A key risk factor identification method and device, a mobile terminal and a storage medium

Abstract

The application discloses a kind of key risk factor identification method, device, mobile terminal and storage medium, the method includes: according to spot market clearing price curve in historical statistical period, obtain multiple same segment curves and generate same segment curve set;Wherein, same segment curve is the same segment of spot market clearing price curve in historical statistical period;According to segmented bidirectional clamping model, the segmented curve shape similarity between each same segment curve in same segment curve set and reference price segmented curve is calculated respectively;According to segmented curve shape similarity, overall similarity is generated, and overall similarity is clustered and analyzed, to generate clustering analysis result;According to clustering analysis result, key risk factor is identified. The accuracy of identifying key risk factor is improved by the embodiment of the application, thereby the safety of power system is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, mobile terminal, and storage medium for identifying key risk factors. Background Technology

[0002] The clearing price in the electricity spot market reflects the time and spatial value of electricity over a short period. Influenced by various risk factors in the spot market, such as market supply and demand, primary energy prices, and market sentiment, the clearing price varies significantly with different degrees of change in these risk factors. This variation is exacerbated by the increasing proportion of renewable energy generation. The volatile electricity price introduces uncertainty to market participants' returns, and can even lead to credit risk as risk exposure increases. Because the power system requires real-time supply and demand balance, defaults by market participants threaten the safe operation of the power grid and severely impact people's production and lives.

[0003] The clearing price in the electricity spot market is influenced by multiple risk factors. Therefore, when the daily risk factors in the electricity spot market are the same or similar, the clearing price will show the same or similar results. However, in actual market operation, the probability of all risk factors being the same or similar is low. Furthermore, the clearing price has varying sensitivities to different risk factors. When the sensitivity to a certain risk factor is low, even if the risk factor changes drastically, it will not be fully reflected in the clearing price. This type of risk belongs to the non-critical risk factors of electricity clearing prices. Because there are many risk factors affecting the clearing price, market risk monitors often cannot comprehensively monitor all risks. Continuously monitoring and controlling non-critical risk factors would waste market resources.

[0004] Existing methods for identifying key risk factors often employ the controlled variable approach, which involves controlling other risk factors to remain constant and examining the impact of changing risk factors on prices to determine whether they qualify as key risk factors. Alternatively, qualitative analysis can be used. However, this approach has drawbacks. Firstly, numerous risk factors influence the clearing price of electricity in the spot market, many of which are difficult or impossible to quantify. Furthermore, the varying frequencies of change and differences in data collection periods among these risk factors make the controlled variable approach challenging. Secondly, the time value of electricity results in diverse price curve profiles, making methods for identifying factors influencing price fluctuations in ordinary commodities unsuitable.

[0005] As can be seen from the above, existing methods for identifying key risk factors have the problem of low accuracy. Summary of the Invention

[0006] This invention provides a method, apparatus, mobile terminal, and storage medium for identifying key risk factors, which improves the accuracy of identifying key risk factors and thus enhances the security of the power system.

[0007] The first aspect of this application provides a method for identifying key risk factors, including:

[0008] Based on the spot market clearing price curves within the historical statistical period, multiple curves of the same segment are obtained and a set of curves of the same segment is generated; among them, the curves of the same segment are the spot market clearing price curves for the same period within the historical statistical period.

[0009] The similarity of the segmented curve shape between each segmented curve in the same segmented curve set and the benchmark price segmented curve is calculated based on the segmented two-way squeeze model.

[0010] The overall similarity is generated based on the similarity of the piecewise curve shapes, and cluster analysis is performed based on the overall similarity to generate the cluster analysis results.

[0011] Key risk factors were identified based on the results of cluster analysis.

[0012] One possible implementation of the first aspect also includes: obtaining a segmented curve of the benchmark price, specifically:

[0013] The spot market clearing price curves within the historical statistical period are normalized to generate the first set of normalized curves.

[0014] Select any one curve from the first set of curves as the benchmark price curve;

[0015] Obtain the benchmark price segmented curve based on the benchmark price curve.

