Electric power system-based global risk intelligent early warning and disposal method and system

Abstract

The embodiment of the invention provides a global risk intelligent early warning and disposal method and system based on a power system, and belongs to the technical field of risk early warning and disposal. The risk intelligent early warning and disposal method comprises the steps that various feature data, including material purchasing feature data, marketing service feature data and engineering project feature data, of a power system are acquired; preprocessing each kind of feature data to construct a training set of each kind of feature data; respectively constructing an early warning model of each kind of feature data; training a corresponding early warning model by adopting the training set of each kind of feature data; according to the method, the early warning models of various feature data are adopted to perform various risk identification and diagnosis on the power system, so that the adaptability and precision of early warning of the power system are effectively improved, and the reliability and efficiency of risk disposal are improved.

Description

A method and system for intelligent early warning and handling of risks across the entire power system Technical Field

[0001] This invention relates to the field of risk warning and response technology, specifically to a method and system for intelligent early warning and response to risks across the entire power system. Background Technology

[0002] As power companies advance their digital transformation, core business scenarios such as material procurement, marketing services, and engineering projects generate massive amounts of business data. Monitoring this data can effectively identify risks and provide timely warnings, thereby improving the security and reliability of business processes.

[0003] Traditional supervision models have significant shortcomings: First, early warning models have poor adaptability, often setting fixed rules for single business scenarios (such as monitoring material risks only through "purchase amount thresholds"), which cannot cope with complex risks across scenarios (such as "the same supplier being involved in both material bidding and project subcontracting"). Furthermore, when business processes are optimized (such as adjustments to the business expansion application process) or policies are updated (such as the release of new bidding management regulations), rules need to be manually redeveloped, resulting in long response cycles. Second, risk assessment lacks quantitative standards, relying on the subjective experience of supervisors, leading to inconsistent handling results for similar suspicious points, and high false alarm and false alarm rates. Third, the handling process is fragmented, with early warnings for each business scenario circulating independently without a unified closed-loop mechanism, resulting in problems such as "repeated verification and overdue handling," increasing the workload of grassroots supervisors and making it difficult to meet the construction requirements of a supervision platform for "coordinated and intelligent re-supervision."

[0004] In the process of realizing this invention, the inventors of this application discovered that the above-mentioned solutions in the prior art have defects such as poor early warning adaptability, poor accuracy, and low efficiency. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for intelligent early warning and handling of risks across the entire power system. This method and system improves the adaptability, accuracy, and efficiency of early warning.

[0006] To achieve the above objectives, embodiments of the present invention provide a method for intelligent early warning and handling of risks across the entire power system, comprising: acquiring multiple characteristic data of the power system, wherein the multiple characteristic data include material procurement characteristic data, marketing service characteristic data, and engineering project characteristic data; preprocessing each type of characteristic data to construct a training set for each type of characteristic data; constructing an early warning model for each type of characteristic data; training the corresponding early warning model using the training set for each type of characteristic data; acquiring multiple real-time characteristic data of the current power system; and obtaining and handling the risk level of the current power system based on the multiple real-time characteristic data of the current power system and the corresponding early warning model.

[0007] Optionally, preprocessing the various feature data includes: handling missing values ​​in the various feature data; handling outliers in the various feature data; and performing feature standardization on the various feature data.

[0008] Optionally, constructing a training set for each type of feature data includes: obtaining the repetition rate of bid document IPs in the material procurement feature data according to formula (1). (1) Among them, This refers to the duplication rate of bid document IPs in the material procurement feature data. The number of unique bid documents with IP addresses within the same project. The number of IPs for all bid documents in the same project; the document creation time difference of the bid documents in the material procurement feature data is obtained according to formula (2). (2) Among them, Create a time difference for the tender documents. , These are the earliest and latest document creation times of the tender documents, respectively; the supplier correlation in the material procurement characteristic data is obtained according to formula (3). (3) Among them, For suppliers and suppliers The degree of correlation, , , These are equity weight, personnel weight, and historical behavior weight. , , The scores are respectively based on equity association, personnel association, and historical behavior association; the duplication rate of the tender document IP, the document creation time difference, and the supplier association are labeled, and a material procurement training set is constructed.

