A method and system for evaluating and early warning of cascading failure probability of integrated energy system

Abstract

This invention discloses a method for assessing and warning of cascading failure probabilities in integrated energy systems, belonging to the technical field of safe and stable operation of integrated energy systems. Addressing the problems in existing technologies such as fuzzy characterization of cascading failure evolution mechanisms, lack of dynamic transition patterns in operating states, static and fixed failure probability assessments, and the inability to adaptively adjust warning thresholds, this invention uses a weighted clustering algorithm with safety weight correction to divide the discrete operating state space. Based on this state space, a Markov state transition probability matrix is ​​constructed, and dynamic updates are achieved by combining a sliding time window and a forgetting factor. The correlation strength coefficient of multi-energy flow coupled failure propagation paths is quantified and incorporated into the Markov chain iterative solution to the cumulative probability of cascading failures. Finally, a two-dimensional dynamic risk model is constructed, and an adaptive fuzzy PID algorithm is used to achieve real-time correction of warning thresholds and hierarchical warnings. Ultimately, this provides quantitative basis and technical support for the safe operation and cascading prevention of integrated energy systems.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system safety and stable operation technology, and in particular to a method and system for assessing and warning of cascading failures in integrated energy systems. Background Technology

[0002] Integrated energy systems, through the interconnection of power grids, gas pipelines, and heating pipelines, combined with the coordinated control of photovoltaic, wind power, energy storage, and multi-energy conversion equipment, achieve complementary optimization of multiple energy flows and improve energy utilization efficiency. However, the multi-energy coupling links within the system, such as electricity-to-gas conversion, gas-to-power generation, and thermoelectric decoupling, significantly increase operational complexity. Local faults in a single device or pipeline can easily propagate cascading through coupling paths, triggering a chain reaction of failures. Therefore, accurately sensing the system's operating status, quantifying the probability of cascading failures, and achieving dynamic early warning are key technical requirements for ensuring the safe and reliable operation of integrated energy systems.

[0003] Currently, most assessment and early warning technologies for cascading failures in integrated energy systems rely on traditional single-energy system analysis methods, which fail to adequately adapt to the complex characteristics of multi-energy coupling and mainly suffer from the following problems.

[0004] First, the accuracy of operational state division is insufficient. Existing methods mostly use manually defined thresholds or equal-weighted K-means clustering for state division, without considering the different impacts of different feature parameters on system safety, and without assigning higher weights to critical operational parameters close to the safety boundary. As a result, the divided state space is difficult to accurately represent the actual fault risk level of the system, and the state completeness and discriminability are insufficient, which cannot provide a reliable basis for subsequent fault evolution analysis.

[0005] Secondly, the fault evolution model has poor dynamic adaptability. The mainstream method uses a Markov state transition probability matrix based on fixed historical data statistics, which fails to effectively cope with the influence of time-varying factors such as load fluctuations, random changes in new energy output, and equipment aging on the state transition law. The model parameters are seriously out of sync with real-time operating conditions, causing the fault probability calculation results to deviate from the actual operating state.

[0006] Third, the propagation characteristics of cascading failures have not been quantified. Existing fault probability calculations do not distinguish the differences in propagation capabilities of different coupling paths such as electric-gas and gas-heat, and lack a quantitative mechanism for fault correlation strength. They simply sum up the probabilities of each fault state, which cannot truly reflect the diffusion law of cascading failures under multi-energy flow coupling, resulting in distorted evaluation results.

[0007] Fourth, the early warning mechanism is rigid and lacks adaptability. Existing early warning thresholds are mostly set based on engineering experience. Under low-risk conditions, they are prone to generating a large number of false alarms, increasing the maintenance burden. Under high-risk critical conditions, early warnings are often delayed, failing to allow sufficient time for fault isolation. It is difficult to balance the accuracy and timeliness of early warnings, and thus cannot meet the needs of online real-time management and control.

[0008] The aforementioned deficiencies make it difficult for existing technologies to achieve accurate perception, dynamic assessment, and timely early warning of cascading failures in integrated energy systems, severely hindering the improvement of system safety. Therefore, how to construct a cascading failure probability assessment and early warning method that can adapt to multi-energy coupling characteristics, possess dynamic adaptive capabilities, and quantify fault propagation paths has become an urgent technical problem to be solved in this field. Summary of the Invention