[0016] In one possible implementation of the first aspect, the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve is calculated according to the segmented two-way squeeze model, specifically as follows:

[0017] Based on the curve downward shift coefficient, curve upward shift coefficient, benchmark price segmented curve, and any segmented curve in the same segmented curve set, set the constraints and objective function of the segmented two-way squeeze model;

[0018] Solve the objective function under the constraints to obtain the shape similarity between each segment curve in the set of segment curves and the benchmark price segment curve.

[0019] In one possible implementation of the first aspect, overall similarity is generated based on the similarity of the piecewise curve shapes, and cluster analysis is performed based on the overall similarity to generate cluster analysis results, specifically as follows:

[0020] Based on the similarity of the piecewise curve shapes, a similarity set is generated;

[0021] Based on the piecewise two-way squeeze model and the similarity set, the overall similarity is calculated; where, the overall similarity refers to the similarity between the benchmark price curve and the non-benchmark price curve in the first curve set;

[0022] Cluster analysis is performed based on overall similarity to generate cluster analysis results.

[0023] In one possible implementation of the first aspect, cluster analysis is performed based on overall similarity to generate cluster analysis results, specifically as follows:

[0024] Generate a similarity vector set based on overall similarity;

[0025] Based on the preset number of neighborhood elements, unprocessed vectors and core vectors are obtained from the similarity vector set;

[0026] Calculate the Mahalanobis distance between the unprocessed vector and the core vector;

[0027] Based on Mahalanobis distance, density-connected unprocessed vectors are grouped into the same cluster, and density-connected unprocessed vectors and core vectors are marked as processed vectors;

[0028] Unprocessed vectors with disjoint densities are labeled as unclassified vectors based on Mahalanobis distance;

[0029] Multiple clustering analysis results are generated based on the processed vectors and the unclassified vectors; among them, the processed vectors belong to different clusters.

[0030] In one possible implementation of the first aspect, key risk factors are identified based on cluster analysis results, specifically as follows:

[0031] Obtain the first risk factor from the processed vectors under different cluster categories;

[0032] If the daily variation value of the first risk factor is greater than the preset threshold, the first risk factor is determined to be a critical risk factor.

[0033] One possible implementation of the first aspect also includes:

[0034] Obtain the second risk factor from the processed vectors within the same cluster;

[0035] If the daily variation value of the second risk factor is greater than a preset threshold, the second risk factor is determined to be a critical risk factor.

[0036] A second aspect of this application provides a key risk factor identification device, including: an acquisition module, a calculation module, an analysis module, and an identification module;

[0037] The acquisition module is used to acquire multiple curves of the same segment based on the spot market clearing price curves within the historical statistical period and generate a set of curves of the same segment; wherein, the curves of the same segment are the spot market clearing price curves of the same period within the historical statistical period.

[0038] The calculation module is used to calculate the similarity of the segmented curve shape between each segmented curve in the same segmented curve set and the benchmark price segmented curve, based on the segmented two-way squeeze model.

[0039] The analysis module is used to generate overall similarity based on the similarity of the piecewise curve shapes, and to perform cluster analysis based on the overall similarity to generate cluster analysis results;

[0040] The identification module is used to identify key risk factors based on the results of cluster analysis.

[0041] A third aspect of this application provides a mobile terminal, including a processor and a memory, wherein the memory stores computer-readable program code, and the processor executes the computer-readable program code to implement the steps of the above-described key risk factor identification method.

[0042] A fourth aspect of this application provides a storage medium that stores computer-readable program code, which, when executed, implements the steps of the above-described key risk factor identification method.

[0043] Compared to existing technologies, this invention provides a method, apparatus, mobile terminal, and storage medium for identifying key risk factors. The method includes: acquiring multiple segmented curves and generating a set of segmented curves based on spot market clearing price curves within a historical statistical period; wherein, the segmented curves are spot market clearing price curves for the same period within the historical statistical period; calculating the segmented curve shape similarity between each segmented curve in the set of segmented curves and the benchmark price segmented curve based on a segmented two-way squeeze model; generating an overall similarity based on the segmented curve shape similarity, and performing cluster analysis based on the overall similarity to generate cluster analysis results; and identifying key risk factors based on the cluster analysis results.