[0009] Optionally, constructing an early warning model for each type of feature data includes: constructing a material procurement early warning model using a random forest algorithm.

[0010] Optionally, constructing a training set for each type of feature data includes: the process time for obtaining the marketing service feature data according to formula (4). (4) Among them, The process duration for the aforementioned marketing service feature data. The end time of the marketing service process. The start time of the marketing service; the completeness of the approval records in the marketing service feature data is obtained according to formula (5). (5) Among them, For the first The completeness of each approval record For the first The number of completed approval records. For the first The standard completion number of each approval record The process duration and the completeness of the approval records are labeled, and a marketing service training set is constructed.

[0011] Optionally, constructing an early warning model for each type of feature data includes: constructing a marketing service early warning model using a logistic regression algorithm.

[0012] Optionally, constructing a training set for each type of feature data includes: obtaining the acceptance-to-fund transfer time difference in the project feature data according to formula (6). (6) Among them, This refers to the time difference between acceptance and asset transfer in the characteristic data of the aforementioned engineering project. For the transfer time of engineering projects, The acceptance time of the project is determined; the project amount in the characteristic data of the project is obtained; the time difference between acceptance and capital transfer and the project amount are labeled, and a training set of the project is constructed.

[0013] Optionally, constructing an early warning model for each type of feature data includes: constructing an early warning model for engineering projects using the gradient boosting tree algorithm.

[0014] Optionally, obtaining and handling the risk level of the current power system based on various real-time characteristic data of the current power system and the corresponding early warning model includes: inputting various real-time characteristic data of the current power system into the corresponding early warning model to output risk characteristics and risk probabilities; obtaining the degree of impact and rectification difficulty corresponding to the risk characteristics; obtaining the risk value according to formula (7), (7) Among them, This is the risk value. , , These are the impact weight, rectification weight, and probability weight, respectively. The degree of impact corresponding to the aforementioned risk characteristics, The difficulty of rectification corresponding to the aforementioned risk characteristics. The process involves: scoring the probability of risk; obtaining the corresponding risk level based on the risk value; determining whether the risk level is high-risk; taking immediate measures if the risk level is high-risk; determining whether the risk level is medium-risk if the risk level is not high-risk; setting a deadline for rectification if the risk level is medium-risk; and issuing an early warning if the risk level is not medium-risk.

[0015] On the other hand, the present invention also provides a power system-wide risk intelligent early warning and handling system, comprising: a data acquisition module connected to the power system for acquiring characteristic data of the power system; and a controller connected to the data acquisition module for executing any of the above-described risk intelligent early warning and handling methods.

[0016] Through the above technical solution, the intelligent early warning and handling method and system based on the overall risk of the power system provided by this invention acquires multiple feature data of the power system, preprocesses each feature data to obtain a corresponding training set, and then constructs an early warning model for each feature data. This model is then trained using the corresponding training set to identify and diagnose multiple real-time feature data of the current power system, thereby determining the risk level of the current power system and taking corresponding handling measures. By employing early warning models based on multiple feature data to identify and diagnose multiple risks in the power system, the adaptability and accuracy of power system early warning are effectively improved, thereby enhancing the reliability and efficiency of risk handling.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the embodiments of the present invention and constitute a part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the drawings: Figure 1 is a flowchart of a method for intelligent early warning and handling of risks across the entire power system according to an embodiment of the present invention; Figure 2 is a flowchart of feature data preprocessing in the method for intelligent early warning and handling of risks across the entire power system according to an embodiment of the present invention; Figure 3 is a flowchart of constructing a training set in the method for intelligent early warning and handling of risks across the entire power system according to an embodiment of the present invention; Figure 4 is a flowchart of obtaining and handling risk levels in the method for intelligent early warning and handling of risks across the entire power system according to an embodiment of the present invention. Detailed Implementation

[0019] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0020] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0021] Figure 1 is a flowchart of a power system-wide risk intelligent early warning and handling method according to an embodiment of the present invention. In Figure 1, the risk intelligent early warning and handling method may include: in step S1, acquiring various characteristic data of the power system, including material procurement characteristic data, marketing service characteristic data, and engineering project characteristic data. Specifically, characteristic data can be acquired from corresponding systems via API interfaces. For example, in a material scenario, tender document characteristics can be acquired from ECP, and purchase order data from ERP; in a marketing scenario, application work orders can be acquired from marketing, and meter reading data can be acquired from the electricity consumption data collection system; in an engineering scenario, project progress can be acquired from the ERP project management module, and acceptance reports can be acquired from the engineering audit system.

