Flexible power control strategy optimization method and system based on dynamic clustering and agent

Abstract

The application discloses a kind of based on dynamic clustering and agent's flexible fee control strategy optimization method and system, it is related to electric power service management technical field.The method includes the following steps: establishing flexible fee control recommendation agent according to user historical power consumption behavior data, and constructing fee control service professional sub-label system;According to the actual power consumption of user, formulate flexible fee control strategy, use clustering analysis algorithm to carry out user group to flexible fee control recommendation agent, and formulate flexible fee control strategy based on group differentiation;According to the flexible fee control recommendation agent, the power consumption characteristics of user are fused and analyzed and predicted using Guangming electric power big model, and the flexible fee control strategy based on user individualization is generated;According to the process optimization of flexible fee control strategy execution situation.The present application can generate differentiated and intelligent fee control scheme based on user electric power image.

Description

Technical Field

[0001] This invention relates to the field of power business management technology, and in particular to a method and system for optimizing flexible cost control strategies based on dynamic clustering and intelligent agents. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] The implementation of fee control strategies directly impacts residents' electricity experience and reflects the company's power service quality and operational efficiency. Currently, there are three main problems with residential fee control implementation: First, user fee control parameter settings are unreasonable. Parameters such as warning thresholds, deduction thresholds, deduction amounts, and overdraft amounts are too homogeneous, failing to adequately match residents' electricity consumption scale and payment habits. This leads to frequent warnings and multiple deductions within a month, increasing fee control management costs and affecting customer service experience. Second, fee control collection costs are high, and the payment conversion rate is low. Third, fee control reminders and power outage periods lack differentiated management. They fail to comprehensively consider customers' electricity consumption habits, livelihood security, credit ratings, and other electricity usage tags, and lack assessment of the impact of power outages on customers, weakening the effectiveness of fee control reminders and thus affecting the customer's fee control service experience.

[0004] With changes in the internal and external environment and the implementation of policies to ensure people's livelihood, the problem of homogenization of current fee control strategies has become increasingly prominent. It is difficult to accurately adapt to the electricity consumption characteristics of different users, and it is also difficult to balance the relationship between electricity fee collection and people's livelihood. In particular, when implementing the requirement that users of people's livelihood should use power outages for payment reminders with caution, a uniform strategy is likely to cause service conflicts and fail to achieve a win-win situation for electricity fee management and user experience.

[0005] In the current fee control management system, the SMS notification templates, voice prompts, and power outage thresholds for residential users are generally standardized. However, due to the relatively lax fee control management at the grassroots level and the lack of necessary verification constraints in the business systems, the early warning mechanisms and automatic deduction strategies for fee control users tend to be similar, leading to a growing problem of homogenized fee control implementation. With changes in the internal and external environment and the implementation of policies to ensure people's livelihoods, the current fee control strategies are showing insufficient flexibility in accurately adapting to the diverse electricity consumption characteristics of users and balancing electricity fee collection with livelihood security, easily leading to service disputes. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a flexible cost control strategy optimization method and system based on dynamic clustering and intelligent agents. This method and system can generate differentiated and intelligent cost control schemes based on user electricity profiles, further improving the level of power supply services and user satisfaction, and promoting the deepening and improvement of the modern electricity billing system.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solution: The first aspect of this invention provides a flexible cost control strategy optimization method based on dynamic clustering and intelligent agents, comprising the following steps: A flexible cost control recommendation intelligent agent is established based on users' historical electricity consumption behavior data, and a professional sub-label system for cost control business is constructed. Flexible cost control strategies are formulated based on users' actual electricity consumption. Cluster analysis algorithms are used to segment users into groups for the flexible cost control recommendation agent, and flexible cost control strategies based on group differences are formulated. First, multidimensional data is preprocessed, then the differences between multidimensional data samples are calculated using distance metric methods. Then, clustering algorithms are used to dynamically formulate differentiated parameters for different groups, resulting in flexible cost control strategies based on group differences. Finally, the Guangming Power big data model is used to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation agent, generating flexible cost control strategies based on user personalization. Optimize processes based on the implementation of flexible cost control strategies.

[0008] Furthermore, a flexible cost control recommendation intelligent agent is established based on users' historical electricity consumption behavior data: Set up multi-dimensional behavioral indicators; Analyze user payment behavior based on behavioral metrics, and label users with corresponding sub-tags based on the analysis results.

[0009] Furthermore, behavioral indicators are divided into normal payment data indicators and abnormal payment data indicators. Normal payment data indicators include payment time preference, payment channel preference, payment amount preference, attention to message channel, attention to message type, sensitivity to power outages, enthusiasm for electricity bill activities, category of fee control application, and suspected invalid deductions. Abnormal payment data indicators include the success rate of early warning reminder payments, the success rate of outbound calls, and the effective duration of outbound calls.

[0010] Furthermore, the specific steps for generating a user-personalized flexible cost control strategy by utilizing the Guangming Power large model to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation intelligent agent are as follows: Based on user demand process analysis, a flexible cost control strategy is designed and a thought process is developed. Based on the user's actual needs, the system invokes the flexible cost control recommendation agent and the flexible cost control strategy deduction thinking chain to dynamically infer the user's electricity consumption characteristics, thereby obtaining a flexible cost control strategy based on the user's personalization.

[0011] Furthermore, the flexible cost control strategy includes a warning flexible strategy, a deduction flexible strategy, and a power outage flexible strategy. The warning flexible strategy is used for recommending warning and reminder content, the deduction flexible strategy is used for recommending content based on deduction status and channel preference, and the power outage flexible strategy is used for recommending content based on user credit evaluation.

[0012] Further process optimizations include optimization of flexible expense control strategy content, optimization of Guangming Power's large model parameters, optimization of flexible expense control recommendation agent labels, and optimization of flexible expense control strategy selection.

[0013] A second aspect of the present invention provides a flexible cost control strategy optimization system based on dynamic clustering and intelligent agents, comprising: The database construction module is configured to build a flexible cost control recommendation intelligent agent based on users' historical electricity consumption behavior data, and to build a cost control business professional sub-label system; The strategy generation module is configured to formulate flexible cost control strategies based on users' actual electricity consumption. It uses clustering analysis algorithms to segment users into groups based on the flexible cost control recommendation agent, and then formulates flexible cost control strategies based on group differentiation. Specifically, it first preprocesses multidimensional data, then calculates the differences between multidimensional data samples using distance metrics, and then dynamically formulates differentiated parameters for different groups using clustering algorithms, resulting in flexible cost control strategies based on group differentiation. Finally, it utilizes the Guangming Power large-scale model to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation agent, generating flexible cost control strategies based on user personalization. The strategy optimization module is configured to optimize processes based on the execution status of the flexible cost control strategy.