[0044] The beneficial effects are as follows: This invention incorporates consideration of the similarity (i.e., time value of electricity) between the segmented curves of the same curve segment and the benchmark price segmented curve during curve clustering. It uses a two-way squeeze principle to calculate the similarity of the segmented clearing price curve shapes, then employs cluster analysis based on shape similarity to perform cluster analysis on the clearing price curves. Finally, it identifies key risk factors based on the cluster analysis results. This invention can accurately identify key risk factors from spot market clearing price curves even when their outlines vary, improving the accuracy of key risk factor identification. By accurately identifying key risk factors and focusing on and promptly managing these factors that affect price risk, the time and manpower required to identify the source of risk when severe price risks occur are reduced, thus lowering the possibility of prolonged damage to the power system and improving power system security. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating a key risk factor identification method provided in an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of a key risk factor identification device provided in an embodiment of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Reference Figure 1 , Figure 1 This is a flowchart illustrating a key risk factor identification method according to an embodiment of the present invention, including S101-S104:

[0049] S101: Based on the spot market clearing price curves during the historical statistical period, obtain multiple curves of the same segment and generate a set of curves of the same segment.

[0050] Among them, the curve of the same segment is the spot market clearing price curve of the same period within the historical statistical period.

[0051] S102: Calculate the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve based on the segmented two-way squeeze model.

[0052] In this embodiment, the method further includes: obtaining a segmented curve of the benchmark price, specifically:

[0053] The spot market clearing price curves within the historical statistical period are normalized to generate a normalized first curve set.

[0054] Select any one curve from the first set of curves as the benchmark price curve;

[0055] The benchmark price segmented curve is obtained based on the benchmark price curve.

[0056] In one specific embodiment, the step of calculating the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve based on the segmented two-way squeeze model specifically involves:

[0057] Based on the curve downward shift coefficient, the curve upward shift coefficient, the benchmark price segmented curve, and any segmented curve in the set of segmented curves, set the constraints and objective function of the segmented two-way squeeze model;

[0058] Solve the objective function under the constraints to obtain the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve.

[0059] S103: Generate overall similarity based on the similarity of the piecewise curve shapes, and perform cluster analysis based on the overall similarity to generate cluster analysis results.

[0060] In this embodiment, the step of generating overall similarity based on the similarity of the piecewise curve shapes, and performing cluster analysis based on the overall similarity to generate cluster analysis results, specifically includes:

[0061] Based on the similarity of the piecewise curve shapes, a similarity set is generated;

[0062] Based on the segmented two-way squeeze model and the similarity set, the overall similarity is calculated; wherein, the overall similarity refers to the similarity between the benchmark price curve and the non-benchmark price curve in the first curve set;

[0063] Cluster analysis is performed based on overall similarity to generate the cluster analysis results.

[0064] In one specific embodiment, the step of performing cluster analysis based on overall similarity to generate the cluster analysis results specifically includes:

[0065] Generate a similarity vector set based on the overall similarity;

[0066] Based on the preset number of neighborhood elements, unprocessed vectors and core vectors are obtained from the similarity vector set;

[0067] Calculate the Mahalanobis distance between the unprocessed vector and the core vector;

[0068] Based on the Mahalanobis distance, the density-connected unprocessed vectors are grouped into the same cluster, and the density-connected unprocessed vectors and the core vector are marked as processed vectors;

[0069] Based on the Mahalanobis distance, the unprocessed vectors whose densities are not connected are marked as unclassified vectors;

[0070] Multiple clustering analysis results are generated based on the processed vectors and the unclassified vectors; wherein the processed vectors belong to different clusters.

[0071] S104: Identify key risk factors based on cluster analysis results.

[0072] In this embodiment, the identification of key risk factors based on the cluster analysis results specifically includes:

[0073] Obtain the first risk factor from the processed vectors under different cluster classes;

[0074] If the daily variation value of the first risk factor is greater than a preset threshold, the first risk factor is determined to be the key risk factor.

[0075] In this embodiment, it also includes:

[0076] Obtain the second risk factor from the processed vectors within the same cluster class;

[0077] If the daily variation value of the second risk factor is greater than the preset threshold, the second risk factor is determined to be the key risk factor.

