Method and system for predicting and continuously monitoring electric power marketing risk

Abstract

The invention provides a method for predicting and continuously monitoring electric power marketing risks. The method comprises the following steps: acquiring a business process with the maximum risk event occurrence probability; in the prior risk early warning stage, risk early warning evaluation is carried out on the acquired business process; in the in-event risk control stage, modeling is carried out on the obtained business process to obtain a directed graph model used for representing a circulation relationship among a plurality of process nodes, and further according to a risk checking algorithm, the process nodes with risk scores greater than a preset threshold value are determined and strategy optimization is carried out; and in the post risk processing stage, continuously monitoring the risk scores of all the process nodes after strategy optimization, and automatically triggering the emergency strategy for correction after monitoring that at least one risk score of a certain process node is still greater than a preset threshold value. By implementing the method and the system, various risk events of the power marketing business network can be covered in an all-around manner in three stages, i.e., before, during and after, so that the risk response capability and the business robustness of an enterprise are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a method and system for predicting and continuously monitoring power marketing risks. Background Art

[0002] With the rapid development of the power market, the risks faced by the power marketing business network are becoming increasingly complex and diverse. These risks not only affect the operational efficiency and customer satisfaction of power companies, but may also pose a serious threat to the compliance and market position of companies.

[0003] At present, traditional risk management methods are often limited to a certain stage or a single dimension, and deal with problems after the fact through inspections and audits, which makes it difficult to achieve comprehensive foresight and continuous monitoring of risks.

[0004] Therefore, under the development trend of automation and intelligence of power systems, there is an urgent need for an innovative risk management method that can comprehensively cover various risk events in the power marketing business network in the three stages of before, during and after the event, so as to improve the company's risk response capabilities and business robustness. Summary of the invention

[0005] The technical problem to be solved by the embodiments of the present invention is to provide a method and system for predicting and continuously monitoring power marketing risks, which can comprehensively cover various risk events of the power marketing business network in the three stages of before, during and after the event, so as to improve the enterprise's risk response capabilities and business robustness.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides a method for predicting and continuously monitoring power marketing risks, the method comprising the following steps:

[0007] Obtain the business process with the highest probability of risk events occurring in the power marketing business network;

[0008] In the pre-risk warning stage, based on the trained global evaluation model, the acquired business processes are evaluated for risk warning;

[0009] In the in-process risk control stage, the acquired business process is modeled to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further, based on the preset risk screening algorithm, the process nodes with risk scores greater than the preset threshold are determined and strategy optimization is performed;

[0010] In the post-risk handling stage, based on a predetermined period of time, the risk scores of all process nodes after strategy optimization are continuously monitored, and when it is monitored that at least one risk score of a process node after strategy optimization is still greater than the preset threshold, the preset emergency strategy is automatically triggered for correction.

[0011] Among them, the business process includes business expansion registration, electricity reading and collection, metering, electricity usage inspection, customer service and electricity usage changes.

[0012] Among them, in the prior risk warning stage, the steps of performing risk warning assessment on the acquired business process based on the trained global evaluation model specifically include:

[0013] Determine the key risk indicators of the acquired business process, and construct all the determined key risk indicators into a one-dimensional vector, and further combine the weights of each key risk indicator to construct a global evaluation model based on a linear regression model;

[0014] The global evaluation model is trained using training samples to obtain a trained global evaluation model; wherein the training samples are composed of historical data of multiple one-dimensional vectors; and the historical data of each one-dimensional vector is composed of historical data of each key risk indicator contained therein;

[0015] Determine the current data of the key risk indicators of the acquired business process to form the current data as a one-dimensional vector, and import the current data of the one-dimensional vector into the trained global evaluation model for risk assessment to obtain a risk assessment value, and automatically trigger an early warning when it is determined that the obtained risk assessment value is greater than a preset early warning threshold.

[0016] Among them, in the in-process risk control stage, the obtained business process is modeled to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further according to a preset risk screening algorithm, the process nodes with risk scores greater than a preset threshold are determined and the steps of strategy optimization specifically include:

[0017] The obtained business process is modeled using Petri nets to obtain a directed graph model for representing the flow relationship between multiple process nodes; the flow relationship between adjacent process nodes is represented by directed arcs;

[0018] According to the directed graph model, a risk screening algorithm based on the risk transfer probability matrix is ​​constructed, and the risk screening algorithm is used to step by step screen each process node, calculate the risk transfer probability of each process node, and further screen out the process nodes whose risk transfer probability is greater than a preset upper limit value;

[0019] The risk probability and risk impact of each selected process node are estimated, and the risk score of each selected process node is obtained based on the estimated risk probability and risk impact of each selected process node;

[0020] If the risk score of a selected process node is greater than the preset threshold, the strategy is automatically optimized; wherein the strategy optimization is related measures for reducing the risk score, including a first measure and a second measure; the first measure is specifically to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact; the second measure is specifically to increase personnel and shorten processing time to reduce the estimated risk probability and risk impact.