[0014] A third aspect of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the steps of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in the first aspect of the present invention.

[0015] A fourth aspect of the present invention provides a computer device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in the first aspect of the present invention.

[0016] A fifth aspect of the present invention provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in the first aspect of the present invention.

[0017] The above one or more technical solutions have the following beneficial effects: This invention discloses a flexible fee control strategy optimization method and system based on dynamic clustering and intelligent agents. Based on user payment behavior, electricity usage habits, electricity credit, and livelihood security data, combined with user electricity consumption and remaining available time, it formulates differentiated flexible fee control strategies for different user groups, including differentiated warning thresholds, deduction amounts, warning messages, and overdraft limits. This ensures electricity bill collection while minimizing the impact on residential users, effectively implementing policy requirements, and improving power supply service levels and user satisfaction. By researching and applying flexible fee control strategies for residential users, this invention further improves the standardization of power outage operations for overdue payments, reduces repeated warnings and multiple deductions within the month, lowers the cost of fee control SMS messages, and increases the success rate of warning reminders and payments, bringing significant direct economic benefits to the company. By constructing a dynamic model and intelligent agents to generate and execute differentiated flexible strategies, it provides electricity customers with a user-friendly fee control service experience. Differentiated power outage strategies will further reduce the number of residential power outage complaints, and user-friendly power outage timing selection will minimize user losses, thereby achieving a dual improvement in power supply service levels and user satisfaction.

[0018] This invention optimizes the bill payment-related tag profiles of electricity customers and constructs a flexible bill control recommendation intelligent agent. Based on the requirements of flexible bill control strategies, it builds a professional sub-tag system for bill control business, encompassing dimensions such as payment behavior, electricity usage habits, and electricity credit. Tag data is dynamically updated according to the tag business definitions, providing accurate data support for subsequent user classification and segmentation. By integrating user behavior data with domain expert knowledge, it constructs an intelligent agent architecture that includes electricity usage feature analysis, scenario-based strategy generation, and autonomous strategy execution. This achieves a paradigm shift from experience-driven to a data- and algorithm-driven approach.

[0019] This invention also enables the recommendation and execution of flexible expense control strategies. Based on the customer tag profile and flexible expense control strategy model, flexible expense control strategies are recommended to customers periodically, including flexible early warning thresholds, message reminder channels, early warning reminder scripts, and reminders for changes to expense control early warning agreements; reminders for changes to deduction thresholds, deduction values, and deduction channels; and flexible overdraft amounts, power outage times, and phone call cancellation scripts. Personalized services are executed according to the flexible expense control strategies, and flexible expense control reminder services are executed based on the user's flexible expense control strategy and the calculated available balance. This includes user flexible expense control strategy query, real-time strategy matching and application, and execution monitoring and feedback loop. Based on the strategy execution status, adjustments are made to SMS early warning strategies, outbound call strategies are dynamically optimized, and power outage messages are simplified and strengthened to further improve work efficiency. For users who have been recommended flexible strategies, staff can use the assistant service to answer questions about the recommendation basis and process; the assistant can also recommend flexible expense control strategies to individual users.

[0020] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a diagram illustrating the overall architecture of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram illustrating the application of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents in Embodiment 1 of the present invention; Figure 3 This is a flowchart of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents in Embodiment 1 of the present invention; Figure 4 This is a business schematic diagram of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents in Embodiment 1 of the present invention. Detailed Implementation

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0025] Example 1: Embodiment 1 of this invention provides a flexible cost control strategy optimization method based on dynamic clustering and intelligent agents, such as... Figures 1 to 3 As shown. This embodiment utilizes a flexible cost control strategy optimization method based on dynamic clustering and intelligent agents to construct a marketing system architecture, as follows. Figure 1 As shown, the system comprises an application layer, a service layer, a model support layer, and a data layer. The application layer is used to implement business applications such as visual tag management, flexible policy management for expense control users, execution of flexible policies for expense control users, optimization and governance of expense control user notification channels, and a flexible expense control recommendation assistant. The application layer implements business functions such as intelligent recommendation for flexible expense control by calling API services from the service layer. The model support layer is used to build a flexible expense control recommendation intelligent agent, utilizing the semantic understanding, intent recognition, cognitive reasoning, and content generation capabilities of the Guangming big data model to recommend suitable policies for expense control users. The data platform in the data layer is used to build a small model for generating flexible expense control policies, optimizing various tags (user habit tags, payment behavior tags, other category tags, etc.), generating flexible policies for expense control users across the entire dataset, and also accessing core marketing data from multiple sources (archival data, electricity bill data, payment data, electricity consumption data, etc.) to provide data support for flexible expense control policy recommendations.

[0026] like Figure 2 As shown, the flexible cost control user interface includes five main categories: tag management, flexible cost control user strategy management, flexible cost control user strategy execution, cost control user notification channel optimization and governance, and flexible cost control recommendation assistant. Tag management can be divided into three functions: electricity usage habit tag optimization, payment behavior tag optimization, and other category tag optimization. Flexible cost control user strategy execution can be divided into three functions: user flexible cost control strategy query, real-time strategy matching and application, and execution monitoring and feedback loop. Flexible cost control user strategy management can be divided into three functions: early warning flexible strategy, automatic deduction flexible strategy, and power outage flexible strategy. The flexible cost control recommendation assistant includes eleven functions: flexible cost control recommendation dialog box, prompt word management, mind chain management, early warning value adjustment suggestion analysis, early warning value recommendation, collection method recommendation, collection script recommendation, recommendation on whether to sign up for automatic deduction, automatic deduction value recommendation, power outage overdraft value recommendation, and recommendation result return.

[0027] like Figure 3 As shown, users can input text and use the intent engine to infer intent, then invoke an intelligent agent to form a flexible expense control service matrix. This matrix further utilizes small business models to generate flexible expense control strategies. By leveraging the general capabilities of the Guangming large model and its toolchain, a flexible expense control intelligent recommendation agent is constructed. This agent then calls the flexible expense control service via API to obtain business data and data from the data platform's small models. This supports the intelligent agent in providing intelligent recommendations for flexible expense control strategies, enabling dynamic early warnings and refined, differentiated early warning services.