[0078] In a preferred embodiment, the specific process of the key risk factor identification method is further described, including S201-S209:

[0079] S201: Statistically analyze the electricity spot market clearing price curve for historical statistical period T days, t = 1, 2, ..., T, where the electricity spot market clearing price curve for a particular day is represented by l. t express.

[0080] S202: Because the electricity spot market clearing price curve fluctuates throughout the 24-hour period, it is divided into N time periods, denoted by n (n = 1, 2, ..., N), using methods such as manual division or division based on peak, flat, and valley periods. The nth segment of the electricity spot market clearing price curve on day t is represented by l. t，n express.

[0081] S203: Normalize the T-line spot market curve.

[0082] One of the normalization methods is:

[0083] l t，min =min(l t，1 lt ，2 , ..., l t，n , ...l t，N );

[0084] l t，max =max(l t，1 , l t，2 , ...l t，n , ..., l t，N );

[0085]

[0086] Among them, l t，n For the nth segment of the spot market clearing price curve on day t, l t，min Let l be the lowest price segment in the N segments of the spot market clearing price curve on day t. t，max p represents the highest price segment in the N segments of the spot market clearing price curve on day t. t，n This is the nth segment of the curve after normalization on day t.

[0087] After normalization, the spot market clearing price curve for day t is represented by p. t , indicating that each segment n is represented by p t , n express:

[0088] p t =[p t，1 p t，2 , ..., p t，n , ..., p t，N ];

[0089] Since the spot market clearing price curves for the historical statistical period T were statistically analyzed, after normalizing all the clearing price curves, a normalized set of curves (i.e., the first set of curves) P is obtained:

[0090] P={p1,p2,…,pt,…,p T}:

[0091] S204: Divide all curves of the same segment on day T into a set (i.e., the set of curves of the same segment), and use U to represent the set of the nth segment of the curve. n express:

[0092] U n ={p 1，n p 2，n, ..., p t，n , ..., p T，n};

[0093] S205: Select any curve from set P as the benchmark price curve.

[0094] To select p T For example, then p constitutes T p T，1 P T，2 , ..., p T，n , ..., P T，N These are the segment sets U1, U2, ..., U... n , ..., U N The benchmark segmented price curve.

[0095] S206: Construct a segmented two-way squeeze model to analyze the similarity of curve shapes among all curves in the same segment on all T days.

[0096] Furthermore, S206 includes: S2061-S2065:

[0097] S2061: Select any segment n and analyze the set of curves U of the same segment. n Similarity in the shape of all curves;

[0098] S2062: Calculation of the set U of curves in the same segment based on a piecewise bidirectional squeeze model n Similarity in shape between any segmented price curve and the benchmark segmented price curve:

[0099] Select the set U of curves of the same segment n Any segment of curve p in t，n Curve downward shift coefficient a t，n Curve upward shift coefficient b t，n Piecewise curve of benchmark price p T，n Set the constraints and objective function of the piecewise bidirectional squeeze model; find the objective function of the piecewise bidirectional squeeze model, and let the final objective function value be d. t，n d t，n This represents the morphological similarity between curves. The solution method is as follows:

[0100] Objective function: min d t，n ;

[0101] Constraints: d t，n =b t，n -a t，n stb t，n >a t，n ;

[0102] a t，n pT，n ≤p t，n ≤b t，n p T，n ;

[0103] Among them, a t，n p T，n The nth segment of the curve p, segmented from the benchmark price curve T，n Same shape and located at p T，n The curve below; b t，n p T，n The nth segment of the curve p, segmented from the benchmark price curve T，n Same shape and located at p T，n The curve above; d t The difference between the downward shift coefficient and the upward shift coefficient of the curve is the smaller the difference, indicating that a... t，n and b t，n The closer, a t，n p T，n and b t，n p T，n The closer the two curves are to each other, the more likely they are to be pt ，n When d t，n When = 0, the objective function reaches its optimal result, i.e., curve p t，n and benchmark segmented price curve p T，n Total overlap.

[0104] S2063: Determine whether to traverse the set U of curves in the same segment. n If all elements in the list are true, proceed to step S2064; otherwise, return to step S2062.

