Power system planning evaluation method and system based on multi-dimensional index decision support

Abstract

The invention discloses an electric power system planning evaluation method and system based on multi-dimensional index decision support, and the method comprises the steps: constructing a scene set containing faults, fluctuation and load growth through obtaining a topological structure, line parameters and historical operation data of an electric power system; generating a topological risk index in the scene set by utilizing power flow calibration and topological risk identification, and dynamically adjusting a multi-dimensional index weight in combination with a phase factor in an impact and recovery stage; further fusing Monte Carlo distribution and Markov distribution through a control variable method to obtain a robust reliability index; mapping impact into stress, mapping function loss into strain, and constructing an elastic curve of the power system; and finally, extracting elastic indexes such as a yield point, ductility, recovery time and a function loss area, and outputting a comprehensive evaluation result of power system planning in combination with a robust reliability index so as to realize scientific comparison and optimization of different planning schemes.

Description

Technical Field

[0001] This invention relates to the field of planning and evaluation technology, specifically to a power system planning and evaluation method and system based on multi-dimensional indicator decision support. Background Technology

[0002] With the ongoing energy transition and the increasing complexity of power systems, power system planning and evaluation are facing new challenges. On the one hand, the large-scale integration of renewable energy sources has led to high volatility and uncertainty in power system operation, making traditional planning and evaluation methods that rely on single load forecasts and static indicators unable to accurately reflect the true state of the system.

[0003] On the other hand, with increasingly complex power grid topologies, local faults can rapidly amplify through cascading effects, triggering widespread blackouts. This makes it difficult to effectively characterize the vulnerability and resilience of power systems using a single reliability index or static risk assessment. Existing research often employs the Monte Carlo method for global reliability estimation, but its convergence speed is slow under extreme tail scenarios; or it uses the Markov chain method for state transition modeling, which, while capable of characterizing details, incurs excessive computational costs in large-scale systems, making it difficult to balance efficiency and accuracy.

[0004] Meanwhile, in multi-indicator comprehensive evaluation, existing methods generally use fixed weights, ignoring the dynamic differences in the contribution of each indicator to the overall vulnerability of the system in different impact stages (such as the failure occurrence and recovery process). Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is to address the issues of the inability to dynamically weight multidimensional indicators in power system planning and evaluation, the contradiction between reliability calculation efficiency and accuracy, and insufficient elastic characterization.

[0007] To address the aforementioned technical problems, this invention provides the following technical solution: a power system planning and evaluation method based on multi-dimensional index decision support, comprising:

[0008] Acquire power system topology, line parameters, and historical operating data, including historical load data, renewable energy output data, and historical failure rates; construct a scenario set based on historical operating data, which includes failure events, random fluctuations, and load growth.

[0009] Under the aforementioned scenario set, a topological risk index is generated to characterize structural vulnerability by utilizing power flow calibration and topological risk identification.

[0010] Based on the preset initial weights of the multidimensional indicators, and combined with the topological risk index and the phase factors of the power system in the impact and recovery phases, the weights of the multidimensional indicators are adjusted to obtain dynamic weights that change with time and scenario.

[0011] Under the aforementioned scenario set, the Monte Carlo distribution and the Markov distribution are unbiasedly fused using the control variable method to obtain the modified robust reliability index.

[0012] The system function level curve is obtained by weighting and fusing the multi-dimensional indicators according to the dynamic weights, and the impact intensity is mapped to stress and the function loss ratio is mapped to strain to form the elasticity curve of the power system.

[0013] The yield point, ductility, recovery time, and functional loss area are extracted from the elastic curve, and combined with the robust reliability index to form a comprehensive evaluation result for power system planning.

[0014] As a preferred embodiment of the power system planning and evaluation method based on multi-dimensional index decision support described in this invention, the scenario set includes: filtering the historical operating data based on each scenario feature label identified in the historical operating data; for each identified scenario feature label, the feature part in the identification process is extracted to obtain the corresponding historical operating data set.

[0015] During the crawling process, contextual features are identified through different data dimensions to obtain the time interval to be crawled. Within the time interval to be crawled, the running data of each dimension is crawled.

[0016] As a preferred embodiment of the power system planning and evaluation method based on multidimensional index decision support described in this invention, wherein: under the scenario set, a linearized model for rapid estimation of DC power flow is used to solve a sparse linear equation set to generate a matrix of phase angles of each node and power flow of each line throughout the entire time series.

[0017] Characteristic analysis is performed on the rapid DC power flow estimation results to obtain representative time periods;

[0018] Selecting the representative time period, the AC power flow is solved using the Newton-Raphson method to achieve accurate calculation of AC power flow, obtaining the node voltage amplitude, node voltage phase angle, and active / reactive power flow of the line. The consistency calibration relationship between DC power flow and AC power flow is established by fitting mapping.

[0019] Based on the calibrated power flow results, the power flow distribution factor and line cut-off factor are calculated, and the power flow distribution factor matrix and line cut-off factor matrix are generated. For each line, a comprehensive risk score is obtained by weighted summation of the maximum relative load rate, the time standard deviation of the relative load rate, and the maximum cut-off impact intensity. The power flow distribution factor matrix, line cut-off factor matrix, and comprehensive risk score are used together as a comprehensive risk assessment indicator.