[0028] This embodiment mainly integrates and interacts with the marketing system, the artificial intelligence platform, and the outbound calling platform. With the marketing system, it mainly obtains user profile data, electricity consumption and billing data, payment data, and SMS message push information. With the artificial intelligence platform, it mainly utilizes the semantic understanding, intent recognition, cognitive reasoning, content generation, and flexible expense control recommendation intelligent agent capabilities of the Guangming Big Data Model. With the outbound calling platform, it mainly integrates outbound calling services.

[0029] Specifically, the following steps are included: Step 1: Establish a flexible cost control recommendation intelligent agent based on users' historical electricity consumption behavior data, and build a cost control business professional sub-label system.

[0030] The flexible prepaid electricity control recommendation intelligent agent in this embodiment refers to an intelligent interaction and decision-making module that is based on users' historical electricity consumption behavior data (including file data, electricity bill data, payment data, electricity consumption data, etc.), integrates a professional sub-label system for prepaid electricity control business, simulates expert decision-making process according to preset business rules, and has the ability to analyze user electricity consumption characteristics, identify prepaid electricity control needs, generate strategy reasoning, push and execute strategies, and provide feedback optimization. It provides label support, data support, and reasoning support for Guangming Power's big model and is the core carrier connecting user needs and prepaid electricity control strategies.

[0031] Step 1.1: Set up multi-dimensional behavioral metrics.

[0032] In one specific implementation, based on the requirements of a flexible cost control strategy, a refined customer tagging system is constructed around dimensions such as payment behavior and electricity usage habits to deeply mine user electricity consumption characteristics. Simultaneously, user tagging results are updated regularly to ensure data timeliness and accuracy, providing a solid data foundation for subsequent user classification and segmentation.

[0033] To accurately characterize users' electricity consumption behavior, a professional sub-label system for fee control is constructed based on subsequent flexible early warning strategies, flexible deduction strategies, and flexible power outage strategies. This system consists of three dimensions: monthly electricity consumption, daily electricity consumption habits, and holiday electricity consumption characteristics. The system is regularly updated in conjunction with business data to provide a basis for the formulation of flexible fee control strategies.

[0034] Specifically, monthly electricity consumption refers to the statistical analysis of the distribution of monthly electricity consumption among residential users across the province, the formulation of business rules for labeling users' monthly electricity consumption, and the classification of users into tiers using quantile methods or tiered intervals, based on regional benchmark values, and the output of interval thresholds. A monthly automatic update task is configured to adjust the labels monthly based on the latest electricity consumption and synchronize them to the label library, ensuring that the labels match the user's current electricity consumption and providing a basis for precise service.

[0035] Daily electricity consumption habits are categorized by weekdays, holidays, and weekends. Hourly electricity consumption data is collected over N consecutive days, and the percentage of electricity consumption in each time period is calculated. Based on this percentage, tags are extracted and set to differentiate consumption habits such as high consumption in the morning and low consumption in the evening, balanced consumption throughout the day, and concentrated consumption at night. A dynamic update mechanism is also established, adjusting the tags monthly based on changes in user behavior to ensure they accurately reflect real-time electricity consumption patterns.

[0036] The characteristics of electricity consumption during holidays need to be clearly defined, covering statutory holidays, weekends, and special occasions, and user electricity consumption data should be collected for the corresponding time periods. The differences in electricity consumption between holidays and weekdays should be compared and analyzed, tags should be extracted and set to distinguish between residential and travel-related holiday electricity consumption characteristics, and these tags should be updated regularly based on business data.

[0037] Based on a refined customer tagging system, behavioral indicators are divided into normal payment data indicators and abnormal payment data indicators. Normal payment data indicators include payment time preference, payment channel preference, payment amount preference, message channel attention, message type attention, power outage sensitivity, electricity bill activity enthusiasm, fee control application category, and suspected invalid deductions.

[0038] The specific details of the indicator are as follows: (1) Payment time preference: Collect user payment record data, extract data such as payment date and specific time period, count the proportion of payment frequency in each time period, identify payment date and time period preference, and update the tags regularly in combination with business data.

[0039] (2) Payment channel preference: Sort out payment channel types and collect user payment data from each channel. Calculate the usage ratio of each channel, identify and label the user's main payment channels and auxiliary channels, and update the labels regularly in combination with business data to provide a basis for channel promotion and service optimization.

[0040] (3) Payment amount preference: Collect users' historical payment records, extract data such as payment amount, whether it is prepaid, and the degree of matching with the current electricity bill. Calculate the average of a single payment, distinguish between small-amount high-frequency, large-amount low-frequency, full-amount accurate, etc., and update the tags regularly in combination with business data to provide a basis for prepaid reminders and payment guidance.

[0041] (4) Message channel attention: Collect user message reach data in recent period of time, count the effective response ratio of each message channel, identify the main attention channels, extract and set tags, and update the tags regularly in combination with business data to provide a basis for accurate message push.

[0042] (5) Message Type Attention: Collect user response data for various messages such as electricity bill reminders, fee control warnings, fee control power outages, and electricity bill collection, and calculate the effective response percentage for each message type to identify user message preferences. Optimize tags based on user payment habits and update tags regularly based on business data.

[0043] (6) Power Outage Sensitivity: Collect historical power outage-related data from users, including complaints during power outages, repair response speed, and the correlation between power outage duration and response intensity. Statistically analyze user reactions under different power outage scenarios, identify sensitive features, and regularly update tags in conjunction with business data to provide a basis for power outage notifications and emergency services.

[0044] (7) Electricity Fee Activity Enthusiasm: Sort out the types of electricity fee activities and collect user activity participation data. Calculate the activity participation rate and the preference for each type of activity, identify characteristics such as high-frequency participation, participation only for full discounts, and never participation, optimize tags in combination with user spending power and electricity consumption scale, and update tags regularly in combination with business data to provide a basis for accurate activity push.

[0045] (8) Fee control application categories: Collect fee control data of users over a period of time, including information such as warning time, payment time, power outage status, and power restoration records. Filter out abnormal situations such as multiple warnings without recharging and no power outage, and classify the fee control application status to provide a basis for the supervision of fee control compliance.