[0105] S2064: The set (i.e., similarity set) of the objective function values ​​of the (T-1) segmented bidirectional squeeze forcing of the nth segment of the electricity spot market clearing price curve. n D n ={d 1，n d 2，n , ..., d t，n , ..., d t-1，n This set represents the similarity between the nth clearing price curve of the electricity spot market on day T and the benchmark segmented price curve over a historical period of T-1 days, after using the nth segment clearing price curve of the electricity spot market on day T as the base segmented price curve. The smaller d is, the more similar the curve shapes are.

[0106] S207: Obtain N sets D1, D2, ..., D''''''''''''''''''''''''" '"N' ... n-1 D n :

[0107] D n ={d1，n d 2，n , ...,d t，n , ...,d T-1，n};

[0108] D n-1 ={d 1，n-1 d 2，n-1 , ...,d t，n-1 , ...,d T-1，n-1};

[0109] …

[0110] D1={d 1，1 d 2，1 , ...,d t，1 , ...,d T-1，1};

[0111] Therefore, the overall similarity S between the electricity spot market clearing price curve and the benchmark price curve on day T within the historical statistical period (T-1) is... t The target value can be represented by the piecewise bidirectional squeeze model:

[0112] S T-1 =[d T-1，1 d T-1，2 , ...,d T-1，n , ...,d T-1，N ];

[0113] S T-2 =[d T-2，1 d T-2，2 , ...,d T-2，n , ...,d T-2，N ];

[0114] …

[0115] S1=[d 1，1 d 1，2 , ...,d 1，n , ...,d 1，N ];

[0116] S T-1 S T-2 S1 represents the set of segment similarity representations of all segmented curves for each day in the past T-1 days and the benchmark segmented price curve for the benchmark day T.

[0117] Because p T As a benchmark price curve, therefore p T Similarity to itself S T A matrix (i.e., a set of similarity vectors) can be represented as:

[0118] S T= [0, 0, ..., 0, ..., 0];

[0119] S208: For S1, S2...S T-2 S T-1 S T Cluster analysis is performed on the similarity vector set.

[0120] Cluster analysis methods include hierarchical clustering and partitioning clustering. Taking density-based clustering as an example within partitioning clustering:

[0121] (1) Set the density clustering neighborhood radius d and the minimum number of elements in the neighborhood MinNum;

[0122] (2) From S1, S2...S T-2 S T-1 S T Choose any vector from the options;

[0123] (3) Determine whether the vector is a core point. A core point is defined as a vector whose neighborhood radius d contains at least MinNum vectors;

[0124] 1) If so, then execute (5);

[0125] 2) If not, then execute (4);

[0126] (4) Search S1, S2...S T-2 S T-1 S T All unprocessed vectors within;

[0127] (5) Calculate the distance between the unprocessed vector and the core vector;

[0128] The distance between any two vectors can be calculated using various methods such as Euclidean distance, Manhattan distance, Chebyshev distance, Minkowski distance, and Mahalanobis distance. Here, we will use Mahalanobis distance as an example to calculate the distance between vectors:

[0129] S t and S t+1 Mahalanobis distance D M，n The calculation method is as follows:

[0130]

[0131]

[0132] Among them, S t Let p be the clearing price curve and benchmark price curve for the electricity spot market on day t. T The matrix formed by the piecewise bidirectional squeeze target values; S t+1The electricity spot market clearing price curve and benchmark price curve p for day t+1 T The matrix formed by the piecewise bidirectional squeeze target values; Λ is S t and S t+1 The covariance matrix.

[0133] (6) Determine whether the vector belongs to direct density reachable / density reachable / density connected based on the calculation result of Mahalanobis distance;

[0134] 1) If so, proceed to step (7);

[0135] 2) If not, return to step (4);

[0136] (7) Group vectors that meet (6) into the same cluster;

[0137] (8) Core vectors and categorized vectors are marked as processed vectors;

[0138] (9) Has the current search for this core vector ended?

[0139] 1) If so, proceed to step (10);

[0140] 2) If not, return to step (2);

[0141] (10) Output all core vectors, clusters (including the processed vectors corresponding to each cluster) and unclassified vectors as the cluster analysis results.