[0009] To address the current problems of ambiguous characterization of cascading failure evolution mechanisms, lack of dynamic transition patterns in operating states, static and fixed failure probability assessments, and non-adaptive optimization of warning thresholds in integrated energy systems, this invention proposes a method and system for cascading failure probability assessment and early warning in integrated energy systems. It achieves refined discrete partitioning of the operating state space through an improved K-means clustering algorithm with safety weight correction, constructs a discrete-time Markov model adapted to multi-energy coupling, and designs a sliding window and forgetting factor to dynamically update the state transition probability matrix. Simultaneously, it quantifies the correlation strength coefficient of multi-energy flow coupling failure propagation paths based on the analytic hierarchy process (AHP) and incorporates it into the Markov chain iterative solution to the cumulative probability of cascading failures. Finally, it constructs a dual-dimensional dynamic risk model of economic loss and energy supply interruption, and employs an adaptive fuzzy PID algorithm to achieve real-time correction of warning thresholds and tiered early warning. This invention effectively reduces the error in cascading failure probability assessment, significantly shortens the early warning response time, and achieves dynamic adaptation to the multi-energy coupling and random fluctuation characteristics of integrated energy systems.

[0010] On the one hand, a method for assessing and warning of the probability of cascading failures in a comprehensive energy system is provided, including: Obtain the characteristic parameters of the integrated energy system during operation, and then standardize the characteristic parameters; Based on the weighted clustering algorithm, the standardized feature parameters are divided to obtain a discrete running state space; Based on the discrete operating state space, the historical operating data of the integrated energy system is mapped to a state, the transition frequency between each state is counted, a Markov state transition probability matrix is ​​constructed, and the state transition probability matrix is ​​iteratively updated using real-time operating data, combined with a sliding time window and a forgetting factor. Typical cascading failure propagation paths of an integrated energy system are obtained, the correlation strength coefficient of each path is calculated, and the correlation strength coefficient is incorporated into the update calculation of the Markov state transition probability matrix to obtain the cumulative probability of cascading failures occurring in the system in the future period. The real-time risk value is calculated based on the cumulative probability. An adaptive fuzzy PID algorithm is used to correct the preset warning threshold in real time. The corrected threshold is compared with the real-time risk value to obtain the warning classification result.

[0011] Furthermore, the integrated energy system includes an electric system, a gas system, a heating system, and an energy storage system.

[0012] Furthermore, based on a weighted clustering algorithm, the standardized feature parameters are divided to obtain a discrete operating state space, specifically: Calculate the degree to which each characteristic parameter deviates from the safe median value after standardization to obtain the safe deviation of each parameter; The initial safety weights of each feature parameter are determined based on the analytic hierarchy process (AHP), and the initial safety weights are corrected according to the safety deviation to obtain the corrected weights. Based on the corrected weights, the weighted Euclidean distance between each data point is calculated to obtain a similarity measure; Based on the similarity metric, the K-means clustering algorithm is used to iteratively cluster the standardized feature parameters. The clustering process is terminated when the change in the cluster center is less than a preset threshold or the maximum number of iterations is reached, and a discrete running state space is output.

[0013] Furthermore, based on the discrete operating state space, the historical operating data of the integrated energy system are mapped to states, the transition frequencies between states are statistically analyzed, and a Markov state transition probability matrix is ​​constructed, specifically as follows: Each set of feature parameters in the historical operation data is mapped to the discrete operation state space to obtain a discrete state sequence; Count the number of times a state transitions from any state to any other state in the historical state sequence; Based on the number of transitions obtained from statistics, calculate the initial transition probability between each state and construct the initial state transition probability matrix.

[0014] Furthermore, the state transition probability matrix is ​​iteratively updated using real-time runtime data, combined with a sliding time window and a forgetting factor; specifically: Set a fixed-duration sliding time window and count the number of transitions of the system between discrete states in real time within the window; Introducing a forgetting factor, the updated state transition probability matrix is ​​calculated using a weighted smoothing update formula, where the weighted smoothing update formula is: , in, Forgetting factor, This represents the number of state transitions from state i to state j within the window. This is the old probability from the previous period.

[0015] Furthermore, typical cascading failure propagation paths of the integrated energy system are obtained, the correlation strength coefficient of each path is calculated, and the correlation strength coefficient is incorporated into the update calculation of the Markov state transition probability matrix to obtain the cumulative probability of cascading failures occurring in the system in the future time period, specifically: Typical cascading failure propagation paths of integrated energy systems are obtained. The importance weight of each coupling link in each path is determined by the analytic hierarchy process (AHP). The coupling strength of each link is determined by combining equipment coupling test data. The correlation strength coefficient of each failure propagation path is calculated in a comprehensive manner. Initialize the current state probability vector, and use the updated state transition probability matrix to perform multi-step iterative prediction to obtain the probability of the system being in each discrete state at each future time. A preset fault state set is defined, and the probability of the system being in the fault state set at each future time is accumulated and corrected by combining the correlation strength coefficient to obtain the cumulative probability of the system experiencing a cascading failure in the future time period.