[0046] (9) Suspected invalid deductions: Review the deduction rules and collect users' deduction records over a period of time, including data such as contract status, reasons for deduction failures, and frequency of manual payments. Filter out abnormal features such as long-term manual payments and frequent deduction failures under the contract status, and update regularly to identify invalid deductions, providing a basis for the formulation of flexible expense control strategies.

[0047] Abnormal payment data indicators include the success rate of early warning reminders for payments, the success rate of outbound calls being answered, and the effective duration of outbound calls being answered.

[0048] (a) Payment success rate of early warning reminders: Extract early warning reminder records of users within a certain period of time, associate them with payment data of the corresponding time period, and calculate the payment completion rate within a fixed time after each reminder. Update the success rate changes regularly to provide a basis for adjusting the reminder strategy.

[0049] (b) Outbound call success rate: Collect user outbound call records, extract data such as the number of calls made, the number of calls answered, and the reasons for missed calls, and calculate the proportion of answered calls to the total number of calls. Regularly update the success rate changes to provide a basis for adjusting outbound call strategies.

[0050] (c) Effective duration of outbound calls: Collect user outbound call record information, extract the connection duration of each call, and calculate the average effective duration and the proportion of each duration interval to provide a basis for improving outbound communication efficiency.

[0051] Step 1.2: Analyze user payment behavior based on behavioral indicators, and label users with corresponding sub-tags based on the analysis results.

[0052] Regarding normal payment data indicators, we statistically analyze the proportion of payment frequency in each time period, identify payment date and time preferences, and regularly update the tags in conjunction with business data to provide a basis for formulating flexible cost control strategies.

[0053] Regarding abnormal payment data metrics, we extract users' early warning reminder records over a certain period of time, associate them with payment data for the corresponding time period, and calculate the payment completion rate within a fixed time after each reminder. Combining the early warning reminder payment success rate, outbound call answering success rate, and effective outbound call answering duration with other payment reminder information, we regularly update the success rate changes to provide data support for adjusting reminder strategies.

[0054] Step 2: Develop a flexible cost control strategy based on the user's actual electricity consumption.

[0055] In one specific implementation, the flexible expense control strategy includes an early warning flexible strategy, a direct debit flexible strategy, and a power outage flexible strategy. The early warning flexible strategy is used for recommending content related to early warning and payment reminders, the direct debit flexible strategy is used for recommending content based on direct debit status and channel preference, and the power outage flexible strategy is used for recommending content related to user credit evaluation.

[0056] The flexible early warning strategy refers to dynamic early warning and collection recommendations based on multiple dimensions of user characteristics, such as monthly electricity consumption, collection rate, and credit rating. By using tags corresponding to the user's average daily electricity bill, suggestions are provided regarding whether the early warning value needs adjustment. When the user needs to adjust the early warning value, the system further differentiates the early warning threshold based on the user's tiered results and monthly electricity consumption tags (e.g., setting a threshold of 20 for electricity consumption between 0-100). Collection methods are recommended based on tags such as payment rate and whitelisted users (e.g., 1 SMS, 2 SMS, 1 voice call, 1 SMS + 1 voice call, 2 voice calls). Collection scripts are recommended based on tags such as the user's credit rating (basic: polite inquiry, basic: reminder notification, basic: urgent urging, additional: adding electricity prepayment activity information, additional: adding e-loan, e-ticket information).

[0057] The flexible deduction strategy dynamically recommends deductions based on multiple dimensions of user characteristics, including suspected invalid deductions and payment channel preferences. If a user is identified as having "suspected invalid deductions" and their "payment channel preference" is online, it is recommended to sign up for a valid payment channel. The recommended deduction value is dynamically calculated by combining the "multiple deductions within the month" tag (e.g., for high-frequency users) and "daily electricity bill" data: set at "monthly electricity bill × 120%". This approach covers fluctuations in electricity consumption while avoiding frequent deductions, accurately matching user habits and deduction needs, and improving cost control efficiency and payment timeliness.

[0058] The flexible power outage strategy is based on a user credit rating system, accurately pushing information on overdraft amounts and available credit. Regarding collection methods, it intelligently recommends the optimal approach by considering user message channel preferences, historical payment reminder success rates, and outbound call answering rates (e.g., prioritizing outbound calls for users with high answering rates). Simultaneously, it integrates telephone complaint ticket records, user credit ratings, and participation in electricity bill prediction activities to customize differentiated collection scripts for different user groups: emphasizing the consequences of default for users with frequent complaints or low credit, and guiding users towards future discount opportunities for those who participated in activities. Finally, by combining users' typical daily electricity usage habits and holiday electricity usage characteristics, it provides sales personnel with suggested time windows for implementing power outage reminders, thereby improving collection efficiency and optimizing customer experience.

[0059] Step 2.1: Use cluster analysis algorithms to segment users into groups for the flexible expense control recommendation agent and formulate flexible expense control strategies based on group differences.

[0060] This embodiment constructs a clustering model by selecting core dimensions such as payment habits, electricity consumption scale, and electricity usage habits to determine the optimal number of clusters. The clustering model has a real-time update function, which can automatically adjust the cluster centers when users add new data or their behavioral characteristics change, ensuring that the clustering results accurately match the latest user characteristics.

[0061] The specific steps are as follows: Step 2.1.1: First, preprocess the multidimensional data.

[0062] This embodiment collects multi-dimensional data such as users' electricity consumption, payment habits, and electricity usage habits over a period of time, and then cleans the data to remove missing values ​​and obviously erroneous data points.

[0063] Step 2.1.2: Calculate the differences between multidimensional data samples using the distance metric method.

[0064] Since most labels are categorical variables, traditional distance metrics fail. Therefore, this embodiment uses Gower distance to measure the difference between samples. For example, for samples X and Y, the Gower distance formula is: .

[0065] in, The Gower distance between samples X and Y It is the first The weights of each variable (the default weight of all variables is 1, which can be adjusted according to business needs, such as setting the weight of important variables to 2). It is the first of two samples The difference between the variables, where p is the total number of variables. The formula for calculating numerical variables is: .

[0066] in, , These are the first two samples respectively. The values ​​of the variables, , These are the maximum and minimum values ​​of the variable, respectively. Categorical variables are calculated using the following formula: .

[0067] For example, the first of samples X and Y If all variables are the same, then Other variables are handled in the same way. Finally, the Gower distance between each sample can be obtained.