[0142] S209: Identify key risk factors based on cluster analysis results.

[0143] (1) Compare the multidimensional risk factors (i.e. the second risk factor) corresponding to the vectors under the same cluster. If the daily difference value of a certain risk factor in the multidimensional risk factors is greater than the preset threshold, then the risk factor is determined to be a non-critical risk factor.

[0144] For example, regarding the risk factor of primary energy prices, a difference of more than 30% between daily primary energy prices is considered a significant difference. Market operators can set this threshold level based on the actual situation of the electricity market in their province. The risk factors mentioned in this embodiment refer to factors that affect the clearing price of the electricity spot market. For example, primary energy prices, supply-demand ratios, weather, and working days can all be considered risk factors. For quantifiable risk factor values, a percentage threshold can be used to measure their differences. For non-quantifiable risk factor values, a zero-one variable can be used to measure their differences, meaning there are only two possible outcomes: difference or no difference.

[0145] By comparing the multidimensional risk factors corresponding to the vectors in the same cluster, if the daily difference of the risk factor value of a certain dimension exceeds the pre-set threshold, it means that even if the risk factor value changes drastically, the cluster analysis results are still classified into the same category. Therefore, it is determined that the risk factor belongs to a non-critical risk factor.

[0146] (2) Compare the risk factors (i.e. the first risk factor) of a certain dimension corresponding to the vector under different clusters. If the daily difference value of a certain risk factor is greater than the preset threshold, then the risk factor is determined to be a key risk factor.

[0147] By comparing the risk factors of a certain dimension corresponding to the vectors under different clusters, if the risk factor values ​​differ significantly from day to day, it indicates that the drastic changes in the risk factor values ​​have led to the cluster analysis results being classified into different categories. Therefore, it is determined that the risk factor belongs to the key risk factor.

[0148] To further explain the key risk factor identification device, please refer to... Figure 2 , Figure 2 This is a schematic diagram of the structure of a key risk factor identification device provided in an embodiment of the present invention, including: an acquisition module 201, a calculation module 202, an analysis module 203, and an identification module 204;

[0149] The acquisition module 201 is used to acquire multiple segment curves and generate a set of segment curves based on the spot market clearing price curves within the historical statistical period; wherein, the segment curves are the spot market clearing price curves for the same period within the historical statistical period.

[0150] The calculation module 202 is used to calculate the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve according to the segmented two-way squeeze model.

[0151] The analysis module 203 is used to generate an overall similarity based on the similarity of the piecewise curve shapes, and to perform cluster analysis based on the overall similarity to generate cluster analysis results;

[0152] The identification module 204 is used to identify key risk factors based on the cluster analysis results.

[0153] In this embodiment, the method further includes: obtaining a segmented curve of the benchmark price, specifically:

[0154] The spot market clearing price curves within the historical statistical period are normalized to generate a normalized first curve set.

[0155] Select any one curve from the first set of curves as the benchmark price curve;

[0156] The benchmark price segmented curve is obtained based on the benchmark price curve.

[0157] In this embodiment, the step of calculating the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve based on the segmented two-way squeeze model specifically involves:

[0158] Based on the curve downward shift coefficient, the curve upward shift coefficient, the benchmark price segmented curve, and any segmented curve in the set of segmented curves, set the constraints and objective function of the segmented two-way squeeze model;

[0159] Solve the objective function under the constraints to obtain the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve.

[0160] In this embodiment, the step of generating overall similarity based on the similarity of the piecewise curve shapes, and performing cluster analysis based on the overall similarity to generate cluster analysis results, specifically includes:

[0161] Based on the similarity of the piecewise curve shapes, a similarity set is generated;

[0162] Based on the segmented two-way squeeze model and the similarity set, the overall similarity is calculated; wherein, the overall similarity refers to the similarity between the benchmark price curve and the non-benchmark price curve in the first curve set;

[0163] Cluster analysis is performed based on overall similarity to generate the cluster analysis results.