[0020] Complex network metrics are generated through topological analysis of node betweenness, network connectivity, and effective graph resistance.

[0021] The comprehensive risk assessment index and the complex network index are used as the topology risk index to characterize structural vulnerability.

[0022] As a preferred embodiment of the power system planning and evaluation method based on multidimensional index decision support described in this invention, the step of adjusting the weights of the multidimensional indexes includes training each index in the topology risk index for the impact phase and recovery phase under different scenarios to generate dynamic weights that change with time and scenarios.

[0023] During the training process, at each time step, the rate of change of each indicator is calculated as a phase factor. The ratio of each phase factor to the standard phase factor is analyzed. The weights of the indicators are adjusted proportionally, and the adjusted weights of all indicators are normalized to obtain the indicator weights used for training.

[0024] In constructing the initial weights of multidimensional indicators, a standard phase factor corresponding to each indicator is pre-constructed.

[0025] As a preferred embodiment of the power system planning and evaluation method based on multidimensional index decision support described in this invention, the global reliability distribution is obtained using the Monte Carlo method.

[0026] In the high-risk tail scenario of the reliability distribution, a Markov chain state transition model is constructed to obtain the accurate reliability distribution;

[0027] The Monte Carlo distribution and the Markov distribution are unbiasedly fused using the control variable method:

[0028] In the non-high-risk tail scenario portion, MC-as-is estimation is used;

[0029] In high-risk tail scenarios, the conditional expectation calculated by MK is used to replace the noise samples of MC.

[0030] As a preferred embodiment of the power system planning and evaluation method based on multidimensional index decision support described in this invention, the six topological indicators of power flow distribution factor matrix, line cut-off factor matrix, comprehensive risk score, node betweenness, network connectivity and effective graph resistance are unified in dimensions, and then weighted summation is performed using the dynamic weights to obtain the ordinate of the elastic curve.

[0031] When unifying the dimensions of each indicator, dimensionality reduction mapping is performed through their respective pre-trained mapping functions to obtain the scalar form of each indicator.

[0032] The impact intensity, representing stress, is obtained by weighting and summing the injection impact, ramp impact, fault / cut-off impact, topology change impact, and operational margin impact according to preset weights.

[0033] Among them, injection impact is the magnitude of load / generation deviation from the baseline; ramp-up impact is the rate of change of load and renewable energy over time; fault / cutover impact is the proportion of capacity withdrawn in the current period; topology change impact is the magnitude of rapid changes in parameters / structure; and operating margin impact is the inverse indicator of safety margin.

[0034] As a preferred embodiment of the power system planning and evaluation method based on multi-dimensional index decision support described in this invention, wherein: a corresponding comprehensive evaluation result is generated for each scenario set;

[0035] Each parameter in the comprehensive evaluation result is analyzed using the threshold method;

[0036] The evaluation results and corresponding scenario sets for those with unqualified screening parameters are output.

[0037] A power system planning and evaluation system based on multidimensional index decision support, as described in this invention, comprises: a data acquisition unit that acquires power system topology, line parameters, and historical operating data; and a scenario set including fault events, random fluctuations, and load growth based on the historical operating data; an identification unit that, under the scenario set, generates a topology risk index to characterize structural vulnerability using power flow calibration and topology risk identification; and an adjustment unit that, based on preset initial weights for the multidimensional indicators, combines the topology risk index with phase factors of the power system's impact and recovery phases to obtain a result that varies with time and scenario conditions. The system comprises: a dynamic weighting of scenario changes; a first evaluation unit, which, under the scenario set, performs unbiased fusion of the Monte Carlo distribution and the Markov distribution using the control variable method to obtain a modified robust reliability index; an analysis unit, which performs weighted fusion of various multidimensional indices according to the dynamic weights to obtain a system function level curve, and maps the impact intensity to stress and the function loss ratio to strain to form the elasticity curve of the power system; and a second evaluation unit, which extracts elasticity indices such as yield point, ductility, recovery time, and function loss area from the elasticity curve, and combines them with the robust reliability index to form a comprehensive evaluation result for power system planning.

[0038] A computer device includes: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.

[0039] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in any one of the present invention.

[0040] The beneficial effects of this invention are as follows: First, the power system planning and evaluation method based on multi-dimensional index decision support provided by this invention improves accuracy while ensuring rapid calculation efficiency through consistency calibration of DC and AC power flows, enabling feasible power flow estimation for large-scale systems. Second, the dynamic weight adjustment method based on the topological risk index and the phase factor of the impact stage can reflect the differential contribution of multi-dimensional indicators to system vulnerability in real time, avoiding the one-sidedness of traditional fixed weights. Third, the unbiased fusion of Monte Carlo and Markov distributions using the control variable method maintains the representativeness of the global distribution while enhancing accuracy under extreme tail scenarios, thus obtaining robust reliability indicators. Finally, this invention introduces a method for constructing power system elasticity curves, mapping impact intensity to stress and functional loss ratio to strain, systematically extracting key elasticity indicators such as yield point, ductility, and recovery time, providing a more comprehensive, dynamic, and interpretable decision-making basis for power system planning. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 The first embodiment of the present invention provides an overall flowchart of a power system planning and evaluation method based on multidimensional index decision support. Detailed Implementation

[0043] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0044] Example 1, referring to Figure 1 As an embodiment of the present invention, a power system planning and evaluation method based on multi-dimensional index decision support is provided, comprising:

[0045] S1: Obtain the power system topology, line parameters, and historical operating data, including historical load data, renewable energy output data, and historical failure rates; construct a scenario set based on the historical operating data, which includes failure events, random fluctuations, and load growth.