[0016] Furthermore, a real-time risk value is calculated based on the cumulative probability, and an adaptive fuzzy PID algorithm is used to correct the preset warning threshold in real time. The corrected threshold is then compared with the real-time risk value to obtain the warning classification result, specifically: Construct a risk quantification model that includes economic losses and energy supply disruptions, and combine it with the cumulative probability calculation system to calculate the real-time risk value. Calculate the risk deviation between the real-time risk value and the preset benchmark warning threshold; Based on the aforementioned risk deviation, the correction amount of the PID control parameters is output in real time through adaptive fuzzy PID fuzzy inference; The correction amount of the warning threshold is calculated based on the corrected PID parameters, and the correction amount is superimposed on the baseline warning threshold to obtain the dynamic warning threshold. The dynamic warning threshold is compared with the real-time risk value, and the corresponding warning result is obtained according to the preset classification rules.

[0017] On the other hand, a comprehensive energy system cascading failure probability assessment and early warning system is provided, including: The state space partitioning module is used to acquire the characteristic parameters of the integrated energy system during operation and perform standardization processing. Based on the weighted clustering algorithm, the standardized characteristic parameters are partitioned to obtain the discrete operating state space. The matrix construction and update module is used to perform state mapping on the historical operating data of the integrated energy system based on the discrete operating state space, count the transition frequency between each state, construct a Markov state transition probability matrix, and dynamically update the state transition probability matrix using real-time operating data, combined with a sliding time window and a forgetting factor. The failure probability calculation module is used to obtain typical cascading failure propagation paths of the integrated energy system, calculate the correlation strength coefficient of each path, and integrate the correlation strength coefficient into the iterative calculation of the Markov chain to solve the cumulative probability of cascading failures occurring in the system in the future period. The dynamic risk assessment and early warning module is used to calculate the real-time risk value based on the cumulative probability, use an adaptive fuzzy PID algorithm to correct the preset early warning threshold in real time, compare the corrected dynamic threshold with the real-time risk value, and output the graded early warning result.

[0018] In another aspect, a computer device is also provided, including a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, characterized in that, when the processor executes the program, it performs the method described in the first aspect.

[0019] In another aspect, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, performs the method described in the first aspect.

[0020] The above technical solution has the following advantages or beneficial effects: Compared with the prior art, the present invention has the following beneficial effects: (1) Regarding the classification of operating states, this invention addresses the shortcomings of traditional equal-weight clustering or manual threshold classification methods in characterizing the criticality of parameter safety by proposing an improved K-means clustering algorithm with safety weight correction. By introducing a safety deviation index and allocating initial safety weights based on the analytic hierarchy process, critical operating parameters close to the fault boundary are weighted and amplified, causing the clustering results to tilt towards high-risk regions. Compared with existing technologies, the discrete operating state space obtained by this invention has the characteristics of mutual exclusion, completeness, and risk distinguishability, significantly improving the accuracy of operating state identification, providing a high-quality state foundation for subsequent Markov modeling, and improving the accuracy of fault risk characterization in state classification.

[0021] (2) Regarding state transition modeling, this invention addresses the problem that a fixed transition probability matrix cannot adapt to the time-varying characteristics of the system by designing an adaptive update mechanism that combines a sliding time window with a forgetting factor. By statistically analyzing the number of state transitions within the window in real time and introducing a forgetting factor to balance historical patterns with real-time changes, online smooth updates of the transition probability matrix are achieved. Compared with existing traditional Markov models that use static matrices, the model of this invention can dynamically respond to time-varying factors such as load fluctuations and random power output from new energy sources, effectively reducing state transition prediction errors and significantly improving the matching degree between the model and real-time operating conditions.

[0022] (3) Regarding the quantification of cascading failure probability. This invention addresses the problem of distorted failure probability calculations caused by the failure to quantify the differences in multi-energy coupling propagation in traditional methods. It proposes a method for calculating the correlation strength coefficient based on the analytic hierarchy process (AHP) and the fusion of equipment coupling test data. By analyzing typical failure propagation paths and decomposing the multi-energy coupling links in each path, the weight and strength of each coupling link are quantified, and the correlation strength coefficient is incorporated as a weighting factor into the Markov chain iterative solution. Compared with existing methods that simply superimpose failure state probabilities, this invention truly reflects the diffusion law of failures along different paths under multi-energy coupling, improves the accuracy of cascading failure probability calculations, and effectively overcomes the problem of distorted evaluation results.