[0022] In step S2, each type of feature data is preprocessed to construct a training set for that type of feature data. Preprocessing can be performed on each type of feature data to improve the reliability and accuracy of the data, thereby enabling the construction of a reliable training set.

[0023] In step S3, a warning model is constructed for each type of feature data. Specifically, corresponding warning models can be constructed for material procurement, marketing difficulties, and engineering projects to improve the accuracy and effectiveness of identifying and issuing warnings for various risks.

[0024] In step S4, the corresponding early warning model is trained using the training set of each feature data.

[0025] In step S5, various real-time characteristic data of the current power system are acquired.

[0026] In step S6, the risk level of the current power system is obtained and addressed based on various real-time characteristic data and corresponding early warning models. Specifically, by inputting the real-time characteristic data of the current power system into the corresponding early warning models, the corresponding risk labels and probabilities are obtained, thus clarifying the current risk level. Based on the current risk level of the power system, targeted and appropriate measures can be taken.

[0027] In steps S1 to S6, material procurement characteristic data, marketing service characteristic data, and engineering project characteristic data from the power system are first acquired and preprocessed to construct training sets for each type of characteristic data. Simultaneously, early warning models for each type of characteristic data are constructed and trained using the corresponding training sets to obtain effective early warning models. Furthermore, various real-time characteristic data of the current power system can be collected and input into the effective early warning model to obtain the current risk type and probability. Based on this risk type and probability, the risk level can be determined, and corresponding measures can be taken to ensure the reliable operation of the power system.

[0028] Traditional power system business data monitoring models suffer from poor early warning adaptability, reliance on the subjective experience of supervisors leading to low early warning accuracy and inefficient early warning response. In this embodiment of the invention, early warning models employing multiple feature data are used to perform various risk identification and diagnosis methods on the power system, effectively improving the adaptability and accuracy of power system early warnings, thereby enhancing the reliability and efficiency of risk response.

[0029] In this embodiment of the invention, after obtaining multiple feature data, it is necessary to preprocess the multiple feature data. The specific preprocessing steps are shown in Figure 2. Specifically, in Figure 2, the feature data preprocessing may include: in step S20, handling missing values ​​for multiple feature data. When key fields (such as project amount, supplier name) are missing, a completion request is automatically pushed to the business department, with a response time of ≤24 hours.

[0030] In step S21, outlier processing is performed on various feature data. Among them, the 3σ principle is used to identify data deviations (such as a single purchase amount exceeding three times the average of similar items), which are then marked and triggered for manual review.

[0031] In step S22, feature standardization processing is performed on various feature data. Specifically, numerical features (such as purchase amount and process duration) are normalized to the [0,1] interval, and textual features (such as supplier relationships) are converted into vectors through One-Hot encoding.

[0032] In this embodiment of the invention, the risks of the power system are divided into three categories: material procurement scenarios, marketing service scenarios, and engineering project scenarios. The characteristic data for these three scenarios correspond to material procurement characteristic data, marketing service characteristic data, and engineering project characteristic data, respectively. Specifically, the risk types / labels for these three scenarios can be further subdivided. For example, the material procurement scenario can include nine risks: excessively high frequency of supermarket-style procurement, risks of supermarket-style procurement exceeding specifications, risks of bid rigging and collusion, risks of illegal invitation to tender, risks of excessively high winning bid prices, risks of materials not being returned when they should be returned, risks of dismantling old materials, risks of frequent transactions with the same related party, and risks of lax supplier management. The marketing service scenario can include eight risks: risks of external transfer of electricity application and installation services, risks of non-compliance with application time limits, risks of corruption related to the "three designations" system, risks of inconsistency between marketing and meter reading data, risks of misappropriation of electricity funds, risks of users exceeding their capacity for a long period of time, risks of electricity theft in marketing, and risks of the authenticity of power supply time for new users. The engineering project scenario can include seven risks: risks of outsourcing core business, risks of illegal subcontracting of engineering projects, risks of false / duplicate projects, risks of long-term non-transfer of construction projects to fixed assets, risks of non-compliance of project acceptance, risks of overdue project final accounts, and risks of inconsistency between project budget and cost.