[0046] The scenario set includes filtering the historical operation data based on each scenario feature tag identified in the historical operation data: for each identified scenario feature tag, the feature part in the identification process is captured to obtain the corresponding historical operation data set.

[0047] During the data capture process, scenario features are identified through different data dimensions (historical load data, renewable energy output data, and historical failure rate) to obtain the time interval to be captured. Within the time interval to be captured, the operational data for each dimension is captured.

[0048] By identifying scenario feature labels from multi-dimensional operational data such as historical load data, renewable energy output data, and historical failure rates, the corresponding characteristic time intervals are filtered and captured to form a corresponding historical operational data set. This method enables the scenario set to cover both normal operation and extreme impact states of the power system, thereby providing a multi-scenario data foundation for subsequent power flow calculations, topology risk identification, and dynamic weight adjustment.

[0049] It's important to understand that the acquired historical operational data is characterized by three dimensions: load data, renewable energy output data, and historical failure rate. In the load data dimension, time series clustering or extreme value thresholding is used to identify peak load, off-peak load, and sudden load increases. In the renewable energy output dimension, slope change detection or quantile filtering is used to identify rapid declines, rapid increases, and prolonged periods of low output for wind and solar power. In the historical failure rate dimension, failure event statistics and Markov state segmentation are used to identify high-frequency failure intervals, cascading failure intervals, and low-risk intervals.

[0050] Based on the identified scenario feature labels, the corresponding time intervals are marked, and data is extracted from historical operational data: when the label is "high load period," the load curve, the renewable energy output curve at the same moment, and line fault statistics for that time interval are extracted; when the label is "wind and solar power drop period," the wind and solar power output curves for that time interval are extracted, and the load level and topology operating status are extracted simultaneously; when the label is "fault interval," fault records and corresponding power flow distribution, node voltage, and line operating data for that interval are extracted. Ultimately, each scenario feature label corresponds to a set of historical operational data, and different sets together constitute a scenario set, serving as input for subsequent power flow calibration and risk assessment.

[0051] S2: Under the aforementioned scenario set, a topological risk index is generated to characterize structural vulnerability by utilizing power flow calibration and topological risk identification.

[0052] Under the aforementioned scenario set, a linearized model for rapid estimation of DC power flow is used to solve a sparse linear equation set, generating a matrix of phase angles of each node and power flow of each line throughout the entire time series.

[0053] Specifically, using the DC power flow, a linearized model is used to solve the sparse linear equation system Bθ=P, where θ is the node phase angle and P is the active power injection.

[0054] A sparse Cholesky / ConjugateGradient solver is used for batch, time-parallel solving. The output node angle θ is... DC (t); Branch road contributes to the tide b represents the DC active power flow of line e at time t; e =1 / x e Indicates the susceptance of line e; x e θ represents the reactance of line e; i ,θ jB represents the phase angle of the nodes at both ends of the line; B represents the node's Laplace matrix (a sparse symmetric matrix composed of network reactances, used for linearized power flow calculations). This represents the DC active power flow (in MW) of line e at time t. t represents the time index (timestamp from the identified scenario segment). e represents the line index (network edge number). i,j represent the node indices (numbers of the buses at both ends of the line).

[0055] Characteristic analysis is performed on the rapid DC power flow estimation results to obtain representative time periods. These representative time periods are then selected, and the Newton-Raphson method is used to solve the AC power flow, achieving accurate calculation of the AC power flow and obtaining node voltage amplitudes, node voltage phase angles, and line active / reactive power flow. The full-time DC results are then subjected to DTW-k-medoids / PELT variable point / clustering to select... (For example, 20-100 typical moments in each game).

[0056] Calculations were performed using full AC power flow (Newton-Raphson / Fast-Decoupled), and the output (in...) (Above): Node voltage magnitude / phase angle Active / Reactive Power Flow of the Line Current, power factor; line / bus constraints (voltage exceeding limits, thermal stability exceeding limits), system loss P loss (t). Extract a small number of representative moments to obtain an accurate solution, correct the systematic bias of DC from the source, and avoid subsequent risk assessments being "based on a distorted foundation".

[0057] A consistent calibration relationship between DC and AC power flow is established by fitting a mapping. Using a small number of precise AC points, a mapping Π is learned to calibrate the full-time results of DC to an approximate AC level for subsequent sensitivity and risk assessment.