[0023] (4) Regarding dynamic early warning, this invention addresses the shortcomings of rigid static early warning thresholds by constructing a dual-dimensional risk model of economic loss and energy supply interruption, and introducing an adaptive fuzzy PID algorithm to achieve real-time correction of the early warning threshold. Using risk deviation as input, the PID parameters are dynamically adjusted through fuzzy inference, forming an adaptive optimization logic where the higher the risk, the lower the threshold. Compared with traditional early warning methods using fixed thresholds, this invention can reduce the false alarm rate under low-risk conditions and shorten the early warning response time under high-risk critical conditions, achieving synergistic optimization of early warning accuracy and timeliness, and reserving sufficient time for fault interruption.

[0024] In summary, this invention can effectively reduce the probability assessment error of cascading failures in integrated energy systems, significantly shorten the early warning response time, and has good adaptability, enabling it to cope with the complex operating characteristics of systems under multi-energy coupling and random fluctuation conditions. It provides reliable quantitative basis and technical support for early warning, cascading interruption, and safety control of cascading failures, and can be widely applied to online safe operation and management scenarios of various integrated energy systems at the park and regional levels. Attached Figure Description

[0025] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0026] Figure 1 This is a flowchart of the method according to Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the system structure of Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of the structure of a computer device according to Embodiment 3 of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0028] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0029] Example 1 This embodiment provides a method for assessing and warning of the probability of cascading failures in an integrated energy system. Figure 1 This is an overall flowchart of the integrated energy system cascading failure probability assessment and early warning method according to Embodiment 1 of the present invention, as follows: Figure 1 As shown, the method includes the following steps: S101: Obtain the characteristic parameters of the integrated energy system during operation and standardize the characteristic parameters; S102: Based on the weighted clustering algorithm, the standardized feature parameters are divided to obtain a discrete operating state space; S103: Based on the discrete operating state space, the historical operating data of the integrated energy system is mapped to a state, the transition frequency between each state is counted, a Markov state transition probability matrix is ​​constructed, and the state transition probability matrix is ​​iteratively updated using real-time operating data, combined with a sliding time window and a forgetting factor. S104: Obtain typical cascading failure propagation paths of the integrated energy system, calculate the correlation strength coefficient of each path, and integrate the correlation strength coefficient into the update calculation of the Markov state transition probability matrix to obtain the cumulative probability of cascading failures occurring in the system in the future period. S105: Calculate the real-time risk value based on the cumulative probability, use an adaptive fuzzy PID algorithm to correct the preset warning threshold in real time, compare the corrected threshold with the real-time risk value, and obtain the warning classification result.

[0030] Traditional equal-weight clustering cannot characterize the safety criticality of parameters and has low state partitioning accuracy. Therefore, this embodiment introduces a parameter safety deviation index to quantify the degree of deviation between feature parameters and the intermediate value of safe operation; it allocates initial safety weights based on the analytic hierarchy process and achieves weighted amplification of critical parameters by correcting the weights through deviation; it constructs a weighted Euclidean distance to replace the traditional Euclidean distance, improves the K-means clustering algorithm, and partitions a mutually exclusive, complete, and risk-distinguishable discrete operating state space.

[0031] Specifically, in step S101: the integrated energy system includes an electric system, a gas system, a heating system, and an energy storage system. In this embodiment, eight types of core parameters that are sensitive to the safety of the integrated energy system, can be collected in real time, and can characterize the operating status of multiple energy flows are selected to form a feature parameter set. Specifically, this includes: the bus voltage of the power system. (kV), branch transmission power (MW), system frequency (Hz); Pipeline pressure of the gas system (MPa), gas flow rate (m³ / h); water supply temperature of the heating system (°C), return water temperature (°C); State of charge of the energy storage system (SOC, %)

[0032] To eliminate the differences in the dimensions of different parameters, the min-max standardization method is used to map each feature parameter to the interval [0,1]. The standardized model is shown in equation (1): (1); in, For the first The first feature parameter The standardized results of the group sampling data For the first The first feature parameter Group of original sampling data, For the first The lower limit for the safe operation of each parameter is determined by the system design specifications. For the first The safe operating limit for each parameter is determined by the system design specifications.