[0021] Among them, in the post-event risk handling stage, the steps of continuously monitoring the risk scores of all process nodes after strategy optimization based on a predetermined time period, and automatically triggering the preset emergency strategy for correction after monitoring that at least one risk score of a process node after strategy optimization is still greater than the preset threshold value specifically include:

[0022] A Kalman filter is used to continuously monitor the risk scores of all process nodes after strategy optimization for a predetermined period of time, so as to obtain the real-time risk scores of all process nodes after strategy optimization for the predetermined period of time;

[0023] If at least one risk score of a process node currently monitored after strategy optimization is still greater than the preset threshold, the preset emergency strategy is automatically triggered for correction; wherein, the emergency strategy includes a first emergency strategy and a second emergency strategy; the first emergency strategy specifically deploys on-site personnel to investigate and correct the currently monitored process node to ensure that the risk score is less than the preset threshold; the second emergency strategy specifically divides the responsibilities of all monitored process nodes point by point, and readjusts the directed graph model to determine the responsibilities and processing time of the personnel at each process node to ensure that the risk score is less than the preset threshold.

[0024] The embodiment of the present invention also provides a system for predicting and continuously monitoring power marketing risks, including:

[0025] A risk business process acquisition unit, used to acquire the business process with the highest probability of occurrence of risk events in the power marketing business network;

[0026] The business process risk prediction unit is used to conduct risk warning assessment on the acquired business process based on the trained global evaluation model in the pre-risk warning stage;

[0027] The business process risk control unit is used to model the acquired business process in the risk control stage to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further determine the process nodes with risk scores greater than a preset threshold and perform strategy optimization based on a preset risk screening algorithm;

[0028] The business process risk continuous monitoring unit is used to continuously monitor the risk scores of all process nodes after strategy optimization based on a predetermined time period in the post-risk processing stage, and automatically trigger the preset emergency strategy for correction after monitoring that at least one risk score of a process node after strategy optimization is still greater than the preset threshold.

[0029] Among them, the business process includes business expansion registration, electricity reading and collection, metering, electricity usage inspection, customer service and electricity usage changes.

[0030] Wherein, the business process risk prediction unit includes:

[0031] A model building module is used to determine the key risk indicators of the acquired business process, and construct all the determined key risk indicators into a one-dimensional vector, and further combine the weights of each key risk indicator to construct a global evaluation model based on a linear regression model;

[0032] A model training module is used to train the global evaluation model using training samples to obtain a trained global evaluation model; wherein the training samples are composed of historical data of multiple one-dimensional vectors; and the historical data of each one-dimensional vector is composed of historical data of each key risk indicator contained therein;

[0033] The risk warning module is used to determine the current data of the key risk indicators of the acquired business process to form the current data as a one-dimensional vector, and import the current data of the one-dimensional vector into the trained global evaluation model for risk assessment to obtain a risk assessment value, and automatically trigger a warning when it is determined that the obtained risk assessment value is greater than a preset warning threshold.

[0034] Wherein, the business process risk control unit includes:

[0035] The process node modeling module is used to model the acquired business process using Petri nets to obtain a directed graph model for representing the flow relationship between multiple process nodes; wherein the flow relationship between adjacent process nodes is represented by directed arcs;

[0036] A risk screening module is used to construct a risk screening algorithm based on a risk transfer probability matrix according to the directed graph model, and use the risk screening algorithm to gradually screen each process node, calculate the risk transfer probability of each process node, and further screen out the process nodes whose risk transfer probability is greater than a preset upper limit value;

[0037] The risk scoring module is used to estimate the risk probability and risk impact of each selected process node, and obtain the risk score of each selected process node based on the estimated risk probability and risk impact of each selected process node;

[0038] A strategy optimization module is used for automatic strategy optimization if the risk score of a selected process node is greater than the preset threshold; wherein the strategy optimization is related measures for reducing the risk score, including a first measure and a second measure; the first measure is specifically to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact; the second measure is specifically to increase personnel and shorten processing time to reduce the estimated risk probability and risk impact.