[0058] Fitting for each line e:

[0059]

[0060] Fitter: Weighted Ridge Regression / Huber Robust Regression (weights are related to loading rates). For full time series... Output

[0061] The power flow distribution factor and line cutoff factor are calculated based on the calibrated power flow results, and the power flow distribution factor matrix and line cutoff factor matrix are generated. For each line, a comprehensive risk score is obtained by weighted summation of the maximum relative load rate, the time standard deviation of the relative load rate, and the maximum cutoff impact intensity. The power flow distribution factor matrix, line cutoff factor matrix, and comprehensive risk score are used together as a comprehensive risk assessment indicator.

[0062] Specifically, PTDF and LODF are calculated under calibration parameters, and a risk score R is constructed based on the calibrated power flow. e Filter key edge sets

[0063]

[0064] (H represents the branch-node angle difference mapping matrix; M represents the node injection selection matrix; PTDF) e,k (Indicates the sensitivity of node / edge injection to line e). Risk scoring example:

[0065]

[0066] R e =αρ e +βσ e +γη e , or Top-k

[0067] H represents the branch-node angle difference mapping matrix (determined by the network topology, mapping node angle differences to branch power). M represents the node injection selection matrix (selecting and balancing the combination of injected / extracted nodes). PTDF represents the power flow distribution factor matrix (the linear sensitivity of each line to small injected disturbances). PTDF e,k This represents the sensitivity element for injection unit k with respect to line e. LODF e,k This represents the line cutoff factor (the impact of cutting off line k on the power flow of line e). This indicates the active power flow of the line after calibration (approximately AC). This represents the upper limit of thermal stability for line e (in MW). ρ e This represents the maximum relative load factor (the relative saturation level at the most critical moment). σ e η represents the time standard deviation of the relative load factor (power flow fluctuation). e This represents the maximum impact intensity of the cutoff (the absolute value of the maximum LODF of any cutline k with respect to e). α, β, and γ represent the risk score weighting coefficients (non-negative, adjustable, used to balance the three impacts). R e This represents the overall risk score of line e (the higher the value, the more critical / vulnerable). τ R This represents the risk threshold (or the top-k percent selected as the critical edge). This represents the critical edge set (the set of high-risk paths). k represents the index of the path that was cut.

[0068] By analyzing topological indicators such as node betweenness, network connectivity, and effective graph resistance, a complex network indicator is generated; the comprehensive risk assessment indicator and the complex network indicator are used as the topological risk index to characterize structural vulnerability.

[0069] S3: Based on the preset initial weights of the multidimensional indicators, and combined with the topological risk index and the phase factors of the power system in the impact and recovery phases, the weights of the multidimensional indicators are adjusted to obtain dynamic weights that change with time and scenario.

[0070] Furthermore, based on the aforementioned topological risk index, each indicator is trained for the impact and recovery phases under different scenarios to generate dynamic weights that change with time and scenario.

[0071] During training, at each time step, the rate of change of each indicator is calculated as a phase factor. The ratio of each phase factor to the standard phase factor is analyzed, and the indicator weights are adjusted proportionally. Then, the adjusted weights of all indicators are normalized to obtain the indicator weights used for training. Specifically, when constructing the initial weights for the multi-dimensional indicators, a standard phase factor corresponding to each indicator is pre-constructed. In other words, this vulnerability is specific to different indicators, thus allowing the generation of weights using this vulnerability.

[0072] It's important to understand that, based on the preset initial weights of the multi-dimensional indicators, a topological risk index and phase factors for the impact and recovery phases are introduced to dynamically adjust the weights of each indicator. Different indicators contribute differently to power system vulnerability under different scenarios. For example, during the impact phase, the power flow distribution factor and line disconnection factor better reflect risk, while during the recovery phase, network connectivity and effective graph resistance better reflect system resilience. By analyzing the rate of change of each indicator step-by-step during training, this rate of change is used as a phase factor, compared with a preset standard phase factor, and the indicator weights are adjusted proportionally. Normalization is then used to ensure that the weights sum to one. The resulting dynamic weights can adjust with time and scenario evolution, thus realistically reflecting the dynamic changes in system vulnerability under different operating conditions. Since vulnerability manifests as the different sensitivities of each indicator to the overall system stability at different stages, this method generates weights by quantifying vulnerability differences, making the comprehensive evaluation results of the multi-dimensional indicators more consistent with the actual operating characteristics of the power system.

[0073] S4: Under the given scenario set, the Monte Carlo distribution and the Markov distribution are unbiasedly fused using the control variable method to obtain the corrected robust reliability index.

[0074] The global reliability distribution is obtained using the Monte Carlo method.

[0075] In the high-risk tail scenario of the reliability distribution, a Markov chain state transition model is constructed to obtain the accurate reliability distribution.

[0076] The Monte Carlo distribution and the Markov distribution are unbiasedly fused using the control variable method:

[0077] In the non-high-risk tail scenario portion, MC-as-is estimation is used.

[0078] In high-risk tail scenarios, the conditional expectation calculated by MK is used to replace the noise samples of MC.

[0079] Specifically, the sample space is divided into two parts based on "tail events": and In the NONTAIL section, the MC is estimated as is; in the TAIL section, the conditional expectation calculated by MK is used to replace the noisy sample of MC.