[0033] Traditional equal-weight clustering cannot characterize the safety criticality of parameters and has low state partitioning accuracy. Therefore, this embodiment introduces a safety deviation index to correct the feature weights and uses weighted Euclidean distance as a similarity measure. Specifically, step S102 includes the following sub-steps: S1021, Calculate the safety deviation. For the standardized characteristic parameters... The degree of deviation from the safety median is defined as the safety deviation, and the calculation formula is: (2); Where 0.5 is the safe median value of the standardized range. A larger value indicates that the parameter is closer to the fault boundary. By calculating the deviation of the parameter from the safe median value, the weight of the critical parameter is amplified, causing the clustering results to tilt towards the fault critical state.

[0034] S1022, Correct feature weights. The initial safety weights for each feature parameter are determined using the analytic hierarchy process (AHP). Normalization satisfies The initial weights are corrected based on the safety deviation, and the correction formula is shown in equation (3): (3); Corrected weights It still satisfies the normalization constraint.

[0035] S1023: Calculate the weighted Euclidean distance. Based on the corrected weights. We construct a weighted Euclidean distance as a similarity measure between samples to replace the traditional equal-weight distance. We iteratively optimize the cluster centers to complete the state space partitioning. The calculation formula is shown in Equation (4): (4); in, For the first The first feature parameter The standardized results of the group sampling data For the first Group and No. The weighted distance of the data sets indicates that the smaller the distance, the more similar the operating states.

[0036] S1024: Clustering Iteration and State Space Generation.

[0037] Set the number of clusters Convergence threshold Maximum number of iterations N max =100; Eight sets of standardized data were randomly selected as the initial cluster centers. Iteratively assign data to the nearest cluster center, and calculate the weighted mean of the cluster to update the center; When the change in cluster centers The iteration terminates when the maximum number of iterations is reached. Output discrete running state space The parameter threshold range and risk level for each state are marked.

[0038] Through the three-step improvements of dimensional standardization, safety weight correction, and weighted distance measurement, this invention achieves a refined division of the operating state. The resulting state is strongly correlated with the system failure risk and has the characteristics of safety orientation, adaptive weighting, and convergence stability.

[0039] A fixed transition probability matrix cannot adapt to the time-varying operating characteristics of a comprehensive energy system, resulting in insufficient model dynamism. Therefore, this embodiment constructs a discrete-time Markov chain for the comprehensive energy system, which aligns with the system's characteristic of no aftereffects. The operating state of the comprehensive energy system at a certain moment is only related to the state at the previous moment and is independent of earlier historical states, satisfying the assumption of no aftereffects of the discrete-time Markov chain (DTMC), as shown in equation (5): (5); in, The discrete time step is 1 minute in this embodiment, consistent with the system data sampling frequency. for System operating state variables at all times Discrete state space Any subset of states in the array, where K is the total number of states. For the system from state Transfer to The one-step transition probability, and all one-step transition probabilities constitute dimensional state transition probability matrix And satisfy , This model uses the transition probability matrix to iteratively solve for the probabilities of future multi-step states, providing a mathematical framework for calculating the probability of cascading failures.

[0040] Specifically, step S103 includes the following sub-steps: S1031, Initial transition probability matrix construction. Based on the system's historical operating data and fault records for the past 3 years, each set of historical data is mapped to the discrete state space obtained in step S102 to obtain a historical state sequence, and the effective number of transitions from state i to state j is counted. Calculate the initial transition probability according to equation (6): (6); This represents the initial state transition probability of the system transitioning from state i to state j, and is ultimately generated. Initial transition probability matrix P init Verify that the sum of probabilities for each row is 1, remove abnormal probability values, and correct them.

[0041] S1032: Real-time data preprocessing. The real-time acquisition unit acquires system operation data at a frequency of 1 minute, and uses the 3σ criterion to eliminate data that meets the requirements. Abnormal data, among which For the first Historical mean and standard deviation of each parameter; For missing values, linear interpolation is used to complete the missing data, as shown in equation (7): (7); After preprocessing is completed, repeat the standardization process in step S101.

[0042] S1033: Sliding window statistics and matrix update. Setting the sliding window size. For each hour, there are 1440 sets of sampled data. The matrix is ​​updated once per hour, and the number of state transitions within the statistical window is counted. Determine the current real-time status of the system. Introducing a forgetting factor Perform a weighted smooth update according to equation (8): (8) in, The real-time updated state transition probability of the system from state i to state j This represents the number of state transitions from state i to state j within the window. The probability is the old probability from the previous period, and α is the forgetting factor. In this embodiment, the preferred value is... =0.7, used to balance the impact of historical statistical patterns and recent real-time changes on the model.

[0043] Finally, output the updated transition probability matrix P. update The core parameters of the model are updated synchronously by the matrix storage unit to support the next round of fault probability calculation. This step realizes online, real-time and smooth updates of the transition probability matrix without parameter jumps, and can effectively adapt to the time-varying characteristics of the system.