[0039] Wherein, the business process risk continuous monitoring unit includes:

[0040] A process node dynamic monitoring module is used to continuously monitor the risk scores of all strategy-optimized process nodes over a predetermined period of time using a Kalman filter to obtain real-time risk scores of all strategy-optimized process nodes over the predetermined period of time;

[0041] An emergency correction module is used to automatically trigger the preset emergency strategy for correction if at least one risk score of a process node currently monitored after strategy optimization is still greater than the preset threshold; wherein the emergency strategy includes a first emergency strategy and a second emergency strategy; the first emergency strategy specifically deploys on-site personnel to investigate and correct the currently monitored process node to ensure that the risk score is less than the preset threshold; the second emergency strategy specifically divides the responsibilities of all monitored process nodes point by point, and readjusts the directed graph model to determine the responsibilities and processing time of the personnel at each process node to ensure that the risk score is less than the preset threshold.

[0042] Implementing the embodiments of the present invention has the following beneficial effects:

[0043] The present invention constructs a comprehensive and progressive risk management strategy of "network, line, and point", which can comprehensively cover various risk events of the power marketing business network in the three stages of before, during, and after the event. It not only improves the risk response capability and business robustness of the enterprise, but also improves the overall standardization of the power marketing business network, while reducing potential risks and customer complaints. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, without paying creative labor, other drawings obtained based on these drawings still belong to the scope of the present invention.

[0045] Figure 1A flow chart of a method for predicting and continuously monitoring power marketing risks provided by an embodiment of the present invention;

[0046] Figure 2 A schematic diagram of the structure of a system for predicting and continuously monitoring power marketing risks provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings.

[0048] like Figure 1 As shown, a method for predicting and continuously monitoring power marketing risks is provided in an embodiment of the present invention, and the method includes the following steps:

[0049] Step S1, obtaining a business process with the highest probability of occurrence of a risk event in the power marketing business network;

[0050] The specific process is to find the business process with the highest probability of risk events in the power marketing business network. The business process includes but is not limited to business expansion registration, reading and verification, metering, power inspection, customer service and power change, which can be flexibly adjusted according to actual conditions.

[0051] It should be noted that risk events include but are not limited to process delays, data errors, and non-compliant operations, which can easily lead to increased customer complaints and potential legal risks. At this time, the probability of risk events is automatically calculated through historical data.

[0052] Step S2: In the pre-risk warning stage, based on the trained global evaluation model, a risk warning assessment is performed on the acquired business process;

[0053] The specific process is as follows: first, determine the key risk indicators of the acquired business process, and construct all the determined key risk indicators into a one-dimensional vector, and further combine the weights of each key risk indicator to construct a global evaluation model based on the linear regression model.

[0054] For example, the key risk indicators of the business process obtained are determined from the three dimensions of risk control, compliance and customer experience. In one example, the key risk indicators corresponding to the business expansion application include the abnormal time rate of high-voltage business expansion installation and disassembly, the abnormal suspension rate of low-voltage business expansion, the proportion of high-voltage temporary power consumption that has not been cleared after the expiration date, the proportion of contracts signed after the high-voltage business expansion is delivered, the customer power consumption expectation achievement rate, and the customer satisfaction rate.

[0055] Furthermore, these key risk indicators are defined as a one-dimensional vector, specifically I = [I1, I2, …, Ii, …, In] T; Where n is the total number of key risk indicators; Ii is the i-th key risk indicator.

[0056] At the same time, the weights wi of these key risk indicators are determined through the Delphi expert survey method, and the following conditions are met:

[0057]

[0058] Furthermore, a global evaluation model is constructed based on the linear regression model, specifically:

[0059]

[0060] Among them, R is the global risk assessment value; α0 is the intercept; αi is the regression coefficient, which is replaced by the weight wi; ∈ is the error term.

[0061] Furthermore, the minimization loss function of the global evaluation model is set as follows:

[0062]

[0063] Among them, m is the number of samples.

[0064] Secondly, the global evaluation model is trained using training samples to obtain a trained global evaluation model; wherein the training samples are composed of historical data of multiple one-dimensional vectors; the historical data of each one-dimensional vector is composed of historical data of each key risk indicator contained therein. That is, the historical data of each key risk indicator is obtained, and the historical data of each key risk indicator at the same time are combined into a one-dimensional vector as a training sample, so that a training sample composed of the historical data of several one-dimensional vectors can be obtained.