[0080] Indicator Function ω represents a single, randomized system event (renewable output, load, fault path, etc.). The true probability of tail events. Using MC frequency (Monte Carlo estimation of the tail probability, i.e., the proportion of samples falling into the tail in N simulations) estimate. MK gives... The precise / high-precision value (establishing a CTMC / DTMC solution or high-precision numerical value in the tail subspace). i represents the index of the simulation number, fused estimator:

[0081]

[0082] First term: MC average of non-tailed samples; Second term: using The tail condition expectation.

[0083]

[0084] E[·]: Mathematical expectation operator. E[X·(1-I)]: Expected contribution of the non-tailed portion. E[I]: Probability of the tailed event (i.e., p). E[X|I=1]: Expected index (i.e., μ) under the condition that the tailed event occurs. tail The above formula confirms that there is no deviation.

[0085] Time-resolved (time-by-time LOLP) for each time t: the tail event definition can be either "LOLP(t) trigger threshold" or "system at critical availability level". In time-by-time analysis, the fusion estimate for time t is:

[0086]

[0087] Similarly, for EENS, the tail segment can be replaced after time-domain integration / summation.

[0088] In summary, the implementation steps are as follows:

[0089] 1. Define tail events Such as "No power supply > q-quantile", "Chain load shedding trigger", etc. (from your LODF / critical edge determination).

[0090] 2. Run MC N times: Record X (i) I (i) (or hourly) ),get

[0091] 3. Construct MK(CTMC / DTMC) in the tail subspace: calculate μ tail .

[0092] 4. Substitute into the formula and output. Advantages: Unbiased, simple, and highly explanatory.

[0093] S5: Based on the dynamic weights, the multi-dimensional indicators are weighted and fused to obtain the system function level curve, and the impact intensity is mapped to stress and the function loss ratio is mapped to strain to form the elasticity curve of the power system.

[0094] After unifying the dimensions of six topology indicators—power flow distribution factor matrix, line cutoff factor matrix, comprehensive risk score, node betweenness, network connectivity, and effective graph resistance—the dynamic weights are used to perform a weighted summation to obtain the ordinate of the elastic curve.

[0095] When unifying the dimensions of each indicator, dimensionality reduction mapping is performed through their respective pre-trained mapping functions to obtain the scalar form of each indicator.

[0096] It should be noted that in this embodiment, the topological indices of "node betweenness, network connectivity, and effective graph resistance" are synthesized from stage S2 and used as the TRI calculation. In the calculation of the topological indices of "node betweenness, network connectivity, and effective graph resistance" mentioned above, the calculation of the three indices is simplified to the calculation of a single index. First, the key edge set E is identified based on the risk-weighted graph of the power grid. c And the edge betweenness number of each of these lines. Averaging is performed to obtain the overall importance of the critical path in power flow. Subsequently, a system-level complex network metric is introduced: effective graph resistance. The topology risk index TRI is used to characterize overall redundancy and transmission efficiency, while connectivity κ measures network integrity under fault conditions. These indices are normalized and weighted using coefficients ν1, ν2, and ν3 to form the final index, which reflects the structural vulnerability of the power system in a given scenario.

[0097]

[0098] TRI(s) represents the Topological Risk Index, a comprehensive quantitative indicator of the structural vulnerability of a power system under scenario s. ν1, ν2, and ν3 represent the composite weights, used to control the contribution ratio of different complex network indicators in TRI; they are generally determined through normalization or training to ensure their sum equals 1. |E c | Represents the vulnerable critical edge set E c The cardinality, i.e. the number of lines contained in the set, is used to average the summation result of the set. This means traversing all critical edges (high-risk routes) and calculating their index values.

[0099] In other alternative embodiments, the three metrics can be calculated separately or combined into one metric in other ways. This represents the normalized value of the betweenness number of line e. It indicates the importance of the line in the power flow path of the power grid; the larger the value, the more crucial the "bridge" role the line plays in the system transmission. This represents the normalized value of the effective graph resistance. It indicates the global connectivity and redundancy of the power network. A larger effective graph resistance indicates lower system redundancy and poorer resilience. κ represents network connectivity, taking values ​​[0,1] in the given scenario. When the network is partitioned / isolated, connectivity decreases, indicating a more vulnerable system. (1-κ) represents a vulnerability measure of connectivity. The lower the connectivity, the higher this value, indicating a more vulnerable system.

[0100] The impact intensity, representing stress, is obtained by weighting and summing the injection impact, ramp impact, fault / cut-off impact, topology change impact, and operating margin impact according to preset weights.

[0101] Among them, injection impact is the magnitude of load / generation deviation from the baseline; ramp-up impact is the rate of change of load and renewable energy over time; fault / cutover impact is the proportion of capacity withdrawn in the current period; topology change impact is the magnitude of rapid changes in parameters / structure; and operating margin impact is the inverse indicator of safety margin.