[0068] Step 2.1.3: Then, use clustering algorithms to dynamically formulate differentiated parameters for different groups to obtain a flexible cost control strategy based on group differences.

[0069] In one specific implementation, this embodiment employs the incremental K-Means clustering method. Incremental K-Means is an improved form of the traditional K-Means clustering algorithm. Its core feature is that it does not require loading all the data at once, but rather processes the data step by step in batches or streaming. This characteristic gives it a significant advantage when processing large-scale data, memory-constrained scenarios, or real-time data streams.

[0070] Step 2.1.3.1: First, determine the number of clusters K. The elbow rule is to calculate the sum of squared distances (WCSS) from all samples in each cluster to the cluster center and plot the curve of WCSS as a function of K. The K value corresponding to the inflection point of the curve is usually the better choice.

[0071] When implementing the incremental K-Means algorithm, the first batch of data, i.e., the initial user sample set, is input, covering users with different electricity consumption and payment behavior characteristics to ensure the representativeness of the initial cluster centers. The traditional K-Means algorithm is then used for clustering to obtain K cluster centers, each corresponding to the mean vector of the core label of a user cluster. Then, for each new batch of input data... That is, newly added user samples or existing user tag updates, calculate The Gower distance from all cluster centers is used to assign the cluster to the nearest one. Then update the cluster using the following formula. Size and cluster centers: , .

[0072] in It is the pre-update cluster Sample size It was assigned to cluster X. Sample size These are the cluster centers before the update. It was assigned to a cluster The mean of the sample. This is the updated sample size. It is the updated cluster center.

[0073] Step 2.1.3.2: Then, for different cluster groups, customized differentiated strategy parameters are made based on the common characteristics of each cluster group.

[0074] Firstly, regarding the parameter settings for the flexible early warning strategy, the core basis is the average cluster center value of three labels: monthly electricity consumption, early warning reminder payment success rate, and power outage sensitivity. Three key parameters are set: early warning threshold, collection method, and collection script. For example, if a user group has an average monthly electricity consumption of 50 kWh (low electricity consumption), a high early warning reminder payment success rate, and high power outage sensitivity, then the early warning threshold is appropriately lowered to "15% of the average electricity bill over the past three months," the collection method is prioritized as "one SMS reminder," and the collection script is a gentle reminder notification to avoid frequent early warnings that could lead to complaints.

[0075] Second, the parameters for the flexible deduction strategy are set based on the cluster center mode / mean of three labels: payment channel preference, payment amount preference, and suspected invalid deductions. Three key parameters are set: deduction channel, deduction amount, and deduction optimization suggestions. For example, if a user group's payment channel preference mode is Alipay, and their payment amount preference mode is small, frequent, and suspected invalid deductions, the recommended deduction channel is Alipay, the deduction amount is set to "monthly average electricity bill × 80%", and the optimization suggestion is to "remind users to update their deduction bank card information to avoid deduction failures, and also recommend binding frequently used payment channels."

[0076] Third, the parameters for the power outage flexibility strategy are set based on the cluster center mode / mean of three labels: power outage sensitivity, credit rating, and daily electricity consumption habits. Three key parameters are set: overdraft amount, power outage timing, and outage call message. For example, if a user group has high power outage sensitivity, a high credit rating, and a daily electricity consumption mode concentrated at night, the overdraft amount is set to "monthly average electricity bill × 30%", the power outage timing is selected as "daytime non-peak electricity consumption period", and the outage call message clearly informs users of the power outage time, power restoration conditions, and contact channels.

[0077] Step 2.2: Utilize the Guangming Power big data model to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation agent, and generate a flexible cost control strategy based on user personalization.

[0078] Step 2.2.1: Based on the user needs process analysis, set up a flexible cost control strategy and develop a thought process chain.

[0079] In one specific implementation, the logical reasoning capabilities of the Guangming Big Data Model are utilized to integrate multi-dimensional user tags and set up a flexible expense control strategy deduction thought chain. This connects the entire structured logical reasoning process of demand identification, tag matching, feature analysis, strategy deduction, and result verification, ensuring the logic, coherence, and accuracy of the strategy deduction. This supports the Big Data Model's dynamic reasoning of personalized "one household, one policy" precise and flexible expense control strategies.

[0080] The flexible cost control strategy deduction thought chain in this embodiment is a structured reasoning framework used to guide the large model to systematically analyze user needs and generate strategies. Its construction process is as follows: Step 2.2.1.1: First, break down the business rules, sort out the entire business process of power fee control, and break down the core rules of each business link. For example, the early warning link needs to consider the scale of electricity consumption and payment channel preferences, and the power outage link needs to consider power outage sensitivity and credit rating. Clarify the tag association requirements corresponding to each rule.

[0081] Step 2.2.1.2: Next, carry out the association of the tag system, and match the decomposed business rules with the sub-tag system of the expense control business one by one to establish a business rule-tag mapping table. For example, the "early warning threshold setting rule" corresponds to the "monthly electricity consumption scale" and "payment amount preference" tags to ensure that the reasoning process is supported by tag data.

[0082] Step 2.2.1.3: Finally, model training and iteration are carried out. The mapping table, label data and historical fee control strategy cases are input into the Guangming Power big model to train the model to master the reasoning logic of demand-label-strategy and form a standardized thinking chain framework. Subsequently, combined with strategy execution feedback data (such as the success rate of early warning reminder payment), the reasoning nodes of the thinking chain and the priority of label matching are iteratively optimized to improve the accuracy of inference.

[0083] The flexible expense control strategy deduction thought chain constructed in this embodiment consists of five nodes: First, accurate demand identification, clarifying the user's core expense control needs (such as adjusting the warning threshold, recommending automatic deduction methods, optimizing the timing of power outage notifications, etc.), and excluding non-expense control needs; Second, targeted tag matching, based on the core needs, selecting the most relevant tags from the expense control business professional sub-tag system, for example, if the need is to adjust the warning threshold, then matching tags such as monthly electricity consumption scale, payment amount preference, and warning reminder payment success rate; Third, user feature analysis, combining the matched tag data to analyze the user's electricity consumption and payment behavior characteristics; Fourth, targeted strategy deduction, based on the feature analysis results and combined with expense control business rules, deducing flexible expense control strategy parameters suitable for the user, such as warning threshold, collection method, and wording; Fifth, result rationality verification, checking whether the deduced strategy parameters match the user's tag characteristics and business rules, and whether there are any logical contradictions (such as whether the warning threshold for low electricity consumption users is too high). If contradictions exist, the analysis is returned until the strategy is reasonable.