[0033] In this embodiment of the present invention, after preprocessing the feature data, it is also necessary to transform and label the feature data to construct training sets for material procurement, marketing services, and engineering projects, respectively. The construction of the training sets can be shown in Figure 3. Specifically, in Figure 3, the construction of the training sets may include: in step S23, obtaining the repetition rate of bid document IPs in the material procurement feature data according to formula (1). (1) Among them, This refers to the duplication rate of bid document IPs in the material procurement feature data. The only bid document IP in the same project The number, that is, the number of distinct IPs counted after removing duplicates. This represents the number of IPs (Individual Bidding Authorities) for all bid documents in the same project. Specifically, the lower the repetition rate of bid document IPs, the higher the IP concentration and the greater the risk. For example, if 5 suppliers submit 10 documents for a project, and these 10 documents come from 5 different IPs, the repetition rate is 0.5.

[0034] In step S24, the document creation time difference of the tender documents in the material procurement feature data is obtained according to formula (2). (2) Among them, Create a time difference for the tender documents. , These are the earliest and latest document creation times for the tender documents, respectively. Specifically, the smaller the difference in document creation times, the higher the risk. For example, if five tender documents were created within five minutes of each other, and the project bidding period is eight days, the time is abnormally concentrated, and the risk is extremely high.

[0035] In step S25, the supplier correlation degree in the material procurement feature data is obtained according to formula (3). (3) Among them, For suppliers and suppliers The degree of correlation, , , These are equity weight, personnel weight, and historical behavior weight. , , These are the scores for equity-related relationships, personnel-related relationships, and historical behavior-related relationships. , The scores are integers. Specifically, equity association scores can include cases where two suppliers share the same legal entity, or where one supplier is a shareholder of the other; in such cases, the equity association score is 1, and vice versa. Personnel association scores can include matching key personnel or senior executives of the two suppliers, with the score determined based on the degree of matching. Historical behavior association scores can include obtaining the frequency of joint bids, specifically the proportion of times two suppliers bid simultaneously in past projects out of the total number of bids. Equity weights, personnel weights, and historical behavior weights can be set according to actual needs.

[0036] In step S26, the duplication rate of bid document IPs, document creation time differences, and supplier relevance are labeled, and a material procurement training set is constructed. Specifically, actual risk labels can be added to the above data to construct the material procurement training set. This training set can include bid-rigging cases and normal bidding data.

[0037] In step S27, the process time for obtaining marketing service feature data according to formula (4) is calculated. (4) Among them, The process time for providing feature data for marketing services The end time of the marketing service process. The start time for marketing services. Specifically, the process duration can be adjusted by excluding holidays and rest days, only calculating the duration on weekdays. If the process duration is too short, it indicates that key steps may be skipped; if the process duration is too long, it indicates that key steps are not going smoothly and have been shelved for an extended period.

[0038] In step S28, the completeness of the approval records in the marketing service feature data is obtained according to formula (5). (5) Among them, For the first The completeness of each approval record For the first The number of completed approval records. For the first The standard completion number of each approval record The number is an integer. Specifically, this approval record includes multiple approval stages. That is, the first The number of approval steps completed in each approval record. That is, the first The number of completed approval steps in each approval record.

[0039] In step S29, the process duration and the completeness of approval records are labeled, and a marketing service training set is constructed. Specifically, actual risk labels can be added to the aforementioned data to construct the marketing service training set. This training set can select off-balance-sheet cases and normal application data.