[0102] For example, when calculating the topology metrics of "node betweenness, network connectivity, and effective graph resistance" separately: Vertical axis (overall vulnerability):

[0103]

[0104] Horizontal axis (stress / impact strength):

[0105]

[0106] s: Scenario number. From the scenario set, used to distinguish different combinations of historical / synthetic operating states (e.g., high load, sudden drop in wind and solar power, critical line maintenance, etc.). t: Time index. Represents the discrete time step (e.g., 5 minutes, 15 minutes) under scenario s, used to construct time series curves. Y s (t): Overall vulnerability (vertical axis). Values ​​range from 0 to 1, with larger values ​​indicating higher structural vulnerability. The scalarized value of the i-th vulnerability component at scenario s and time t. All z i After standardization using 0-1, the dimensions are unified, and they can be directly added. PTDF s (t): Power flow distribution factor matrix under scenario s and time t. Characterizes the linear impact of unit injection / extraction on power flow along each line. ΔP s (t): The injection change vector relative to the baseline (net injection difference between generation and load). Used to transform PTDF into power flow increments falling within the line space. ||·|| p : Norm of vector p. Commonly used is p=2 (energy-type measure) or p=∞ (maximum component, conservative approach), used to measure the intensity of overall power flow redistribution caused by injected disturbances. Norm(·): Robust 0-1 normalization operator. It is recommended to use a linear mapping based on the historical 5% / 95th percentile and truncate the external values ​​to avoid extreme values ​​dominating the composite score. LODF s (t)[e,k]: Line cutoff factor. In scenario s and at time t, the relative change in power flow of line e after cutting off line k reflects the sensitivity to cascading redistribution. e,k |...|: Take the maximum value of the amplitude of all affected lines e and cut lines k to obtain the upper bound of the intensity of the most unfavorable tangent effect. A comprehensive risk score for line e (e.g., maximum relative load factor, load factor standard deviation, weighted sum of maximum shedding impact). Used to aggregate risks at the component level. Agg{·}: For line-by-line risk R e Aggregate functions. Robust statistics such as the Top-k average or 95th percentile can be used to highlight high tail risk without being overly diluted by the mean. Key edge set The average edge betweenness. Measured by the proportion of "bridging" lines in the shortest / most important path; a higher value indicates that the structure is more likely to break at that point. Critical edge set. A subset of highly vulnerable lines selected based on a comprehensive risk and sensitivity threshold (e.g., high ρ, high |LODF|). κ s(t): Network connectivity (0–1). Can be defined as the percentage of nodes in the most connected component, or a normalized version of connectivity. Smaller values ​​indicate a closer proximity to network fragmentation / isolation. 1-κ s (t): Connectivity loss metric. Explicitly representing the direction of "lower connectivity means greater fragility" facilitates summing with other components in the same direction. Effective resistance. Reflects the average effective resistance of the entire network / transmission redundancy; a higher value indicates fewer redundant paths and higher detour costs. α i : Weighting coefficients on the vertical axis. Non-negative and normalized (summing to 1), controlling the relative weights of the six components in the overall vulnerability score; can be fixed or determined through offline training. σ s (t): Stress / Impact Intensity (horizontal axis). It only reflects the strength of external / triggered disturbances and does not directly include structural vulnerability. The value ranges from 0 to 1, with a larger value indicating a stronger impact in the current period. Injection deviation intensity. ||ΔP s The normalized value of (t)||2 measures the deviation of total injection / load from the baseline. load,s (t): System load power time series under scenario s. Used to calculate the ramp rate. P RES,s (t): Time series of renewable energy output (wind / solar, etc.) under scenario s. Used to calculate ramp rate and uncertain disturbances. The time derivative of load and regenerative capacity (numerically expressed as the difference ΔP / Δt). The larger the norm, the steeper the ramp and the stronger the instantaneous impact on the system. Climbing rate intensity. This is calculated by adding the load to the renewable climbing norm and normalizing the result, reflecting the dynamic impact of rapid changes on the system. F max Line thermal stability upper limit (MVA or MW). Used to characterize capacity benchmark. ∑ outaged F max / ∑ all F max The percentage of capacity out of service during the current period (fault / cutover intensity). The higher the percentage, the greater the pressure on the system. Fault / Cutoff Impact. A standardized quantity based on the above capacity percentages, providing a direct measure of the impact of current outages on available transmission capacity. B: Node Electrical Navigation Laplace Matrix (a sparse symmetric matrix formed by line reactances, the basis of DC power flow). A baseline quantity for topology parameters. B s (t): The change in electrical nanolatatus relative to the baseline (e.g., line commissioning / decommissioning, parameter changes, B changes due to topology switching). ||·|| F Frobenius norm: Used to measure the energy of the overall change of a matrix (the square root of the sum of the squares of its elements). Topology change intensity. ||ΔB s (t)|| F / ||B|| FThe normalized value reflects the relative magnitude of rapid changes in parameters / structure. M margin,s (t): Comprehensive operational safety margin (a combination of indicators such as minimum voltage margin, total spinning reserve, reactive power margin, etc.). The larger the value, the safer the operation. The operating margin is inversely inverted. By using the reciprocal, the direction of "the smaller the margin, the greater the impact" is reversed to "the larger the value, the more dangerous," which facilitates superposition with other impact components in the same direction. Operating margin shock. After standardization, the measurement captures the external pressure of "thinning safety cushion". ω j : The weights of each impact component on the horizontal axis. Non-negative and normalized (summing to 1), controlling the relative contribution of the five types of external impacts to the total stress, which can be fixed or calibrated through historical events.