[0065] Finally, the current data of the key risk indicators of the acquired business process is determined to form the current data as a one-dimensional vector, and the current data of the one-dimensional vector is imported into the trained global evaluation model for risk assessment to obtain a risk assessment value, and when it is determined that the obtained risk assessment value is greater than the preset warning threshold, an early warning is automatically triggered.

[0066] For example, the current real-time data is input into the trained global evaluation model for risk assessment to obtain the real-time risk assessment value Rreal. If the real-time risk assessment value Rreal> the preset warning threshold Talert, the warning is automatically triggered.

[0067] Step S3: In the in-process risk control stage, the acquired business process is modeled to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further, according to a preset risk screening algorithm, the process nodes with risk scores greater than a preset threshold are determined and strategy optimization is performed;

[0068] The specific process is as follows: first, the acquired business process is modeled using Petri nets to obtain a directed graph model for representing the flow relationship between multiple process nodes; wherein the flow relationship between adjacent process nodes is represented by directed arcs.

[0069] For example, a business process in the power marketing business network is modeled using Petri nets to generate a directed graph model Gprocess, which is in the form of: Gprocess = (N, E); where N represents the set of process nodes, and E represents the set of arcs; each process node N j Corresponding to a certain link in the business process, through arc E jk Indicates the flow relationship of business.

[0070] Secondly, according to the directed graph model, a risk screening algorithm based on the risk transfer probability matrix is ​​constructed. The risk screening algorithm is used to step by step screen each process node, calculate the risk transfer probability of each process node, and further screen out the process nodes whose risk transfer probability is greater than the preset upper limit value.

[0071] For example, based on the directed graph model Gprocess, we build a risk screening algorithm Acheck and use the risk transfer probability matrix P jk Each process node is checked step by step, and the risk transfer probability of each process node is calculated.

[0072]

[0073] Among them, P j is the process node N j The error probability P t It is the risk transfer probability of risk events being transmitted to the terminal node.

[0074] Further, the risk transfer probability P t >Process node with preset upper limit value P0.

[0075] Next, the risk probability and risk impact of each selected process node are estimated, and the risk score of each selected process node is obtained based on the estimated risk probability and risk impact of each selected process node.

[0076] For example, assign a risk probability interval from 1 to 5, and score according to the probability of occurrence of existing events and incremental events to obtain the risk probability L of each process node. risk ; The risk impact is graded from 1 to 5, corresponding to the hazard levels of negligible, minor, moderate, severe, and critical. The highest score among the evaluation criteria of "error amount", "impact level", "social image", and "audit expert correction" is taken to obtain the risk impact S of each process node.risk ;

[0077] By formula R risk =L risk *S risk , calculate the risk score R of each process node risk .

[0078] Finally, if the risk score of a selected process node is greater than a preset threshold, the strategy is automatically optimized; wherein the strategy optimization is related measures used to reduce the risk score, including but not limited to the first measure and the second measure; the first measure is specifically to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact; the second measure is specifically to increase the number of personnel and shorten the processing time to reduce the estimated risk probability and risk impact.

[0079] For example, set high risk to R risk >=15, medium risk is 5= <R risk <15 and low risk is 1 <R risk <5, so the risk score R risk >When the preset threshold is 15, the automatic strategy optimization is required to increase the number of personnel and shorten the processing time to reduce the estimated risk probability and risk impact;

[0080] Alternatively, the automatic strategy optimization is required to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact, that is, execute the correction algorithm Acorrect to adjust the directed graph model to obtain the adjusted directed graph model The overall risk can be reduced. At this time, a new risk transfer probability is generated So that:

[0081] Step S4: In the post-risk handling stage, based on a predetermined time period, the risk scores of all process nodes after strategy optimization are continuously monitored, and when it is monitored that at least one risk score of a process node after strategy optimization is still greater than the preset threshold, the preset emergency strategy is automatically triggered for correction.

[0082] The specific process is as follows: first, a Kalman filter is used to continuously monitor the risk scores of all process nodes after strategy optimization within a predetermined time period to obtain real-time risk scores of all process nodes after strategy optimization within the predetermined time period.

