Electric charge sensitive user accurate analysis method based on AI algorithm analysis

Abstract

The invention discloses an electric charge sensitive user accurate analysis method based on AI algorithm analysis, and the method comprises the steps: obtaining the historical data of the power consumption behavior of a user, and calculating an adjustment coefficient according to the maximum value, the minimum value and the intermediate value of a variable, and the upper limit Xup and the lower limit Xdown of a preset value range; when the adjustment coefficient exceeds a preset threshold value, calculating a compensation amount, and synchronously correcting Xup and Xdown to generate an optimized value range; constructing an electricity charge sensitive user analysis model based on the optimized data; and performing user identification on the real-time power consumption behavior data by using the electricity charge sensitive user analysis model.

Description

Technical Field

[0001] The present invention relates to the technical field of power data analysis, and in particular to a method for accurately analyzing electricity-cost-sensitive users based on AI algorithm analysis. Background Art

[0002] In the field of electricity user management, identifying price-sensitive users has always been a key focus of the industry. These users react significantly to price fluctuations, often proactively adjusting their electricity usage behavior in response to price fluctuations—for example, reducing consumption during peak periods or switching to more economical packages. Existing technologies for identifying these users rely primarily on two approaches: one is simple classification based on fixed thresholds, such as labeling users with monthly average electricity usage exceeding a certain threshold as "highly sensitive users"; the other uses traditional clustering algorithms to group payment records or total electricity usage. While these methods can achieve basic classification, they have significant limitations. Static thresholds fail to capture users' dynamic responses to electricity prices. For example, users may be highly price-sensitive during peak air conditioning usage in the summer, but return to more normal levels during other seasons. A single label cannot capture these fluctuations. Cluster analysis is often limited to aggregate statistics, rarely incorporating key factors such as price policy adjustments and time-of-use price differences, resulting in a one-sided identification of behavioral patterns.

[0003] Existing technologies represented by the public patent CN115760252A further expose the pain points of the industry. Although this solution introduces a user rating mechanism, its core is still centered around optimizing collection efficiency: dividing users into levels by statistically analyzing historical arrears records, and triggering differentiated power outage strategies accordingly. This method is essentially post-management. It neither establishes a dynamic correlation model between user behavior and real-time electricity prices, nor converts the identification results into pre-services. For example, the system cannot predict the possible load transfer behavior of users after the adjustment of electricity price policies, and lacks energy-saving guidance for highly sensitive users or value-added incentives for less sensitive users. This "heavy collection, light service" model has caused a serious disconnect between user tags and subsequent services, which not only reduces user experience, but also restricts the optimal allocation of power resources.

[0004] A deeper technical gap lies in the lack of real-time and multidimensional behavior analysis. Current systems generally rely on quarterly or monthly historical data, making it difficult to respond promptly to user behavior changes brought about by changes in electricity price policies. For example, after the implementation of the new tiered electricity price policy, some users may quickly adjust their electricity usage habits within a week, but the system often lags for several weeks to capture this change. At the same time, the gradient differences in user sensitivity (such as high sensitivity, moderate sensitivity, and low sensitivity) are often simplified into binary classifications in existing technologies, lacking precise quantitative methods. This rough division directly leads to a mismatch of marketing resources - excessive energy-saving recommendations are pushed to low-sensitivity users, or ineffective standardized services are adopted for high-sensitivity users, ultimately affecting the effectiveness of electricity management.

[0005] These issues have collectively created a long-standing dilemma for the industry: insufficient accuracy in identifying users who are sensitive to electricity costs, a lack of personalized services, and difficulty improving the efficiency of two-way interactions between power companies and users. Breaking through the constraints of static rules and lagging analysis to establish a dynamic, accurate, and service-oriented identification system has become a pressing technical challenge. Summary of the Invention

[0006] To address the shortcomings and deficiencies of existing technologies, the present invention provides an AI-based method for accurately analyzing electricity-sensitive users. Its core innovation lies in the construction of a dynamic data optimization and model closed-loop collaborative mechanism. By analyzing the distribution characteristics of electricity usage behavior data (such as peaks, valleys, and median values) in real time, an adjustment coefficient is dynamically generated and an adaptive range correction is triggered. When the data deviates abnormally from the preset threshold, the system automatically calculates the compensation amount and synchronously adjusts the upper and lower limits of the value range, effectively eliminating extreme value interference and significantly improving the quality of input data. The optimized data drives convolutional neural network (CNN) model training, combined with real-time model performance evaluation (accuracy, recall, and other indicators) and iterative optimization, to form a closed-loop technical chain: data cleaning → range correction → model training → accuracy improvement. Finally, based on the identification results, a tiered marketing strategy is implemented (such as energy-saving guidance for highly sensitive users and points incentives for mildly sensitive users), achieving a complete process from behavior identification to service delivery.