[0044] In step S104, typical cascading failure propagation paths of the integrated energy system are obtained, the correlation strength coefficient of each path is calculated, and the correlation strength coefficient is incorporated into the update calculation of the Markov state transition probability matrix to obtain the cumulative probability of cascading failures occurring in the system in the future time period. This specifically includes the following sub-steps: S1041: Fault Propagation Path Analysis. This embodiment takes three typical cascading fault paths L={L1,L2,L3} of an integrated energy system as an example for analysis, breaking down the multi-energy coupling links of each path, wherein: Power equipment failure → gas turbine unit shutdown → interruption of heat supply; Gas pipeline failure → gas unit shutdown → power shortage → insufficient heat supply; Energy storage failure → power fluctuation → gas turbine frequency regulation overload → heat supply fluctuation.

[0045] S1042: Calculation of Association Strength Coefficient. For each fault propagation path, the association strength is quantified using the analytic hierarchy process (AHP) combined with equipment coupling test data. The formula for calculating the single-path association strength coefficient is shown in equation (9): (9); in, For the first The association strength coefficient of each fault path, with a value of For path The number of coupling links; For the first The weights of each coupling link are determined by the analytic hierarchy process (AHP) and satisfy the following conditions: For the first The coupling strength of each component's equipment is determined experimentally. Based on experimental experience, this embodiment recommends: electro-electric coupling. Gas-heat coupling Electric-energy storage coupling .

[0046] S1043: Iterative solution for cascading failure probabilities. Let the initial time be... The system is in a state The initial state probability vector π0 is a 1×8 dimensional vector, where the corresponding The position is 1, and the rest are 0.

[0047] Using the updated transition probability matrix P update Perform multiple iterations according to equation (10) to calculate the state probability vector for the future T=24 hours: (10); , for The system is in a state at any given time. The probability of.

[0048] Define fault state set Combined with the correlation strength coefficient Calculate the cumulative probability of cascading failures within 24 hours. As shown in equation (11): (11); Understandably, a fault state set is a set of dangerous states that, once entered, are determined to have caused a fault or are about to trigger a cascading failure.

[0049] Static warning thresholds are prone to false alarms and delayed warnings, resulting in poor warning performance. To address this, this embodiment establishes a dual-dimensional risk assessment model based on economic losses and energy supply interruptions, replacing a single probability risk indicator. Using the deviation between the risk value and the benchmark threshold as input, fuzzy logic rules are used to correct PID control parameters in real time. The dynamic output of the threshold correction amount forms an adaptive optimization logic where higher risks result in lower thresholds, and is complemented by a four-level graded warning standard.

[0050] Specifically, in step S105: the real-time risk value is calculated based on the cumulative probability, the preset warning threshold is corrected in real time using an adaptive fuzzy PID algorithm, and the corrected threshold is compared with the real-time risk value to obtain the warning classification result, including the following sub-steps: S1051: Calculation of dynamic risk value in two dimensions. Integrating the two risk dimensions of economic loss and energy supply interruption, the real-time risk value R of the system is calculated according to formula (12): (12); in, As a weight for economic losses, For supply interruption weight, satisfy To estimate the economic loss (in ten thousand yuan) due to the failure, including losses from equipment repair, power outages, gas outages, and heating outages, The number of users (in ten thousand) affected by the energy supply interruption is determined by the system's energy supply range.

[0051] S1052: Adaptive fuzzy PID threshold correction. Define risk deviation. ,in Using 10,000 yuan as the baseline warning threshold, PID parameters are output in real time through fuzzification, fuzzy inference, and defuzzification. , , The correction amount. The PID control unit calculates the threshold correction amount according to equation (13) based on the corrected PID parameters. : (13); The correction amount is superimposed on the baseline warning threshold to obtain the dynamic warning threshold. : (13); S1053: Graded Early Warning Judgment. The early warning judgment unit matches four early warning levels based on the ratio of the dynamic threshold to the real-time risk value, obtaining the graded early warning result for the integrated energy system fault, where: Blue alert (low risk, normal operation and maintenance): ; Yellow alert (medium risk, close monitoring): ; Orange alert (higher risk, proactive measures required): ; Red Alert (High Risk, Emergency Response): .

[0052] The system ultimately outputs corresponding early warning signals and prevention and control suggestions, achieving graded early warning of faults in the integrated energy system. This adaptive fuzzy PID threshold optimization mechanism does not require the establishment of a precise risk-threshold mathematical model. It can dynamically adjust the early warning sensitivity according to the real-time risk level, reduce the false alarm rate under low-risk conditions, and provide early warnings under high-risk critical conditions, allowing sufficient time for fault prevention and control.