[0083] Finally, if at least one risk score of a process node currently monitored after strategy optimization is still greater than the preset threshold, the preset emergency strategy will be automatically triggered for correction; among them, the emergency strategy includes but is not limited to the first emergency strategy and the second emergency strategy; the first emergency strategy is to deploy on-site personnel to investigate and correct the currently monitored process nodes to ensure that the risk score is less than the preset threshold; the second emergency strategy is to divide the responsibilities of all monitored process nodes point by point, and readjust the directed graph model to determine the responsibilities and processing time of the personnel at each process node to ensure that the risk score is less than the preset threshold.

[0084] For example, assuming that the real-time data stream shows that the risk score of process node A has dropped slightly, but is still above the preset threshold (i.e., the warning value), the risk score needs to be updated through the Kalman filter and monitoring continues.

[0085] (1) Point-by-point early warning response: Once the real-time risk score of process node A continues to be higher than the preset threshold, the emergency strategy is automatically triggered and on-site staff are deployed to conduct further investigation and correction to ensure that the risk score of process node A decreases and is less than the preset threshold.

[0086] (2) Point-by-point optimization strategy: once the real-time risk score of process node A continues to be higher than the preset threshold, the emergency strategy is automatically triggered, and the genetic algorithm is used to optimize process node A, including increasing on-site personnel and shortening time to reduce the risk score and / or determining the responsibilities of the person in charge and his or her department through an overall planning model and clearly recording them through a responsibility allocation matrix to ensure that the risk score is less than the preset threshold.

[0087] It should be noted that the Bayesian network can be used to evaluate the effectiveness of the response measures for each high-risk process node, update the risk professional matrix and feed back to the global evaluation model to ensure that changes in each "point" can affect the overall risk control strategy.

[0088] In an embodiment of the present invention, an application scenario of a method for predicting and continuously monitoring power marketing risks is further described as follows:

[0089] Assume that a power company has discovered in its marketing business network that there is a high probability of risk events occurring in the business process of "business expansion and installation". Such risk events include but are not limited to process delays, data errors, and non-compliant operations, which can easily lead to increased customer complaints and potential legal risks.

[0090] The first step is to define six key risk indicators related to "business expansion registration" during the risk control phase, including: process delay rate (I1), data entry error rate (I2), document compliance check (I3), customer satisfaction (I4), complaint handling time (I5), and internal audit pass rate (I6). These key risk indicators are uniformly defined as a one-dimensional vector form: I = [I1, I2, ..., I6] T , and determine its weight wi through the Delphi expert survey method to ensure that the total is

[0091] Secondly, based on the above indicators, a global evaluation model of the linear regression model is constructed, specifically:

[0092] R=α0+α1*I1+α2*I2+...+α6*I6

[0093] Then, the one-dimensional vector data monitored in real time is input into the global evaluation model. If the obtained risk assessment value Rreal> the preset warning threshold Talert, the warning is automatically triggered.

[0094] The second step, during the risk control stage, firstly, the business process of "business expansion registration" was modeled using Petri nets, generating a directed graph model Gprocess, including multiple process nodes, such as preliminary application (N1), document review (N2) and on-site investigation (N3).

[0095] Secondly, using the risk transfer probability matrix P jk , the risk screening algorithm Acheck is used to step by step screen each process node in the business process of "business expansion and registration" and calculate the risk transfer probability P t .

[0096] Assume that in process node N3 (i.e., on-site inspection), it is found that the probability of risk event transfer to process node N4 (i.e., audit passing) is high, that is, the risk transfer probability P of process node N3 t >Preset upper limit value P0. At this time, record that process node N3 has risks, and continue to check other process nodes.

[0097] Next, for the high-risk process node N3, the corresponding risk probability L is estimated. risk (assuming the result is 4 points), and the risk impact level S risk (Assume the result is 5 points);

[0098] By formula R risk =L risk *S risk , and obtain the risk score R of process node N3 risk=20. Given that the high risk is R risk >=15, it is necessary to start strategy optimization for process node N3. For example, for process node N3, execute the correction algorithm Acorrect to readjust the directed graph model and update the transition probability matrix to reduce the risk of process node N3. At this time, the new directed graph model Reduce the risk transmission rate of this node to an acceptable range.

[0099] Step 3: In the post-risk processing stage, Kalman filter dynamic monitoring is implemented for high-risk process node N3. Assume that the real-time data stream shows that the risk score of process node N3 has slightly decreased, but is still above the preset threshold. Therefore, the risk score of process node N3 is updated through the Kalman filter and continues to be monitored. Point-by-point early warning response, once the real-time risk score of process node N3 continues to be higher than the preset threshold, the emergency strategy will be automatically triggered, and on-site staff will be deployed for further investigation and correction.