[0007] The present invention specifically adopts the following technical solutions:

[0008] A precise analysis method for electricity cost-sensitive users based on AI algorithm analysis, including:

[0009] Obtain historical data on user electricity consumption behavior, based on the maximum, minimum, and intermediate values ​​of the variables, and the upper limit of the preset value range X up and the lower limit X down , calculate the adjustment coefficient;

[0010] When the adjustment coefficient exceeds the preset threshold, the compensation amount is calculated and X is corrected synchronously. up and X down Generate optimized value range;

[0011] Build an electricity cost-sensitive user analysis model based on the optimized data;

[0012] The electricity cost-sensitive user analysis model is used to perform user identification on real-time electricity usage behavior data.

[0013] Furthermore, the generation of the adjustment coefficient includes:

[0014] Detecting data distribution anomalies: Dynamically generating a logarithmic calculation benchmark based on the deviation relationship between the maximum, minimum, and median values ​​of the variable and the upper and lower limits of the preset range;

[0015] Quantify extreme value deviations:

[0016] The first deviation parameter is used to characterize the abnormal degree of the maximum value exceeding the preset upper limit;

[0017] The second deviation parameter is used to characterize the abnormal degree of the minimum value exceeding the preset lower limit;

[0018] Fusion abnormality indicator: Based on the logarithmic calculation benchmark, the first deviation parameter and the second deviation parameter, an adjustment coefficient is output to trigger a range correction.

[0019] Furthermore, the generation of the compensation amount includes:

[0020] Evaluate the regulation status: Generate a range correction intensity factor based on the proportional relationship between the regulation coefficient and the preset threshold;

[0021] Analyze the density of data distribution: Generate a distribution stability factor based on the concentration of data points relative to the median value except for extreme values;

[0022] Dynamic compensation control: determining the compensation base value by comparing the difference between the range correction intensity factor and the distribution stability factor;

[0023] Adaptive discreteness scaling: The compensation base value is calibrated based on the fluctuation characteristics of the data set and the final compensation amount is output.

[0024] Furthermore, the calculation of the compensation amount satisfies the following conditions:

[0025] Based on the proportional relationship between the adjustment coefficient and the preset threshold, constructing the first square root term;

[0026] Construct the second square root term based on the distribution deviation of data points from the median value except for the extreme values;

[0027] generating a dynamic correction factor by the difference between the first square root term and the second square root term;

[0028] Combined with the discrete degree calculation after data standardization, the compensation amount is output.

[0029] Furthermore, after obtaining the historical data of the user's electricity usage behavior, the historical data of the user's electricity usage is cleaned and normalized, and then the adjustment coefficient is calculated.

[0030] Furthermore, the electricity cost-sensitive user analysis model is a convolutional neural network model.

[0031] Furthermore, the maximum value and the minimum value of the variable correspond to the peak value and the valley value data of the applied electricity respectively.

[0032] Furthermore, the synchronous correction X up and X down include:

[0033] Scale the upper and lower limits of the preset range by multipliers (1+Q) and (1-Q) respectively;

[0034] Take the maximum value of the scaled value as the correction upper limit X up , the minimum value is used as the correction lower limit X down , generate an optimized value range, where Q is the compensation amount.

[0035] And, a precise analysis system for electricity cost-sensitive users, comprising:

[0036] The data dynamic processing unit is configured to perform:

[0037] Obtain historical data on user electricity consumption behavior, based on the maximum, minimum, and intermediate values ​​of the variables, and the upper limit of the preset value range X up and the lower limit X down , calculate the adjustment coefficient; when the adjustment coefficient exceeds the preset threshold, calculate the compensation amount and correct X synchronously up and X down Generate optimized value range;

[0038] The model building unit is configured to: build an electricity cost-sensitive user analysis model based on the optimized data;

[0039] The real-time identification unit is configured to use the electricity cost-sensitive user analysis model to perform user identification on the real-time electricity usage behavior data.

[0040] And, an electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0041] A non-transitory computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0042] Compared with the prior art, the present invention and its preferred embodiments have at least the following beneficial effects:

[0043] Dynamic data optimization improves model accuracy

[0044] By analyzing the distribution characteristics of electricity usage data (such as peaks, valleys, and median values) in real time, the system dynamically generates adjustment coefficients and triggers a range correction mechanism. When data deviates abnormally from preset thresholds, the system automatically calculates compensation and simultaneously adjusts the upper and lower limits of the value range, effectively eliminating interference from extreme values ​​and significantly improving input data quality. This innovation solves the problem of residual noise caused by traditional static thresholds or simple cleaning, providing a highly reliable data foundation for subsequent model training.