[0053] In summary, this embodiment constructs a complete technical closed loop from state perception and probability assessment to dynamic early warning through the synergistic effect of four innovative mechanisms: refined state division, dynamic model updating, propagation path quantification, and adaptive threshold optimization. It effectively solves the adaptability problem of existing technologies in multi-energy coupling scenarios and provides reliable technical support for early warning and safety control of cascading failures in integrated energy systems.

[0054] Example 2 This embodiment provides a system for assessing and warning of the probability of cascading failures in an integrated energy system, such as... Figure 2 As shown, it includes: The state space partitioning module is used to acquire the characteristic parameters of the integrated energy system during operation and perform standardization processing. Based on the weighted clustering algorithm, the standardized characteristic parameters are partitioned to obtain the discrete operating state space. The matrix construction and update module is used to perform state mapping on the historical operating data of the integrated energy system based on the discrete operating state space, count the transition frequency between each state, construct the Markov state transition probability matrix, and dynamically update the state transition probability matrix using real-time operating data, combined with a sliding time window and a forgetting factor. The failure probability calculation module is used to obtain typical cascading failure propagation paths of integrated energy systems, calculate the correlation strength coefficient of each path, and integrate the correlation strength coefficient into the iterative calculation of the Markov chain to solve the cumulative probability of cascading failures occurring in the system in the future period. The dynamic risk assessment and early warning module is used to calculate the real-time risk value based on the cumulative probability, use an adaptive fuzzy PID algorithm to correct the preset early warning threshold in real time, compare the corrected dynamic threshold with the real-time risk value, and output the graded early warning result.

[0055] It should be noted that each module in this embodiment corresponds one-to-one with each step in Embodiment 1, and their specific implementation processes are the same, so they will not be repeated here.

[0056] Example 3 This embodiment provides a computer device, such as... Figure 3 As shown, the system includes a computer-readable storage medium 1003, a processor 1001, a communication interface 1002, and a computer program stored on the computer-readable storage medium 1003 and executable on the processor 1001. The processor 1001, communication interface 1002, and computer-readable storage medium 1003 can be connected via a bus or other means. The communication interface 1002 is used to receive and send data. When the processor 1001 executes the program, it implements the steps in the domain-adaptive UAV fault diagnosis method described in Embodiment 1 above.

[0057] Those skilled in the art will recognize that the units and algorithm steps described in connection with the various examples of this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0058] Example 4 This embodiment also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the method described in Embodiment 1.

[0059] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for assessing and warning of cascading failures in an integrated energy system, characterized in that, include: Obtain the characteristic parameters of the integrated energy system during operation, and then standardize the characteristic parameters; Based on the weighted clustering algorithm, the standardized feature parameters are divided to obtain a discrete running state space; Based on the discrete operating state space, the historical operating data of the integrated energy system is mapped to a state, the transition frequency between each state is counted, a Markov state transition probability matrix is ​​constructed, and the state transition probability matrix is ​​iteratively updated using real-time operating data, combined with a sliding time window and a forgetting factor. Typical cascading failure propagation paths of an integrated energy system are obtained, the correlation strength coefficient of each path is calculated, and the correlation strength coefficient is incorporated into the update calculation of the Markov state transition probability matrix to obtain the cumulative probability of cascading failures occurring in the system in the future period. The real-time risk value is calculated based on the cumulative probability. An adaptive fuzzy PID algorithm is used to correct the preset warning threshold in real time. The corrected threshold is compared with the real-time risk value to obtain the warning classification result.

2. The method for assessing and warning of cascading failures in an integrated energy system according to claim 1, characterized in that, The integrated energy system includes an electric system, a gas system, a heating system, and an energy storage system.

3. The method for assessing and warning of cascading failures in an integrated energy system according to claim 1, characterized in that, Based on the weighted clustering algorithm, the standardized feature parameters are divided to obtain a discrete operating state space, specifically: Calculate the degree to which each characteristic parameter deviates from the safe median value after standardization to obtain the safe deviation of each parameter; The initial safety weights of each feature parameter are determined based on the analytic hierarchy process (AHP), and the initial safety weights are corrected according to the safety deviation to obtain the corrected weights. Based on the corrected weights, the weighted Euclidean distance between each data point is calculated to obtain a similarity measure; Based on the similarity metric, the K-means clustering algorithm is used to iteratively cluster the standardized feature parameters. The clustering process is terminated when the change in the cluster center is less than a preset threshold or the maximum number of iterations is reached, and a discrete running state space is output.