[0100] Alternatively, optimize the strategy point by point, using genetic algorithms to optimize the process node N3 strategy, including increasing the number of on-site survey personnel and shortening the survey time to reduce the risk score. And / or divide the responsibilities point by point, through the integer programming model, determine the responsibilities of the survey team leader and his department, and clearly record them through the responsibility allocation matrix to ensure the implementation of the measures.

[0101] like Figure 2 As shown, a system for predicting and continuously monitoring power marketing risks is provided in an embodiment of the present invention, including:

[0102] The risk business process acquisition unit 110 is used to acquire the business process with the highest probability of occurrence of risk events in the power marketing business network;

[0103] The business process risk prediction unit 120 is used to perform risk warning evaluation on the acquired business process based on the trained global evaluation model in the pre-risk warning stage;

[0104] The business process risk control unit 130 is used to model the acquired business process in the in-process risk control stage to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further determine the process nodes with risk scores greater than a preset threshold and perform strategy optimization according to a preset risk screening algorithm;

[0105] The business process risk continuous monitoring unit 140 is used to continuously monitor the risk scores of all process nodes after strategy optimization based on a predetermined time period in the post-risk processing stage, and automatically trigger the preset emergency strategy for correction after monitoring that at least one risk score of a process node after strategy optimization is still greater than the preset threshold.

[0106] Among them, the business process includes business expansion registration, electricity reading and collection, metering, electricity usage inspection, customer service and electricity usage changes.

[0107] The business process risk prediction unit 120 includes:

[0108] A model building module is used to determine the key risk indicators of the acquired business process, and construct all the determined key risk indicators into a one-dimensional vector, and further combine the weights of each key risk indicator to construct a global evaluation model based on a linear regression model;

[0109] A model training module is used to train the global evaluation model using training samples to obtain a trained global evaluation model; wherein the training samples are composed of historical data of multiple one-dimensional vectors; and the historical data of each one-dimensional vector is composed of historical data of each key risk indicator contained therein;

[0110] The risk warning module is used to determine the current data of the key risk indicators of the acquired business process to form the current data as a one-dimensional vector, and import the current data of the one-dimensional vector into the trained global evaluation model for risk assessment to obtain a risk assessment value, and automatically trigger a warning when it is determined that the obtained risk assessment value is greater than a preset warning threshold.

[0111] The business process risk control unit 130 includes:

[0112] The process node modeling module is used to model the acquired business process using Petri nets to obtain a directed graph model for representing the flow relationship between multiple process nodes; wherein the flow relationship between adjacent process nodes is represented by directed arcs;

[0113] A risk screening module is used to construct a risk screening algorithm based on a risk transfer probability matrix according to the directed graph model, and use the risk screening algorithm to gradually screen each process node, calculate the risk transfer probability of each process node, and further screen out the process nodes whose risk transfer probability is greater than a preset upper limit value;

[0114] The risk scoring module is used to estimate the risk probability and risk impact of each selected process node, and obtain the risk score of each selected process node based on the estimated risk probability and risk impact of each selected process node;

[0115] A strategy optimization module is used for automatic strategy optimization if the risk score of a selected process node is greater than the preset threshold; wherein the strategy optimization is related measures for reducing the risk score, including a first measure and a second measure; the first measure is specifically to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact; the second measure is specifically to increase personnel and shorten processing time to reduce the estimated risk probability and risk impact.

[0116] The business process risk continuous monitoring unit 140 includes:

[0117] A process node dynamic monitoring module is used to continuously monitor the risk scores of all strategy-optimized process nodes over a predetermined period of time using a Kalman filter to obtain real-time risk scores of all strategy-optimized process nodes over the predetermined period of time;

[0118] An emergency correction module is used to automatically trigger the preset emergency strategy for correction if at least one risk score of a process node currently monitored after strategy optimization is still greater than the preset threshold; wherein the emergency strategy includes a first emergency strategy and a second emergency strategy; the first emergency strategy specifically deploys on-site personnel to investigate and correct the currently monitored process node to ensure that the risk score is less than the preset threshold; the second emergency strategy specifically divides the responsibilities of all monitored process nodes point by point, and readjusts the directed graph model to determine the responsibilities and processing time of the personnel at each process node to ensure that the risk score is less than the preset threshold.