[0164] In one specific embodiment, the step of performing cluster analysis based on overall similarity to generate the cluster analysis results specifically includes:

[0165] Generate a similarity vector set based on the overall similarity;

[0166] Based on the preset number of neighborhood elements, unprocessed vectors and core vectors are obtained from the similarity vector set;

[0167] Calculate the Mahalanobis distance between the unprocessed vector and the core vector;

[0168] Based on the Mahalanobis distance, the density-connected unprocessed vectors are grouped into the same cluster, and the density-connected unprocessed vectors and the core vector are marked as processed vectors;

[0169] Based on the Mahalanobis distance, the unprocessed vectors whose densities are not connected are marked as unclassified vectors;

[0170] Multiple clustering analysis results are generated based on the processed vectors and the unclassified vectors; wherein the processed vectors belong to different clusters.

[0171] In one specific embodiment, identifying key risk factors based on the cluster analysis results specifically includes:

[0172] Obtain the first risk factor from the processed vectors under different cluster classes;

[0173] If the daily variation value of the first risk factor is greater than a preset threshold, the first risk factor is determined to be the key risk factor.

[0174] In one specific embodiment, it further includes:

[0175] Obtain the second risk factor from the processed vectors within the same cluster class;

[0176] If the daily variation value of the second risk factor is greater than the preset threshold, the second risk factor is determined to be the key risk factor.

[0177] A specific embodiment of the present invention provides a mobile terminal, including a processor and a memory, wherein the memory stores computer-readable program code, and the processor executes the computer-readable program code to implement the steps of the above-described key risk factor identification method.

[0178] A specific embodiment of the present invention provides a storage medium that stores computer-readable program code, which, when executed, implements the steps of the above-described method for identifying key risk factors.

[0179] In this embodiment of the invention, the acquisition module obtains multiple segmented curves based on the spot market clearing price curves within a historical statistical period and generates a set of segmented curves; wherein, the segmented curves are the spot market clearing price curves for the same period within the historical statistical period; the calculation module calculates the segmented curve shape similarity between each segmented curve in the set of segmented curves and the benchmark price segmented curve based on a segmented two-way squeeze model; the analysis module generates an overall similarity based on the segmented curve shape similarity and performs cluster analysis based on the overall similarity to generate cluster analysis results; and the identification module identifies key risk factors based on the cluster analysis results.

[0180] This invention incorporates consideration of the similarity (i.e., time value of electricity) between the segmented curves of the same curve segment and the benchmark price segmented curve during curve clustering. It uses a two-way squeeze principle to calculate the similarity of the segmented clearing price curve shapes, then employs cluster analysis based on shape similarity to cluster the clearing price curves. Finally, it identifies key risk factors based on the cluster analysis results. This invention can accurately identify key risk factors from spot market clearing price curves even when their outlines vary, improving the accuracy of key risk factor identification. By accurately identifying key risk factors and focusing on and promptly managing these factors that affect price risk, it reduces the time and manpower required to identify the source of risk when severe price risks occur, thereby reducing the possibility of prolonged damage to the power system and improving power system security.

[0181] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for identifying key risk factors, characterized in that, include: Based on the spot market clearing price curves within the historical statistical period, multiple curves of the same segment are obtained and a set of curves of the same segment is generated; wherein, the curves of the same segment are the spot market clearing price curves for the same period within the historical statistical period; The similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve is calculated based on the segmented two-way squeeze model; wherein, the constraints and objective function of the segmented two-way squeeze model are set based on the curve downward shift coefficient, the curve upward shift coefficient, the benchmark price segmented curve, and any segmented curve in the set of segmented curves; An overall similarity is generated based on the piecewise curve shape similarity, and cluster analysis is performed based on the overall similarity to generate cluster analysis results; wherein, the overall similarity is generated through the piecewise bidirectional squeeze model; the cluster analysis results are determined by calculating the Mahalanobis distance between the unprocessed vector and the core vector; the unprocessed vector and the core vector are obtained based on the overall similarity; Key risk factors are identified based on the clustering analysis results; wherein the key risk factors include a first risk factor and a second risk factor; the first risk factor is obtained from processed vectors under different clusters; the second risk factor is obtained from processed vectors under the same cluster; the processed vectors are determined based on the unprocessed vectors and the core vectors.