[0107] S6: Extract the elasticity indices of yield point, ductility, recovery time, and functional loss area from the elastic curve, and combine them with the robust reliability indices to form a comprehensive evaluation result for power system planning.

[0108] Generate a corresponding comprehensive evaluation result for each scenario set; analyze each parameter in the comprehensive evaluation result using a threshold method; and output the evaluation results and corresponding scenario sets that fail to meet the parameter requirements.

[0109] Yield point: Determined by analyzing the inflection point of the curve as it transitions from the "initial elasticity stage" to the "increased vulnerability stage". In practice, the curve is first smoothed in chronological order, and then the changes in the slope and curvature of the curve are observed. When a clear inflection point appears, it indicates that the system is gradually entering a stage of rapid functional decline from being able to withstand impact; this point is the yield point.

[0110] Ductility: Ductility indicates how much compressive strength a system can retain after yielding. It is determined by comparing the degree of functional loss at the yield point with the maximum functional loss the system can ultimately withstand. Greater ductility indicates stronger compressive strength after yielding.

[0111] Recovery time: Recovery time refers to the time required for system performance to recover to near normal functional levels after it has dropped to its lowest point. In practice, a functional level threshold can be set (e.g., 95% of normal power supply capacity), and then the time difference between the lowest point and the curve reaching that level again can be calculated.

[0112] Functional loss area: The functional loss area represents the cumulative loss of the system throughout the entire shock and recovery process. In practice, the portion below the normal level on the function curve can be considered the "loss area," and the area of ​​this region is calculated as the total loss. The larger the area, the more severe the overall shock impact on the system.

[0113] After obtaining the indicator results for each scenario, they are standardized to avoid incomparability due to different units and value ranges. Then, thresholds are set for each indicator: hard thresholds are used for mandatory indicators (such as recovery time not being too long and failure probability not exceeding the upper limit); for tolerable indicators, a certain degree of fluctuation is allowed, controlled by setting upper and lower limits. Finally, the evaluation results for each scenario are checked one by one: if all key indicators are within the threshold range, the planning scheme corresponding to that scenario is deemed qualified; if some indicators exceed the threshold, the scenario is marked as unqualified, and the results are output.

[0114] On the other hand, this embodiment also provides a power system planning and evaluation system based on multi-dimensional index decision support, which includes:

[0115] The system comprises the following components: a data acquisition unit, a data collection unit, and an analysis unit. The data acquisition unit obtains the power system topology, line parameters, and historical operating data. Based on this historical data, a scenario set is constructed, including fault events, random fluctuations, and load increases. An identification unit, within this scenario set, uses power flow calibration and topology risk identification to generate a topology risk index characterizing structural vulnerability. An adjustment unit, based on preset initial weights for multidimensional indicators, combines the topology risk index with phase factors indicating the power system's impact and recovery phases to adjust the weights of the multidimensional indicators, resulting in dynamic weights that change over time and depending on the scenario. A first evaluation unit, within this scenario set, uses the control variable method to unbiasedly fuse the Monte Carlo distribution and the Markov distribution to obtain a corrected robust reliability index. An analysis unit, based on the dynamic weights, performs weighted fusion of the multidimensional indicators to obtain a system function level curve, mapping the impact intensity to stress and the function loss ratio to strain, forming the power system's elasticity curve. A second evaluation unit extracts elasticity indicators such as yield point, ductility, recovery time, and function loss area from the elasticity curve, and combines these with the robust reliability index to form a comprehensive evaluation result for power system planning.

[0116] If the above functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0117] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0118] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0119] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0120] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A power system planning and evaluation method based on multi-dimensional index decision support, characterized in that, include: Acquire power system topology, line parameters, and historical operating data, including historical load data, renewable energy output data, and historical failure rates; construct a scenario set based on historical operating data, which includes failure events, random fluctuations, and load growth. Under the aforementioned scenario set, a topological risk index is generated to characterize structural vulnerability by utilizing power flow calibration and topological risk identification. Based on the preset initial weights of the multidimensional indicators, and combined with the topological risk index and the phase factors of the power system in the impact and recovery phases, the weights of the multidimensional indicators are adjusted to obtain dynamic weights that change with time and scenario. Under the aforementioned scenario set, the Monte Carlo distribution and the Markov distribution are unbiasedly fused using the control variable method to obtain the modified robust reliability index. The system function level curve is obtained by weighting and fusing the multi-dimensional indicators according to the dynamic weights, and the impact intensity is mapped to stress and the function loss ratio is mapped to strain to form the elasticity curve of the power system. The yield point, ductility, recovery time, and functional loss area are extracted from the elastic curve, and combined with the robust reliability index to form a comprehensive evaluation result for power system planning.

2. The power system planning and evaluation method based on multi-dimensional index decision support as described in claim 1, characterized in that: The scenario set includes filtering the historical operation data based on each scenario feature tag identified in the historical operation data: for each identified scenario feature tag, the feature part in the identification process is extracted to obtain the corresponding historical operation data set. During the crawling process, contextual features are identified through different data dimensions to obtain the time interval to be crawled. Within the time interval to be crawled, the running data of each dimension is crawled.