[0045] Closed-loop iteration of AI models enhances recognition accuracy

[0046] A convolutional neural network (CNN) model is constructed based on the optimized data. Real-time performance evaluation (accuracy, recall, and other metrics) is combined with an iterative optimization mechanism to form a closed-loop technology cycle: "data cleaning → range correction → model training → accuracy improvement." This design transcends the limitations of traditional single-shot modeling. By continuously optimizing the recognition logic through dynamic feedback, it deeply captures the behavioral characteristics of electricity-sensitive users and significantly improves classification accuracy.

[0047] Tiered service strategy improves electricity management efficiency

[0048] Translating identification results into personalized marketing strategies: For example, highly sensitive users can be provided with energy-saving guidance and habit optimization solutions, while less sensitive users can be given flexible guidance such as point-based incentives. This seamless integration of "precise identification and tiered services" solves the disconnect between user tags and service delivery in existing technologies, effectively improving user satisfaction and the efficiency of power resource allocation.

[0049] Full process automation reduces operation and maintenance costs

[0050] From dynamic data processing and model optimization to service strategy generation, the entire process requires no human intervention. This automated closed-loop design significantly reduces the operational and maintenance burden of traditional solutions, which require manual threshold calibration and model parameter adjustment. This approach is particularly suitable for high-concurrency analysis scenarios involving large numbers of users in power companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0052] Figure 1 This is a schematic diagram of a method for accurately analyzing electricity-cost-sensitive users based on AI algorithm analysis according to an embodiment of the present invention;

[0053] Figure 2 This is a flow chart of a method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to an embodiment of the present invention;

[0054] Figure 3 This is a flow chart of the core steps of an embodiment of the present invention. DETAILED DESCRIPTION

[0055] Hereinafter, specific embodiments of the present application will be described in detail with reference to the accompanying drawings. Based on these detailed descriptions, those skilled in the art will be able to clearly understand the present application and implement the present application. Without violating the principles of the present application, the features of different embodiments may be combined to obtain new implementations, or certain features of certain embodiments may be substituted to obtain other preferred implementations.

[0056] To make the features and advantages of the present invention more clearly understood, the following embodiments are specifically described in detail with reference to the accompanying drawings.

[0057] In order to solve the problem that the existing technology cannot accurately analyze electricity-sensitive users, cannot effectively identify electricity-sensitive users, and cannot provide users with personalized electricity services, resulting in poor electricity management and reduced efficiency and effectiveness of electricity services, an embodiment of the present invention provides a precise analysis solution for electricity-sensitive users based on AI algorithm analysis.

[0058] like Figure 3 As shown, the core implementation of this solution includes:

[0059] Obtain historical data on user electricity consumption behavior, based on the maximum, minimum, and intermediate values ​​of the variables, and the upper limit of the preset value range X up and the lower limit X down , calculate the adjustment coefficient;

[0060] When the adjustment coefficient exceeds the preset threshold, the compensation amount is calculated and X is corrected synchronously. up and X down Generate optimized value range;

[0061] Build an electricity cost-sensitive user analysis model based on the optimized data;

[0062] The electricity cost-sensitive user analysis model is used to perform user identification on real-time electricity usage behavior data.

[0063] In order to facilitate those skilled in the art to better understand the scheme of the present invention, the reference steps for its specific implementation are provided as follows: Figure 1 、 Figure 2 As shown:

[0064] Step S1: Data collection and processing:

[0065] Collecting user electricity usage history data based on big data, processing the user electricity usage history data based on big data, and determining user electricity usage feature data based on big data;

[0066] As a preferred solution of this embodiment, the user's electricity consumption history data based on big data is collected and the following operations are performed:

[0067] Monitor and collect users' electricity usage behavior to obtain users' electricity usage behavior data;

[0068] Among them, the user's electricity usage behavior includes the user's electricity usage time, electricity consumption and electricity price;

[0069] Monitor and collect users' electricity costs and obtain users' electricity cost data;

[0070] Among them, based on the user's electricity usage behavior data and the user's electricity cost data, the user's electricity usage history data based on big data is determined.