2. The method of claim 1, wherein, Also includes: To obtain the benchmark price segmented curve, specifically: The spot market clearing price curves within the historical statistical period are normalized to generate a normalized first curve set. Select any one curve from the first set of curves as the benchmark price curve; The benchmark price segmented curve is obtained based on the benchmark price curve.

3. The method of claim 2, wherein, The step of calculating the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve based on the segmented two-way squeeze model is as follows: Based on the curve downward shift coefficient, the curve upward shift coefficient, the benchmark price segmented curve, and any segmented curve in the set of segmented curves, set the constraints and objective function of the segmented two-way squeeze model; Solve the objective function under the constraints to obtain the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve.

4. The method of claim 3, wherein, The process of generating an overall similarity based on the similarity of the piecewise curve shapes, and performing cluster analysis based on the overall similarity to generate cluster analysis results, specifically involves: Based on the similarity of the piecewise curve shapes, a similarity set is generated; Based on the segmented two-way squeeze model and the similarity set, the overall similarity is calculated; wherein, the overall similarity refers to the similarity between the benchmark price curve and the non-benchmark price curve in the first curve set; Cluster analysis is performed based on overall similarity to generate the cluster analysis results.

5. The method of claim 4, wherein, The step of performing cluster analysis based on overall similarity to generate the cluster analysis results is as follows: Generate a similarity vector set based on the overall similarity; Based on the preset number of neighborhood elements, unprocessed vectors and core vectors are obtained from the similarity vector set; Calculate the Mahalanobis distance between the unprocessed vector and the core vector; Based on the Mahalanobis distance, the density-connected unprocessed vectors are grouped into the same cluster, and the density-connected unprocessed vectors and the core vector are marked as processed vectors; Based on the Mahalanobis distance, the unprocessed vectors whose densities are not connected are marked as unclassified vectors; Multiple clustering analysis results are generated based on the processed vectors and the unclassified vectors; wherein the processed vectors belong to different clusters.

6. The method for identifying key risk factors according to claim 5, characterized in that, The identification of key risk factors based on the cluster analysis results specifically includes: Obtain the first risk factor from the processed vectors under different cluster classes; If the daily variation value of the first risk factor is greater than a preset threshold, the first risk factor is determined to be the key risk factor.

7. The method of claim 6, wherein, Also includes: Obtain the second risk factor from the processed vectors within the same cluster class; If the daily variation value of the second risk factor is greater than the preset threshold, the second risk factor is determined to be the key risk factor.

8. A critical risk factor identification apparatus characterized by, include: The module includes an acquisition module, a calculation module, an analysis module, and a recognition module. The acquisition module is used to acquire multiple segmented curves and generate a set of segmented curves based on the spot market clearing price curves within the historical statistical period; wherein, the segmented curves are the spot market clearing price curves for the same period within the historical statistical period. The calculation module is used to calculate the similarity of the segmented curve shape between each segmented curve in the set of segmented curves and the benchmark price segmented curve, according to the segmented two-way squeeze model; wherein, the constraints and objective function of the segmented two-way squeeze model are set based on the curve downward shift coefficient, the curve upward shift coefficient, the benchmark price segmented curve, and any segmented curve in the set of segmented curves; The analysis module is used to generate an overall similarity based on the piecewise curve shape similarity, and to perform cluster analysis based on the overall similarity to generate cluster analysis results; wherein, the overall similarity is generated through the piecewise bidirectional squeeze model; the cluster analysis results are determined by calculating the Mahalanobis distance between the unprocessed vector and the core vector; the unprocessed vector and the core vector are obtained based on the overall similarity; The identification module is used to identify key risk factors based on the clustering analysis results; wherein, the key risk factors include a first risk factor and a second risk factor; the first risk factor is obtained from processed vectors under different clusters; the second risk factor is obtained from processed vectors under the same cluster; the processed vectors are determined based on the unprocessed vectors and the core vectors.

9. A mobile terminal, characterized by The device includes a processor and a memory, the memory storing computer-readable program code, and the processor executing the computer-readable program code to implement the steps of a key risk factor identification method according to any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores computer-readable program code, which, when executed, implements the steps of a key risk factor identification method according to any one of claims 1 to 7.

Images