3. The power system planning and evaluation method based on multi-dimensional index decision support as described in claim 2, characterized in that: Under the aforementioned scenario set, a linearized model for rapid estimation of DC power flow is used to solve a sparse linear equation set, generating a matrix of phase angles of each node and power flow of each line throughout the entire time series. Characteristic analysis is performed on the rapid DC power flow estimation results to obtain representative time periods; Selecting the representative time period, the AC power flow is solved using the Newton-Raphson method to achieve accurate calculation of AC power flow, obtaining the node voltage amplitude, node voltage phase angle, and active / reactive power flow of the line. The consistency calibration relationship between DC power flow and AC power flow is established by fitting mapping. The power flow distribution factor and line cutoff factor are calculated based on the calibrated power flow results, and the power flow distribution factor matrix and line cutoff factor matrix are generated. For each line, a comprehensive risk score is obtained by weighted summation of the maximum relative load rate, the time standard deviation of the relative load rate, and the maximum cutoff impact intensity. The power flow distribution factor matrix, the line cutoff factor matrix, and the comprehensive risk score are used together as comprehensive risk assessment indicators. Complex network metrics are generated through topological analysis of node betweenness, network connectivity, and effective graph resistance. The comprehensive risk assessment index and the complex network index are used as the topology risk index to characterize structural vulnerability.

4. The power system planning and evaluation method based on multi-dimensional index decision support as described in claim 3, characterized in that: The method of adjusting the weights of multidimensional indicators includes training each indicator in the topological risk index for the impact and recovery phases under different scenarios to generate dynamic weights that change with time and scenarios. During the training process, at each time step, the rate of change of each indicator is calculated as a phase factor. The ratio of each phase factor to the standard phase factor is analyzed. The weights of the indicators are adjusted proportionally, and the adjusted weights of all indicators are normalized to obtain the indicator weights used for training. In constructing the initial weights of multidimensional indicators, a standard phase factor corresponding to each indicator is pre-constructed.

5. The power system planning and evaluation method based on multi-dimensional index decision support as described in claim 4, characterized in that: The global reliability distribution is obtained using the Monte Carlo method; In the high-risk tail scenario of the reliability distribution, a Markov chain state transition model is constructed to obtain the accurate reliability distribution; The Monte Carlo distribution and the Markov distribution are unbiasedly fused using the control variable method: In the non-high-risk tail scenario portion, MC-as-is estimation is used; In high-risk tail scenarios, the conditional expectation calculated by MK is used to replace the noise samples of MC.

6. The power system planning and evaluation method based on multi-dimensional index decision support as described in claim 5, characterized in that: After unifying the dimensions of six topology indicators—power flow distribution factor matrix, line cutoff factor matrix, comprehensive risk score, node betweenness, network connectivity, and effective graph resistance—the dynamic weights are used to perform a weighted summation to obtain the ordinate of the elastic curve. When unifying the dimensions of each indicator, dimensionality reduction mapping is performed through their respective pre-trained mapping functions to obtain the scalar form of each indicator. The impact intensity, representing stress, is obtained by weighting and summing the injection impact, ramp impact, fault / cut-off impact, topology change impact, and operational margin impact according to preset weights. Among them, injection impact is the magnitude of load / generation deviation from the baseline; ramp-up impact is the rate of change of load and renewable energy over time; fault / cutover impact is the proportion of capacity withdrawn in the current period; topology change impact is the magnitude of rapid changes in parameters / structure; and operating margin impact is the inverse indicator of safety margin.

7. The power system planning and evaluation method based on multi-dimensional index decision support as described in claim 6, characterized in that: Generate a corresponding comprehensive evaluation result for each scenario set; Each parameter in the comprehensive evaluation result is analyzed using the threshold method; The evaluation results and corresponding scenario sets for those with unqualified screening parameters are output.

8. A power system planning and evaluation system based on multidimensional index decision support using the method described in any one of claims 1-7, characterized in that: The data acquisition unit obtains the power system topology, line parameters, and historical operating data; based on the historical operating data, it constructs a scenario set that includes fault events, random fluctuations, and load growth. The identification unit, under the scenario set, uses power flow calibration and topology risk identification to generate a topology risk index to characterize structural vulnerability; The adjustment unit adjusts the weights of the multidimensional indicators based on the preset initial weights of the multidimensional indicators, combined with the topological risk index and the phase factors of the impact and recovery phases of the power system, to obtain dynamic weights that change with time and scenario. The first evaluation unit, under the scenario set, performs unbiased fusion of the Monte Carlo distribution and the Markov distribution using the control variable method to obtain the corrected robust reliability index. The analysis unit performs weighted fusion of various multi-dimensional indicators according to the dynamic weights to obtain the system function level curve, and maps the impact intensity to stress and the function loss ratio to strain to form the elasticity curve of the power system; the second evaluation unit extracts the elasticity indicators of yield point, ductility, recovery time and function loss area from the elasticity curve, and combines them with the robust reliability indicators to form a comprehensive evaluation result of the power system planning.

9. A computer device, comprising: Memory and processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, it implements the steps of a power system planning and evaluation method based on multi-dimensional index decision support.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of a power system planning and evaluation method based on multidimensional index decision support.

Images