[0084] Step 2.2.2: Based on the user's actual needs, invoke the flexible cost control recommendation agent and the flexible cost control strategy deduction thinking chain to perform dynamic reasoning on the user's electricity consumption characteristics, and obtain a flexible cost control strategy based on the user's personalization.

[0085] Step 2.2.2.1: Receive user input.

[0086] In this embodiment, the user input is in natural language and is not limited to any format, such as: "Is the XXXX user warning strategy appropriate?"

[0087] Step 2.2.2.2: Perform keyword recognition and intent understanding on the user input to obtain the specific requirements of the flexible expense control business.

[0088] This embodiment uses natural language processing technology, combined with preset business categories (expense control flexible strategy Q&A / other Q&A), to deeply analyze the core business needs of user questions and clarify their category.

[0089] First, natural language preprocessing is performed. The user's input is segmented and stop words are removed to extract core keywords relevant to the business. For example, for a user input asking "Is the warning threshold in the warning strategy for user XXX appropriate?", the core keywords "warning strategy" and "warning threshold" are extracted after preprocessing. Second, the extracted features are matched against a pre-defined database of flexible cost control strategies. If the user's question involves queries, evaluations, generation, or adjustments to cost control strategies such as warnings, automatic deductions, and power outages, it is classified as a flexible cost control strategy question. If the user's intent is clearly outside this scope (e.g., checking electricity bills, inquiring about business procedures), or if no core topic is matched, it is categorized as "other questions".

[0090] It should be noted that this embodiment does not describe the requirements other than the flexible expense control business requirements; therefore, it is sufficient to identify the flexible expense control business requirements through natural language processing.

[0091] Step 2.2.2.3: Determine the specific areas and intentions based on the specific needs of the flexible expense control business.

[0092] In this embodiment, the expense control business domain corresponds to the flexible expense control strategy: for example, if the question involves warning values ​​or collection methods, it is determined to be related to the flexible warning strategy. If the question involves overdraft values ​​or power outages, it is determined to be related to the power outage strategy. If the question involves deduction values ​​or deductions, it is determined to be related to the deduction strategy.

[0093] Step 2.2.2.4: Based on the domain and intent judgment results, make flexible cost control recommendation agent calls.

[0094] Based on the identified user information and business segmentation, key information is extracted, and APIs are called to obtain user tag data (such as electricity consumption range, payment channel preference, historical payment reminder records, etc.).

[0095] Step 2.2.2.5: Based on the user tag information in the flexible expense control recommendation agent, call the early warning strategy recommendation thinking chain, and optimize the user's user-personalized flexible expense control strategy based on the needs of the business segment.

[0096] To optimize the flexible early warning strategy, the first step is to recommend threshold values. If a user's current early warning threshold does not match the electricity consumption range label, it is recommended to adjust the early warning threshold to "10% of the average electricity bill over the past 3 months" (adjusted according to actual business rules). Next, recommendations for collection methods are provided. For example, if the user prefers WeChat / State Grid online payment channels over the past 3 months, send an internal message. If they do not have the above preferences but have already linked their account to State Grid online, send an internal message from the State Grid website. In other cases, if the historical 24-hour timely collection rate is 100%, send an SMS; otherwise, send a voice message. Finally, recommendations for collection scripts are provided. For example, if there are telephone complaint records in the past six months, use SMS template A. If there are no complaints but the user has not participated in electricity prepayment activities within the past six months, use SMS template B. If there are no complaints and the user has participated in prepayment activities, use template C for small-amount, high-frequency users; otherwise, use template D.

[0097] To optimize the flexible automatic deduction strategy, the first step is to recommend deduction channels. For users with invalid automatic deductions, the signed-up channels are adjusted, and account information is updated. For example, if a user's "suspected invalid deduction" label is "yes" (i.e., they have signed up for automatic deductions but have been manually paying for a long time and have frequent deduction failures), and their payment channel preference label is "online channels (Alipay / WeChat)," it is recommended that the user cancel the original invalid automatic deduction channel and re-sign up for the corresponding online preferred channel. Next, the deduction amount is recommended, dynamically calculated based on electricity consumption and payment preferences to cover electricity fluctuations and avoid frequent deductions. For example, for users with stable monthly electricity consumption and a payment amount preference of "sufficient and accurate," the deduction amount is set to "105% to 110% of the average electricity bill of the past 3 months" to ensure that it just covers the current electricity bill. Third, optimization suggestions are provided to address issues such as deduction failures and invalid automatic deductions, offering targeted improvement solutions to improve the success rate of automatic deductions. For example, if a user frequently experiences deduction failures and the label shows "deduction failure due to insufficient account balance," the user is reminded to set up an automatic deduction account balance warning or bind a backup payment channel to avoid deduction failures due to insufficient balance.

[0098] To optimize the power outage flexibility strategy, the first step is to recommend overdraft amounts. This involves setting personalized overdraft amounts based on user credit ratings, power outage sensitivity, and monthly electricity consumption, balancing cost control risks with user convenience. For example, if a user has a "high credit rating," "high power outage sensitivity," and "medium or higher" monthly electricity consumption, the recommended overdraft amount would be "30%-40% of the average electricity cost over the past three months." The second step is to recommend power outage timings. Based on user daily electricity usage habits, holiday electricity usage characteristics, and power outage sensitivity, the strategy selects the time of day with the least impact on the user, avoiding outages during peak periods. Power outages during peak electricity consumption periods, such as when a user's daily electricity usage habit is tagged as "concentrated electricity consumption at night," should be scheduled during off-peak hours in the morning (9-11 am). Additionally, if a user's power outage sensitivity tag is "highly sensitive" and they have no historical complaint records, an extra emergency reminder can be added 2 hours before the outage to confirm whether the user can pay on time. Thirdly, a power outage call script is recommended, matching a template based on the user's credit rating, complaint records, and the level of attention they receive from news channels. This script clearly indicates the power outage and restoration information. For example, if a user's credit rating is "high credit" and they have no historical complaint records, template E should be used.

[0099] It should be noted that the templates AE in this embodiment are all preset template codes, and the content can be customized according to the actual situation.