[0040] In step S30, the time difference between acceptance and asset transfer in the project characteristic data is obtained according to formula (6). (6) Among them, This refers to the time difference between acceptance and asset transfer in the characteristic data of engineering projects. For the transfer time of engineering projects, This refers to the acceptance time of the project. Specifically, the time difference between acceptance and capital transfer can be understood as the calculation of the actual capital transfer period. If the time difference between acceptance and capital transfer is too long or exceeds the deadline, the risk is greater.

[0041] In step S31, the project amount is obtained from the project characteristic data. The larger the project amount, the greater the impact of failure to promptly transfer it to fixed assets on the balance sheet and profit and loss statement.

[0042] In step S32, the time difference between acceptance and capitalization and the project amount are labeled, and a training set for engineering projects is constructed. Specifically, actual risk labels can be added to the aforementioned data to construct the training set for engineering projects. This training set can select cases that have not yet been transferred to fixed assets and data from cases that have been transferred to fixed assets.

[0043] In this embodiment of the invention, for material procurement scenarios, a random forest algorithm can be used to construct a material procurement early warning model, such as a "bid rigging and collusion identification model".

[0044] For marketing service scenarios, logistic regression algorithms can be used to build marketing service early warning models, such as the "business expansion and installation off-site circulation model".

[0045] For engineering project scenarios, gradient boosting tree algorithm can be used to build early warning models for engineering projects, such as the "construction in progress model".

[0046] Furthermore, the training process for the above model can include first dividing the data (training set: validation set: test set = 7:2:1), then initializing the model parameters (e.g., random forest tree depth 5, iteration count 100); training the model using the training set, then validating the effect, and if the effect is good, then solidifying the model and ensuring that the model performance meets the standards.

[0047] Furthermore, the aforementioned model can be dynamically iterated and optimized. Iteration triggers can include changes in business rules, a decline in model performance, and the addition of new risk scenarios. Specifically, changes in business rules could include a company issuing new bidding methods that adjust the "invitation to tender limit," requiring updates to the corresponding warning threshold. A decline in model performance could include a false positive rate >10%, a false negative rate >6%, or an accuracy <88% on the validation set for two consecutive weeks. New risk scenarios could include the addition of "risks associated with the dismantling of old materials in new energy projects," requiring the addition of features and samples.

[0048] The iteration methods for the above model can include parameter adjustment. Threshold parameters (such as the application process duration threshold) can be modified online through the threshold management module without retraining. Incremental training can also be performed on newly added features (such as the subsidy amount for new energy projects) using 5000+ new samples, with training time ≤ 2 hours. After iterating the model, it must pass a "historical case backtesting test" (with an accuracy rate ≥ 92% for identifying 200+ known violation cases) before it can be deployed online.

[0049] In this embodiment of the invention, after acquiring various real-time characteristic data of the current power system and the corresponding early warning model, the risk level of the current power system can be determined. The specific steps are shown in Figure 4. Specifically, in Figure 4, the intelligent risk early warning and handling method may further include: in step S60, inputting various real-time characteristic data of the current power system into the corresponding early warning model to output risk characteristics and risk probabilities. The risk characteristics may include risk types, and the risk probability is the probability that the risk may occur.

[0050] In step S61, the degree of impact and the difficulty of rectification corresponding to the risk characteristics are obtained. The degree of impact and the difficulty of rectification can be determined based on the risk type and specific indicators; different risks have different degrees of impact and difficulties of rectification.

[0051] Specifically, the Analytic Hierarchy Process (AHP) can be used to quantify the weights of the three dimensions: impact level, rectification difficulty, and probability of occurrence. See Table 1 for details. Table 1: Weights of the Three Dimensions and Scoring Criteria

[0052] In Table 1, the scores for the degree of impact and the difficulty of rectification can be determined based on specific risk indicator examples, while the probability of occurrence can be determined based on the risk probabilities output by each model. The risk probabilities can be matched with the scoring criteria. For example, if the risk probability is 90%, then it is 90 points.