[0119] Implementing the embodiments of the present invention has the following beneficial effects:

[0120] The present invention constructs a comprehensive and progressive risk management strategy of "network, line, and point", which can comprehensively cover various risk events of the power marketing business network in the three stages of before, during, and after the event. It not only improves the risk response capability and business robustness of the enterprise, but also improves the overall standardization of the power marketing business network, while reducing potential risks and customer complaints.

[0121] It is worth noting that in the above system embodiment, the various system modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the various functional modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0122] A person skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as ROM / RAM, disk, CD-ROM, etc.

[0123] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for predicting and continuously monitoring power marketing risks, characterized in that: The method comprises the following steps: Obtain the business process with the highest probability of risk events occurring in the power marketing business network; In the pre-risk warning stage, based on the trained global evaluation model, the acquired business processes are evaluated for risk warning; In the in-process risk control stage, the acquired business process is modeled to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further, based on the preset risk screening algorithm, the process nodes with risk scores greater than the preset threshold are determined and strategy optimization is performed; In the post-risk handling stage, based on a predetermined period of time, the risk scores of all process nodes after strategy optimization are continuously monitored, and when it is monitored that at least one risk score of a process node after strategy optimization is still greater than the preset threshold, the preset emergency strategy is automatically triggered for correction.

2. The method for predicting and continuously monitoring power marketing risks according to claim 1, characterized in that: The business process includes business expansion registration, electricity metering, metering, electricity usage inspection, customer service and electricity usage changes.

3. The method for predicting and continuously monitoring power marketing risks according to claim 1, characterized in that: In the prior risk warning stage, the steps of performing risk warning assessment on the acquired business process based on the trained global evaluation model specifically include: Determine the key risk indicators of the acquired business process, and construct all the determined key risk indicators into a one-dimensional vector, and further combine the weights of each key risk indicator to construct a global evaluation model based on a linear regression model; The global evaluation model is trained using training samples to obtain a trained global evaluation model; wherein the training samples are composed of historical data of multiple one-dimensional vectors; and the historical data of each one-dimensional vector is composed of historical data of each key risk indicator contained therein; Determine the current data of the key risk indicators of the acquired business process to form the current data as a one-dimensional vector, and import the current data of the one-dimensional vector into the trained global evaluation model for risk assessment to obtain a risk assessment value, and automatically trigger an early warning when it is determined that the obtained risk assessment value is greater than a preset early warning threshold.

4. The method for predicting and continuously monitoring power marketing risks according to claim 3, characterized in that: In the in-process risk control stage, the obtained business process is modeled to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further according to a preset risk screening algorithm, the process nodes with risk scores greater than a preset threshold are determined and the strategy optimization steps are specifically included: The obtained business process is modeled using Petri nets to obtain a directed graph model for representing the flow relationship between multiple process nodes; the flow relationship between adjacent process nodes is represented by directed arcs; According to the directed graph model, a risk screening algorithm based on the risk transfer probability matrix is ​​constructed, and the risk screening algorithm is used to step by step screen each process node, calculate the risk transfer probability of each process node, and further screen out the process nodes whose risk transfer probability is greater than a preset upper limit value; The risk probability and risk impact of each selected process node are estimated, and the risk score of each selected process node is obtained based on the estimated risk probability and risk impact of each selected process node; If the risk score of a selected process node is greater than the preset threshold, the strategy is automatically optimized; wherein the strategy optimization is related measures for reducing the risk score, including a first measure and a second measure; the first measure is specifically to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact; the second measure is specifically to increase personnel and shorten processing time to reduce the estimated risk probability and risk impact.

5. The method for predicting and continuously monitoring power marketing risks according to claim 4, characterized in that: In the post-event risk handling stage, the steps of continuously monitoring the risk scores of all process nodes after strategy optimization based on a predetermined time period, and automatically triggering the preset emergency strategy for correction after monitoring that at least one risk score of a process node after strategy optimization is still greater than the preset threshold value specifically include: A Kalman filter is used to continuously monitor the risk scores of all process nodes after strategy optimization for a predetermined period of time, so as to obtain the real-time risk scores of all process nodes after strategy optimization for the predetermined period of time; If at least one risk score of a process node currently monitored after strategy optimization is still greater than the preset threshold, the preset emergency strategy is automatically triggered for correction; wherein, the emergency strategy includes a first emergency strategy and a second emergency strategy; the first emergency strategy specifically deploys on-site personnel to investigate and correct the currently monitored process node to ensure that the risk score is less than the preset threshold; the second emergency strategy specifically divides the responsibilities of all monitored process nodes point by point, and readjusts the directed graph model to determine the responsibilities and processing time of the personnel at each process node to ensure that the risk score is less than the preset threshold.

