i.e., the midpoint of the interval (0.68 + 0.80) / 2 = 0.74 is calculated, resulting in a clear evaluation value for system flexibility. The same operation is performed on the other three primary indicators, assuming that the clear evaluation values ​​obtained are 0.90, 0.85, and 0.78, respectively. Finally, based on the preset (determined by the analytic hierarchy process) primary indicator evaluation weights (e.g., power supply reliability 0.4, system flexibility 0.3, green and low-carbon 0.15, supply and demand coordination 0.15), a weighted sum was calculated to obtain the final comprehensive evaluation value: (0.90×0.4)+(0.74×0.3)+(0.85×0.15)+(0.78×0.15)=0.8265. Ultimately, the dispatch center obtained a comprehensive supply and demand balance effectiveness evaluation value of 0.8265 at time T. Simultaneously, the dispatcher also noted that the centroid interval [0.68, 0.80] for system flexibility was relatively wide, which intuitively indicates that this effectiveness is most affected by current uncertainties and requires close attention in subsequent dispatching. This embodiment fully verifies the feasibility and superiority of the present invention in actual power grid operation scenarios, which integrates the processing of highly uncertain and dynamically coupled relationships and obtains robust evaluation results.

[0029] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic. Includes, characterized in that: Collect historical and real-time operational data of the new power system, and construct a supply and demand balance effectiveness evaluation index system based on the historical operational data; For the uncertain input indicators in the indicator system, an interval type II fuzzy membership function is used for fuzzification to generate an interval type II fuzzy set representing its uncertainty footprint; a fuzzy cognitive graph is constructed with each indicator in the indicator system as a node, and the real-time running data is input into the trained long short-term memory network model to generate the dynamic weight matrix of the fuzzy cognitive graph at the current time. The dynamic weight matrix is ​​used to represent the dynamic time-varying nonlinear coupling relationship between each indicator node. The interval type II fuzzy set and the dynamic weight matrix generated at the current time are input into the update iteration mechanism to map and output the fuzzy state vector; The fuzzy state vector is defuzzified to generate a supply-demand balance effectiveness evaluation value.

2. The novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic according to claim 1, characterized in that, The specific details of constructing the supply and demand balance effectiveness evaluation index system based on the historical operational data are as follows: Analyze the historical operational data to construct a hierarchical evaluation indicator system that includes a primary indicator layer and a secondary indicator layer; The primary indicator layer includes power supply reliability, system flexibility, green and low-carbon level, and supply and demand coordination. The indicators in the secondary indicator layer correspond to the indicators in the primary indicator layer. The indicators corresponding to the power supply reliability include average outage time, voltage qualification rate, and frequency qualification rate; the indicators corresponding to the system flexibility include system net load ramp rate, renewable energy absorption rate, and flexibility resource adjustability margin; and the indicators corresponding to the green and low-carbon level include clean energy power generation penetration rate and carbon emission factor per unit of power supply. The indicators corresponding to the supply and demand coordination include the load peak shaving and valley filling rate and the amount of demand-side response participation.

3. The novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic according to claim 2, characterized in that, The specific process for generating the interval type-II fuzzy set representing its uncertainty footprint is as follows: The uncertainty input indicators include renewable energy output forecast data and load forecast data; Acquire and analyze the historical prediction error sequence of the aforementioned uncertainty input indicators in the historical operating data, wherein the historical prediction error sequence consists of the deviation between the historical predicted value and the historical actual value; Statistical analysis was performed on the historical prediction error sequence to determine its probability distribution characteristics. Based on the distribution characteristics, the interval type II fuzzy membership function is constructed. The interval type II fuzzy membership function is jointly defined by the upper membership function and the lower membership function to form the interval type II fuzzy set.

4. The novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic according to claim 2, characterized in that, The specific process of constructing a fuzzy cognitive map using each indicator in the indicator system as a node is as follows: The first-level indicator layer and the second-level indicator layer are set as concept nodes of the fuzzy cognitive graph. The causal relationship topology between the concept nodes is determined based on the historical operation data, and an adjacency matrix representing the topology is constructed. The causal relationship topology includes the horizontal coupling relationship between the secondary indicator nodes and the vertical aggregation relationship between the secondary indicator nodes and their corresponding primary indicator nodes. The adjacency matrix defines whether there is a connection between nodes, and a fuzzy cognitive graph is constructed based on the adjacency matrix.

5. A novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic according to claim 1, characterized in that, The specific process of generating the dynamic weight matrix of the fuzzy cognitive map at the current moment is as follows: The Long Short-Term Memory (LSTM) network model is trained offline using the historical operational data to learn and fit the nonlinear coupling relationship pattern between the indicator nodes in the indicator system over time. The time series data of the indicators corresponding to the indicator nodes of the fuzzy cognitive map in the real-time operational data are then input into the trained LSM network model. The LSM network model performs forward inference calculations based on the currently input real-time indicator time series data and outputs a dynamic weight matrix.

6. The novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic according to claim 1, characterized in that, The specific process of mapping and outputting the fuzzy state vector is as follows: The interval type II fuzzy set is used as the initial fuzzy state of the corresponding uncertainty input index node in the fuzzy cognitive graph, and the initial states of other nodes are set to form the initial fuzzy state vector. The fuzzy state vector is iteratively calculated using the fuzzy cognitive graph iterative reasoning rule based on the dynamic weight matrix. The iterative calculation is based on interval type II fuzzy arithmetic operations, and the fuzzy state value of each index node is updated at each time step. After each iteration, it is determined whether the fuzzy state vector satisfies a preset convergence condition. The convergence condition is usually that the change in the fuzzy state vector between two consecutive iterations is less than a preset threshold. When the convergence condition is met, the iteration stops, and the fuzzy state vector in the final stable state is used as the output of the update iteration mechanism.

7. A novel power system supply and demand balance effectiveness evaluation method based on fuzzy logic according to claim 2, characterized in that, The specific process for generating the comprehensive supply and demand balance effectiveness assessment value is as follows: Extract the type II fuzzy state values ​​of each interval corresponding to the first-level index layer from the fuzzy state vector; For each extracted primary index, the interval type II fuzzy state value is subjected to defuzzification calculation, which includes type simplification and centroid calculation; the type simplification adopts a type simplification algorithm to simplify the interval type II fuzzy state value into an interval type I fuzzy set; The centroid calculation involves obtaining the clear evaluation value of the primary indicators by calculating the midpoint of the centroid interval; and then weighting and summing the clear evaluation values ​​of all primary indicators with their corresponding evaluation weights to obtain the final comprehensive supply and demand balance effectiveness evaluation value.

Images