[0053] In step S62, the risk value is obtained according to formula (7). (7) Among them, This is the risk value. , , These are the impact weight, rectification weight, and probability weight, respectively. The degree of impact (score) corresponding to the risk characteristics. The difficulty of rectification (score) corresponding to the risk characteristics. This is a score representing the probability of risk. Specifically, , , The impact weight, rectification weight, and probability weight can be 0.5, 0.3, and 0.2, respectively.

[0054] In step S63, the corresponding risk level is obtained based on the risk value. Specifically, a risk value greater than or equal to 80 points can be set as high risk, such as bid rigging or large-scale misappropriation of electricity funds; a risk value greater than or equal to 50 points but less than 80 points can be set as medium risk, such as overdue installation applications or supermarket-style procurement exceeding specifications; and a risk value less than 50 points can be set as low risk, such as slight over-purchasing or minor discrepancies in meter readings.

[0055] In step S64, it is determined whether the risk level is high risk.

[0056] In step S65, if the risk level is determined to be high, immediate measures should be taken. For high-risk situations, immediate intervention, focused supervision, and automatic reporting to company-level management (such as the head of risk control and relevant supervisors) are required. If necessary, a special working group should be established. Specifically, measures may include: for extremely high-risk behaviors such as bid rigging, the system can automatically suspend related business processes; requiring the responsible unit to develop a specific rectification plan within 3-5 working days, clearly defining rectification measures, responsible persons, and completion deadlines, and submitting it for senior management approval; including 100% of the cases in the manual review checklist of the audit or risk control department for on-site or off-site verification; and having the company office or risk control department supervise the rectification, with automatic escalation of warnings and accountability for failure to rectify within the deadline.

[0057] Finally, the closed loop is determined after the root cause of the risk has been eliminated, the rectification results have been accepted by the special working group, and the relevant supporting materials have been archived for future reference.

[0058] In step S66, if the risk level is not determined to be high risk, then it is determined whether the risk level is medium risk.

[0059] In step S67, if the risk level is determined to be medium risk, a rectification period is set. For medium risk cases, the task can be pushed to the business department head and departmental risk control personnel for rectification within a specified timeframe. Specifically, measures may include: the system automatically generates a rectification task sheet and assigns it to the responsible personnel; the responsible personnel must complete the rectification within 7-10 working days according to the pre-set standardized rectification guidelines; the rectification results (such as explanations and supporting materials) must be submitted to the departmental risk control personnel for online review through the system; the handling of medium risk matters is included in the departmental performance evaluation and reported regularly at business analysis meetings.

[0060] Finally, after the rectification measures were implemented and the department's risk control officer reviewed and confirmed them, the system task status was updated to "closed," confirming the closed loop.

[0061] In step S68, if the risk level is determined to be not medium risk, an early warning reminder is issued. For low-risk cases, a notification can be sent, allowing for self-management, and the system can directly notify the operational staff and their direct supervisors. Specifically, measures may include: the system sending reminders via to-do tasks, SMS, or office software; requiring responsible personnel to complete rectification or provide a reasonable explanation within 10-15 working days; and the risk control department conducting random checks on the handling results of low-risk matters at a certain percentage (e.g., 10%) to supervise the quality of rectification.

[0062] Finally, the system confirms the closed loop after the responsible personnel submit the handling receipt and the system automatically closes the warning; if it is not handled within the time limit, it will be automatically upgraded to a medium-risk warning.

[0063] By dividing business data in the power system into three scenarios / feature data and building early warning models for each, the adaptability and accuracy of early warnings can be effectively improved. Simultaneously, the closed-loop process shortens the high-risk early warning handling cycle, reduces the workload of repetitive verification at the grassroots level, and improves the work efficiency of supervisory personnel. Furthermore, the archiving of early warning data and handling results forms a "risk feature library + handling template" (such as a "bid-rigging and collusion handling process template"), shortening the training cycle for newly hired supervisory personnel and reducing reliance on the "experience of veteran employees."