6. A system for predicting and continuously monitoring power marketing risks, characterized in that: include: A risk business process acquisition unit, used to acquire the business process with the highest probability of occurrence of risk events in the power marketing business network; The business process risk prediction unit is used to conduct risk warning assessment on the acquired business process based on the trained global evaluation model in the pre-risk warning stage; The business process risk control unit is used to model the acquired business process in the risk control stage to obtain a directed graph model for characterizing the flow relationship between multiple process nodes, and further determine the process nodes with risk scores greater than a preset threshold and perform strategy optimization based on a preset risk screening algorithm; The business process risk continuous monitoring unit is used to continuously monitor the risk scores of all process nodes after strategy optimization based on a predetermined time period in the post-risk processing stage, and automatically trigger the preset emergency strategy for correction after monitoring that at least one risk score of a process node after strategy optimization is still greater than the preset threshold.

7. The system for predicting and continuously monitoring power marketing risks according to claim 6, characterized in that: The business process includes business expansion registration, electricity metering, metering, electricity usage inspection, customer service and electricity usage changes.

8. The system for predicting and continuously monitoring power marketing risks according to claim 6, characterized in that: The business process risk prediction unit includes: A model building module is used to determine the key risk indicators of the acquired business process, and construct all the determined key risk indicators into a one-dimensional vector, and further combine the weights of each key risk indicator to construct a global evaluation model based on a linear regression model; A model training module is used to train the global evaluation model using training samples to obtain a trained global evaluation model; wherein the training samples are composed of historical data of multiple one-dimensional vectors; and the historical data of each one-dimensional vector is composed of historical data of each key risk indicator contained therein; The risk warning module is used to determine the current data of the key risk indicators of the acquired business process to form the current data as a one-dimensional vector, and import the current data of the one-dimensional vector into the trained global evaluation model for risk assessment to obtain a risk assessment value, and automatically trigger a warning when it is determined that the obtained risk assessment value is greater than a preset warning threshold.

9. The system for predicting and continuously monitoring power marketing risks according to claim 8, characterized in that: The business process risk control unit includes: The process node modeling module is used to model the acquired business process using Petri nets to obtain a directed graph model for representing the flow relationship between multiple process nodes; wherein the flow relationship between adjacent process nodes is represented by directed arcs; A risk screening module is used to construct a risk screening algorithm based on a risk transfer probability matrix according to the directed graph model, and use the risk screening algorithm to gradually screen each process node, calculate the risk transfer probability of each process node, and further screen out the process nodes whose risk transfer probability is greater than a preset upper limit value; The risk scoring module is used to estimate the risk probability and risk impact of each selected process node, and obtain the risk score of each selected process node based on the estimated risk probability and risk impact of each selected process node; A strategy optimization module is used for automatic strategy optimization if the risk score of a selected process node is greater than the preset threshold; wherein the strategy optimization is related measures for reducing the risk score, including a first measure and a second measure; the first measure is specifically to adjust the directed graph model and update the risk transfer probability matrix to reduce the estimated risk probability and risk impact; the second measure is specifically to increase personnel and shorten processing time to reduce the estimated risk probability and risk impact.

10. The system for predicting and continuously monitoring power marketing risks according to claim 9, characterized in that: The business process risk continuous monitoring unit includes: A process node dynamic monitoring module is used to continuously monitor the risk scores of all strategy-optimized process nodes over a predetermined period of time using a Kalman filter to obtain real-time risk scores of all strategy-optimized process nodes over the predetermined period of time; An emergency correction module is used to automatically trigger the preset emergency strategy for correction if at least one risk score of a process node currently monitored after strategy optimization is still greater than the preset threshold; wherein the emergency strategy includes a first emergency strategy and a second emergency strategy; the first emergency strategy specifically deploys on-site personnel to investigate and correct the currently monitored process node to ensure that the risk score is less than the preset threshold; the second emergency strategy specifically divides the responsibilities of all monitored process nodes point by point, and readjusts the directed graph model to determine the responsibilities and processing time of the personnel at each process node to ensure that the risk score is less than the preset threshold.

Images