[0100] Step 2.2.2.6: Generate a specific response based on the optimized flexible cost control strategy.

[0101] Leveraging the natural language generation capabilities of Guangming Power's large-scale model, specific responses for flexible cost control strategies are generated, such as: "Based on this user's electricity consumption, the following flexible cost control early warning strategy is recommended. Current threshold: XX yuan → Recommended adjustment to XX yuan (based on: electricity consumption range label); Collection method: Prioritize reminders via [channel]; Recommended script: [Specific script content]." The recommended script can be set based on preset templates.

[0102] Step 2.2.2.7: Set up a fallback mechanism.

[0103] If the generated answer does not meet the requirements, set an automatic reply, such as: "Dear, I'm sorry, I'm still learning and cannot provide you with a more accurate answer at the moment."

[0104] Step 3: Optimize the process based on the implementation of the flexible cost control strategy.

[0105] In one specific implementation, process optimization includes optimizing the content of flexible expense control strategies, optimizing the parameters of the Guangming Power large-scale model, optimizing the tags of the flexible expense control recommendation agent, and optimizing the selection of flexible expense control strategies. The implementation of flexible expense control strategy adaptation and adjustment supports services such as issuing warnings when users' spending falls below the flexible threshold, classifying overdraft limits based on user credit and issuing power outage notifications accordingly, determining power outage timing based on user electricity usage habits, recommending user-personalized flexible expense control strategies when handling expense control business online, pushing differentiated notification scripts according to user tags, and one-click electricity bill reporting, further improving the user's expense control service experience. The flexible expense control strategies mentioned in this step include flexible expense control strategies based on group differentiation and flexible expense control strategies based on user personalization.

[0106] This embodiment also establishes a flexible expense control strategy recommendation dialog box. When business personnel need to readjust expense control strategies for certain users, they can do so through a dialogue with the flexible expense control assistant tool.

[0107] Step 3.1: Implementation of flexible cost control strategy.

[0108] It supports querying the output results of the flexible expense control strategy model, such as user flexible warning thresholds and overdraft amounts. Business systems can display relevant data via API calls, providing a data foundation for the accurate triggering and application of the strategy. Flexible expense control strategies can be implemented based on actual circumstances. Expense control warning trigger: When a user's available balance is lower than the warning threshold configured for their group, the system automatically triggers a warning message push. The notification message is set according to the characteristics of the group and sent through the user's preferred main channel.

[0109] Overdraft limit management: When a user's available balance is less than zero, the system automatically grants a corresponding temporary overdraft limit based on the overdraft limit rules configured for their group, allowing the user to continue using electricity within the limit; when a user's available balance is negative and the preset overdraft conditions are met, the system notifies the user of the overdraft status and repayment requirements through designated channels.

[0110] Power outage timing decision and notification: When a power outage is necessary, the system combines the characteristic tags of the user's group and their historical electricity consumption habit tags to intelligently select the power outage time with the least impact on the user, and sends a clear power outage notification in advance through high-priority channels. The notification wording is set according to the characteristics of the group.

[0111] Business strategy recommendation: When users handle expense control-related business, the system intelligently recommends the most suitable flexible expense control strategy options based on the user's current group and tag profile on the business handling interface (such as suggesting adjustments to the warning threshold and prompting optimal deduction settings), to help users make decisions that better suit their own needs.

[0112] Step 3.2: Implement a closed loop of monitoring and feedback.

[0113] Step 3.2.1: Real-time monitoring dashboard.

[0114] Establish a monitoring platform for the implementation of flexible cost control strategies to track key performance indicators in real time, including but not limited to: the number and percentage of early warnings triggered by each group, the payment success rate after early warnings, the overdraft limit utilization rate, and the number of power outages during different time periods.

[0115] Step 3.2.2: Query abnormal execution status.

[0116] Monitor and record any abnormal situations during the execution of user flexible policies, and provide relevant anomaly query functions.

[0117] Step 3.2.3: Generate a cost control analysis report with one click.

[0118] It supports different periods such as daily, weekly, monthly, and yearly data extraction, using either scheduled or manual methods, based on preset templates to automatically extract business data and generate analysis reports with one click.

[0119] Step 3.2.4: Feedback data synchronization.

[0120] The strategy execution results data are synchronized to the model training library for scenarios such as verifying the effectiveness of the strategy model and dynamically updating user behavior tags.

[0121] Step 3.3: Optimization and Governance of Expense Control User Notification Channels Based on payment conversion rates, optimize alert SMS strategies for users with high SMS awareness and proactive payment behavior, while guiding users to switch to automatic deductions or use alternative notifications such as WeChat or in-app messages to reduce SMS costs. Analyze users' historical outbound call success rates and call durations, and adjust strategies for users with multiple failed outbound calls to reduce outbound call pressure. Optimize the power outage notification outbound call template, simplifying information, adjusting word order, highlighting key information, and improving notification delivery efficiency. Specific optimizations include: Step 3.3.1: Adjust the SMS alert strategy.

[0122] Based on user SMS alert records, a user group with high SMS awareness and active payment activity is selected, ensuring that this group receives SMS notifications first. For users with low SMS response rates and low payment conversion rates, the frequency of SMS messages sent to these users is gradually reduced. The number of SMS messages sent and the cost reduction are analyzed quarterly to verify the effectiveness of the strategy.

[0123] Step 3.3.2: Dynamic optimization of outbound calling strategy.

[0124] Integrate users' outbound call records over a recent period, extract metrics such as outbound call success rate, and set outbound call failure thresholds. For users who reach the threshold, suspend regular outbound calls and switch to sending SMS summaries, prioritizing information delivery through high-profile channels. Establish an outbound call monitoring mechanism to track daily outbound call volume and failure rate, and evaluate the improvement in outbound call efficiency monthly.

[0125] Step 3.3.3: Simplify and enhance power outage messages.

[0126] By adjusting the word order and tone of the power outage outbound call template, redundant information was simplified, and key information such as the power outage address, estimated outage time, and contact channels were highlighted at the beginning to ensure users hear effective information in the shortest possible time. Template testing was conducted with different user groups, and the information confirmation rate before and after optimization was compared. The template was iterated based on the test results.