[0064] In this embodiment of the present invention, the risk warning case in the material procurement scenario can be as follows: I. Risk of excessively high frequency of supermarket-style procurement in the material procurement scenario: Ⅰ. Failure to establish asset cards for accepted materials: 1. Obtain ERP acceptance form information, acceptance time, and asset card.

[0065] 2. Compare and analyze whether asset cards have been established for the materials on the acceptance forms, and determine whether asset cards have not been established even if the acceptance time is >30 days.

[0066] II. Material procurement abnormally exceeds the average procurement value of previous years: 1. Obtain the ECP purchase requisition information of the previous year, including the name and quantity of materials.

[0067] 2. Obtain the purchase quantity of the materials in the previous two years and calculate the average.

[0068] 3. Compare the previous year's material procurement quantity with the average value. If the difference exceeds 10%, an early warning will be issued.

[0069] II. Risks of lax supplier management in material procurement scenarios: I. Determining whether "shadow companies" have not been truthfully reported: 1. Obtaining information on leaders / employees and their relatives (generally only spouses, parents, children, parents-in-law, and siblings are maintained) from the human resources system.

[0070] 2. Obtain information on companies where leaders / employees and their relatives are listed as legal representatives or shareholders on platforms such as the National Enterprise Credit Information Publicity System, Qichacha, and Tianyancha, and ensure that the companies' business scope is related to State Grid's business and their registered addresses are in the city where their employers are located.

[0071] 3. Compare and analyze whether there are any businesses or enterprises that have not been reported. If so, issue a warning.

[0072] II. Intending to or already using one's position to seek personal gain: 1. Obtaining information about leaders / employees and their relatives in the human resources system (generally only spouses, parents, children, parents-in-law, and siblings are maintained).

[0073] 2. Obtain information on companies where the supervisor / employee and their relatives are listed as legal representatives or shareholders on platforms such as the National Enterprise Credit Information Publicity System, Qichacha, and Tianyancha, and that the companies' business scope is related to State Grid's business, with their registered addresses in the city where the supervisor / employee is located.

[0074] 3. Comparative analysis to determine whether the obtained business names are in the supplier's inventory.

[0075] 4. Starting from the warning time, obtain relevant information about the companies that signed procurement contracts within the past four years from the time of the warning in the legal system.

[0076] 5. Comparative analysis: If the obtained business name exists in the supplier's inventory but there is no purchase contract in the legal system, an early warning will be issued.

[0077] 6. Comparative analysis: Check if the obtained business name exists in the company that signed the contract. If it does, issue a warning.

[0078] On the other hand, the present invention also provides a power system-wide intelligent risk early warning and handling system, which may include a data acquisition module and a controller.

[0079] The data acquisition module connects to the power system to collect characteristic data from the power system. The controller connects to the data acquisition module to execute any of the above-mentioned intelligent risk warning and handling methods.

[0080] Through the above technical solution, the intelligent early warning and handling method and system based on the overall risk of the power system provided by this invention acquires multiple feature data of the power system, preprocesses each feature data to obtain a corresponding training set, and then constructs an early warning model for each feature data. This model is then trained using the corresponding training set to identify and diagnose multiple real-time feature data of the current power system, thereby determining the risk level of the current power system and taking corresponding handling measures. By employing early warning models based on multiple feature data to identify and diagnose multiple risks in the power system, the adaptability and accuracy of power system early warning are effectively improved, thereby enhancing the reliability and efficiency of risk handling.

[0081] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0082] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams.

[0083] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0084] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0085] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0086] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0087] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0088] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0089] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for intelligent early warning and handling of risks across the entire power system, characterized in that, include: The system acquires various characteristic data of the power system, including material procurement characteristic data, marketing service characteristic data, and engineering project characteristic data; preprocesses each type of characteristic data to construct a training set for each type of characteristic data; constructs an early warning model for each type of characteristic data; trains the corresponding early warning model using the training set for each type of characteristic data; acquires various real-time characteristic data of the current power system; and determines and handles the risk level of the current power system based on the various real-time characteristic data of the current power system and the corresponding early warning model.