4. The method for assessing and warning of cascading failures in an integrated energy system according to claim 1, characterized in that, Based on the discrete operating state space, the historical operating data of the integrated energy system are mapped to states, the transition frequencies between states are statistically analyzed, and a Markov state transition probability matrix is ​​constructed, specifically as follows: Each set of feature parameters in the historical operation data is mapped to the discrete operation state space to obtain a discrete state sequence; Count the number of times a state transitions from any state to any other state in the historical state sequence; Based on the number of transitions obtained from statistics, calculate the initial transition probability between each state and construct the initial state transition probability matrix.

5. The method for assessing and warning of cascading failures in an integrated energy system according to claim 1, characterized in that, Using real-time runtime data, combined with a sliding time window and a forgetting factor, the state transition probability matrix is ​​iteratively updated and calculated; specifically: Set a fixed-duration sliding time window and count the number of transitions of the system between discrete states in real time within the window; Introducing a forgetting factor, the updated state transition probability matrix is ​​calculated using a weighted smoothing update formula, where the weighted smoothing update formula is: ， in, The real-time updated state transition probability of the system from state i to state j Forgetting factor, This represents the number of state transitions from state i to state j within the window. This is the old probability from the previous period.

6. The method for assessing and warning of cascading failures in an integrated energy system according to claim 1, characterized in that, Typical cascading failure propagation paths of an integrated energy system are obtained, the correlation strength coefficient of each path is calculated, and the correlation strength coefficient is incorporated into the update calculation of the Markov state transition probability matrix to obtain the cumulative probability of cascading failures occurring in the system in the future time period, specifically: Typical cascading failure propagation paths of integrated energy systems are obtained. The importance weight of each coupling link in each path is determined by the analytic hierarchy process (AHP). The coupling strength of each link is determined by combining equipment coupling test data. The correlation strength coefficient of each failure propagation path is calculated in a comprehensive manner. Initialize the state probability vector at the current moment, and use the updated state transition probability matrix to perform multi-step iterative prediction to obtain the probability of the system being in each discrete state at each future moment; A preset fault state set is defined, and the probability of the system being in the fault state set at each future time is accumulated and corrected by combining the correlation strength coefficient to obtain the cumulative probability of the system experiencing a cascading failure in the future time period.

7. The method for assessing and warning of cascading failures in an integrated energy system according to claim 1, characterized in that, The real-time risk value is calculated based on the cumulative probability. An adaptive fuzzy PID algorithm is used to correct the preset warning threshold in real time. The corrected threshold is compared with the real-time risk value to obtain the warning classification result, specifically: Construct a risk quantification model that includes economic losses and energy supply disruptions, and combine it with the cumulative probability calculation system to calculate the real-time risk value. Calculate the risk deviation between the real-time risk value and the preset benchmark warning threshold; Based on the aforementioned risk deviation, the correction amount of the PID control parameters is output in real time through adaptive fuzzy PID fuzzy inference; The correction amount of the warning threshold is calculated based on the corrected PID parameters, and the correction amount is superimposed on the baseline warning threshold to obtain the dynamic warning threshold. The dynamic warning threshold is compared with the real-time risk value, and the corresponding warning result is obtained according to the preset classification rules.

8. A system for assessing and warning of the probability of cascading failures in an integrated energy system, characterized in that, include: The state space partitioning module is used to acquire the characteristic parameters of the integrated energy system during operation and perform standardization processing. Based on the weighted clustering algorithm, the standardized characteristic parameters are partitioned to obtain the discrete operating state space. The matrix construction and update module is used to perform state mapping on the historical operating data of the integrated energy system based on the discrete operating state space, count the transition frequency between each state, construct a Markov state transition probability matrix, and dynamically update the state transition probability matrix using real-time operating data, combined with a sliding time window and a forgetting factor. The failure probability calculation module is used to obtain typical cascading failure propagation paths of the integrated energy system, calculate the correlation strength coefficient of each path, and integrate the correlation strength coefficient into the iterative calculation of the Markov chain to solve the cumulative probability of cascading failures occurring in the system in the future period. The dynamic risk assessment and early warning module is used to calculate the real-time risk value based on the cumulative probability, use an adaptive fuzzy PID algorithm to correct the preset early warning threshold in real time, compare the corrected dynamic threshold with the real-time risk value, and output the graded early warning result.

9. A computer device comprising a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the method for probabilistic assessment and early warning of cascading failures in an integrated energy system as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps in the method for probabilistic assessment and early warning of cascading failures in an integrated energy system as described in any one of claims 1-7.

Images