[0127] Based on user profiles including payment behavior, electricity usage habits, electricity credit, and social welfare data, combined with data on user electricity consumption and remaining usable time, a flexible cost control strategy generation intelligent agent is constructed. This agent can accurately identify different user groups and dynamically generate and match highly personalized cost control strategies. By setting flexible parameters such as differentiated warning thresholds, warning messages, push channels, and overdraft limits, the agent achieves the goal of refined operation with a "one policy per household" approach, improving the user's electricity experience and enhancing the intelligence and humanization of electricity bill collection. Figure 4The diagram illustrates a specific business process, including: residential users activating the smart prepayment function; calculating the user's available balance based on the prepaid electricity fees collected on the monthly settlement date; obtaining user baseline flexible strategy parameters, such as warning thresholds and overdraft limits; and acquiring the user's on-site electricity usage status for benchmark comparison. Based on a comparison of the available balance with the warning thresholds and overdraft limits, appropriate strategies, message formats, and notification channels are selected. Finally, user feedback is analyzed and recorded based on the prepayment control implementation.

[0128] It should be noted that the content of this embodiment is based on user-authorized data and known publicly available data for analysis and training, and does not involve user privacy. All data used has been agreed upon by the user.

[0129] Example 2: Embodiment 2 of the present invention provides a flexible cost control strategy optimization system based on dynamic clustering and intelligent agents, comprising: The database construction module is configured to build a flexible cost control recommendation intelligent agent based on users' historical electricity consumption behavior data, and to build a cost control business professional sub-label system; The strategy generation module is configured to formulate flexible cost control strategies based on users' actual electricity consumption. It uses clustering analysis algorithms to segment users into groups based on the flexible cost control recommendation agent, and then formulates flexible cost control strategies based on group differentiation. Specifically, it first preprocesses multidimensional data, then calculates the differences between multidimensional data samples using distance metrics, and then dynamically formulates differentiated parameters for different groups using clustering algorithms, resulting in flexible cost control strategies based on group differentiation. Finally, it utilizes the Guangming Power large-scale model to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation agent, generating flexible cost control strategies based on user personalization. The strategy optimization module is configured to optimize processes based on the execution status of the flexible cost control strategy.

[0130] Example 3: Embodiment 3 of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the steps of the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in Embodiment 1 of the present invention.

[0131] Example 4: Embodiment 4 of the present invention provides a computer device, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the steps in the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in Embodiment 1 of the present invention.

[0132] Example 5: Embodiment 5 of the present invention provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in Embodiment 1 of the present invention.

[0133] The steps and methods involved in Examples 2, 3, 4 and 5 above correspond to those in Example 1. For specific implementation methods, please refer to the relevant description section of Example 1.

[0134] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0135] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0136] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A flexible cost control strategy optimization method based on dynamic clustering and intelligent agents, characterized in that, Includes the following steps: A flexible cost control recommendation intelligent agent is established based on users' historical electricity consumption behavior data, and a professional sub-label system for cost control business is constructed. Flexible cost control strategies are formulated based on users' actual electricity consumption. Cluster analysis algorithms are used to segment users into groups for the flexible cost control recommendation agent, and flexible cost control strategies based on group differences are formulated. First, multidimensional data is preprocessed, then the differences between multidimensional data samples are calculated using distance metric methods. Then, clustering algorithms are used to dynamically formulate differentiated parameters for different groups, resulting in flexible cost control strategies based on group differences. Finally, the Guangming Power big data model is used to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation agent, generating flexible cost control strategies based on user personalization. Optimize processes based on the implementation of flexible cost control strategies.

2. The flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in claim 1, characterized in that, Establish a flexible cost control recommendation agent based on users' historical electricity consumption data: Set up multi-dimensional behavioral indicators; Analyze user payment behavior based on behavioral metrics, and label users with corresponding sub-tags based on the analysis results.

3. The flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in claim 2, characterized in that, Behavioral indicators are divided into normal payment data indicators and abnormal payment data indicators. Normal payment data indicators include payment time preference, payment channel preference, payment amount preference, attention to message channel, attention to message type, sensitivity to power outages, enthusiasm for electricity bill activities, category of fee control application, and suspected invalid deductions. Abnormal payment data indicators include the success rate of early warning reminder payments, the success rate of outbound calls, and the effective duration of outbound calls.

4. The flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in claim 1, characterized in that, The specific steps for generating a user-personalized flexible cost control strategy by using the Guangming Power large model to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation intelligent agent are as follows: Based on user demand process analysis, a flexible cost control strategy is designed and a thought process is developed. Based on the user's actual needs, the system invokes the flexible cost control recommendation agent and the flexible cost control strategy deduction thinking chain to dynamically infer the user's electricity consumption characteristics, thereby obtaining a flexible cost control strategy based on the user's personalization.

5. The flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in claim 1, characterized in that, Flexible cost control strategies include early warning flexible strategies, direct debit flexible strategies, and power outage flexible strategies. The early warning flexible strategy is used for recommending content related to early warning and payment reminders. The direct debit flexible strategy is used for recommending content based on direct debit status and channel preferences. The power outage flexible strategy is used for recommending content related to user credit evaluation.

6. The flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in claim 1, characterized in that, The process optimization includes optimization of the flexible cost control strategy content, optimization of the parameters of the Guangming Power large model, optimization of the label of the flexible cost control recommendation agent, and optimization of the selection of flexible cost control strategy.

7. A flexible cost control strategy optimization system based on dynamic clustering and intelligent agents, characterized in that, include: The database construction module is configured to build a flexible cost control recommendation intelligent agent based on users' historical electricity consumption behavior data, and to build a cost control business professional sub-label system; The strategy generation module is configured to formulate flexible cost control strategies based on users' actual electricity consumption. It uses clustering analysis algorithms to segment users into groups based on the flexible cost control recommendation agent, and then formulates flexible cost control strategies based on group differentiation. Specifically, it first preprocesses multidimensional data, then calculates the differences between multidimensional data samples using distance metrics, and then dynamically formulates differentiated parameters for different groups using clustering algorithms, resulting in flexible cost control strategies based on group differentiation. Finally, it utilizes the Guangming Power large-scale model to perform fusion analysis and prediction of user electricity consumption characteristics based on the flexible cost control recommendation agent, generating flexible cost control strategies based on user personalization. The strategy optimization module is configured to optimize processes based on the execution status of the flexible cost control strategy.

8. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1-6.

10. A computer device, characterized in that, include: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the flexible cost control strategy optimization method based on dynamic clustering and intelligent agents as described in any one of claims 1-6.

Images