2. The risk intelligent early warning and handling method according to claim 1, characterized in that, Preprocessing the various feature data includes: handling missing values ​​in the various feature data; handling outliers in the various feature data; and performing feature standardization on the various feature data.

3. The risk intelligent early warning and handling method according to claim 1, characterized in that, Constructing a training set for each of the aforementioned feature data includes: obtaining the repetition rate of bid document IPs in the material procurement feature data according to formula (1). (1) Among them, This refers to the duplication rate of bid document IPs in the material procurement feature data. The number of unique bid documents with IP addresses within the same project. The number of IPs for all tender documents in the same project; the document creation time difference of the tender documents in the material procurement feature data is obtained according to formula (2). (2) Among them, Create a time difference for the tender documents. 、 These are the earliest and latest document creation times of the tender documents, respectively; the supplier correlation in the material procurement characteristic data is obtained according to formula (3). (3) Among them, For suppliers and suppliers The degree of correlation, 、 、 These are equity weight, personnel weight, and historical behavior weight, respectively. 、 、 The scores are respectively based on equity association, personnel association, and historical behavior association; the duplication rate of the tender document IP, the document creation time difference, and the supplier association are labeled, and a material procurement training set is constructed.

4. The risk intelligent early warning and handling method according to claim 3, characterized in that, The early warning model for each type of feature data is constructed as follows: a material procurement early warning model is constructed using the random forest algorithm.

5. The risk intelligent early warning and handling method according to claim 1, characterized in that, Constructing a training set for each of the aforementioned feature data includes: the process time for obtaining the marketing service feature data according to formula (4). (4) Among them, The process duration for the aforementioned marketing service feature data. The end time of the marketing service process. The start time of the marketing service; the completeness of the approval records in the marketing service feature data is obtained according to formula (5). (5) Among them, For the first The completeness of each approval record For the first The number of completed approval records. For the first The standard number of completed approval records The process duration and the completeness of the approval records are labeled, and a marketing service training set is constructed.

6. The risk intelligent early warning and handling method according to claim 5, characterized in that, The construction of early warning models for each type of feature data includes: constructing a marketing service early warning model using a logistic regression algorithm.

7. The risk intelligent early warning and handling method according to claim 1, characterized in that, Constructing a training set for each of the aforementioned feature data includes: obtaining the time difference between acceptance and capital transfer in the project feature data according to formula (6). (6) Among them, This refers to the time difference between acceptance and asset transfer in the characteristic data of the aforementioned engineering project. For the transfer time of engineering projects, The acceptance time of the project is determined; the project amount in the characteristic data of the project is obtained; the time difference between acceptance and capital transfer and the project amount are labeled, and a training set of the project is constructed.

8. The risk intelligent early warning and handling method according to claim 5, characterized in that, The construction of early warning models for each type of feature data includes: constructing an early warning model for engineering projects using the gradient boosting tree algorithm.

9. The risk intelligent early warning and handling method according to claim 1, characterized in that, Based on the various real-time characteristic data of the current power system and the corresponding early warning model, the risk level of the current power system is obtained and the following measures are taken: inputting the various real-time characteristic data of the current power system into the corresponding early warning model to output risk characteristics and risk probabilities; obtaining the degree of impact and rectification difficulty corresponding to the risk characteristics; obtaining the risk value according to formula (7), (7) Among them, This is the risk value. 、 、 These are the impact weight, rectification weight, and probability weight, respectively. The degree of impact corresponding to the aforementioned risk characteristics, The difficulty of rectification corresponding to the aforementioned risk characteristics. The process involves: scoring the probability of risk; obtaining the corresponding risk level based on the risk value; determining whether the risk level is high-risk; taking immediate measures if the risk level is high-risk; determining whether the risk level is medium-risk if the risk level is not high-risk; setting a deadline for rectification if the risk level is medium-risk; and issuing an early warning if the risk level is not medium-risk.

10. A power system-wide risk intelligent early warning and response system, characterized in that, include: A data acquisition module, connected to the power system, is used to collect characteristic data from the power system; The controller, connected to the data acquisition module, is used to execute the risk intelligent early warning and handling method as described in any one of claims 1-9.

Images