[0071] As a preferred solution of this embodiment, the user electricity consumption history data based on big data is processed by performing the following operations:

[0072] Obtain user electricity consumption history data based on big data;

[0073] Cleans and processes user electricity consumption history data based on big data, including:

[0074] Conduct consistency checks on user electricity consumption history data based on big data;

[0075] Check whether the user's electricity consumption history data meets the requirements based on the reasonable value range and mutual relationship of each variable in the user's electricity consumption history data;

[0076] Remove inconsistent data in the user's electricity usage history data that is beyond the normal range, logically unreasonable, or contradictory;

[0077] Process invalid and missing values ​​in user electricity consumption history data based on big data;

[0078] Remove invalid and missing data from the user's electricity consumption history data that is not valuable for accurate analysis of electricity-sensitive users;

[0079] Identify users who are sensitive to electricity costs and accurately analyze valuable historical electricity usage data;

[0080] Normalize the historical electricity consumption data of users who are sensitive to electricity charges after cleaning to accurately analyze the valuable data;

[0081] Eliminate the differences in the dimensions and value ranges between historical electricity consumption data of users that are valuable for accurate analysis of electricity cost-sensitive users, making different indicators comparable;

[0082] Determine the standardized historical data of electricity consumption of users;

[0083] Perform feature extraction on the user electricity consumption history data after normalization;

[0084] Analyze the user's electricity consumption patterns from the standardized user electricity consumption history data and identify the electricity cost-sensitive user patterns;

[0085] Determine user electricity consumption characteristic data based on big data.

[0086] Specifically, based on the reasonable value range and mutual relationship of each variable in the user's electricity consumption history data, check whether the user's electricity consumption history data meets the requirements, including:

[0087] Extracting the upper limit and lower limit of the preset initial reasonable value range corresponding to each variable in the user's electricity consumption history data;

[0088] Extracting the data value corresponding to each variable in the user's electricity usage history data, as well as the maximum value, minimum value and median value of each variable in the user's electricity usage history data;

[0089] The evaluation coefficient of each variable is obtained by using the maximum value, minimum value and intermediate value of each variable in the user's electricity consumption history data in combination with the upper limit and lower limit corresponding to the initial reasonable value range; wherein the adjustment coefficient is obtained by the following formula:

[0090]

[0091] Among them, P represents the adjustment coefficient; X max 、X min and X z Represents the maximum value, minimum value and median value of the variable respectively; X up and X down Respectively represent the upper limit and lower limit of the initial reasonable value range;

[0092] Comparing the evaluation coefficient of each variable with a preset evaluation coefficient threshold;

[0093] When the evaluation coefficient of the variable is lower than the preset evaluation coefficient threshold corresponding to each variable, the initial reasonable value range is used as the reasonable value range corresponding to each variable;

[0094] When the evaluation coefficient of the variable is not lower than the preset evaluation coefficient threshold corresponding to each variable, the initial reasonable value range is adjusted.

[0095] The technical solution achieves the following: By performing a range check on each variable in a user's electricity usage history data, the data is ensured to be within a reasonable range. This helps identify and eliminate outliers or erroneous data, thereby improving data quality and reliability. Dynamically Adjusting the Reasonable Value Range: This solution not only uses a preset initial reasonable value range (upper and lower limits) as a benchmark, but also dynamically adjusts these ranges by calculating an evaluation coefficient (P) for each variable and comparing it with a preset threshold. This adaptive approach more accurately reflects actual data characteristics, enhancing the system's flexibility and adaptability. Enhanced Intelligent Data Processing: This solution automatically calculates the evaluation coefficient using mathematical formulas (such as the provided adjustment coefficient formula) and automatically adjusts the reasonable value range based on the results, reducing the need for manual intervention and improving data processing efficiency and accuracy. Improved Data Analysis Accuracy: By ensuring data compliance and accuracy, this solution provides a more solid foundation for subsequent data analysis (such as electricity consumption forecasting, load analysis, and energy management). Based on high-quality data, the accuracy and reliability of analytical results will be significantly improved. Support decision-making: The adjusted reasonable value range not only helps identify anomalies in the data, but also provides a scientific basis for energy management, electricity bill calculation, troubleshooting, etc., supporting more accurate and efficient decision-making. Enhance the robustness of the system: Through strict inspection and dynamic adjustment of data, this technical solution can enhance the robustness of the entire system, reduce the risk of errors or failures caused by data problems, and improve the stability and reliability of the system. In summary, this technical solution improves the quality and reliability of data by dynamically adjusting the reasonable value range of user electricity consumption history data, providing strong support for subsequent data analysis and decision-making, while enhancing the flexibility and robustness of the system.

[0096] Specifically, when the evaluation coefficient of each variable is not lower than the preset evaluation coefficient threshold corresponding to each variable, the initial reasonable value range is adjusted, including:

[0097] When the evaluation coefficient of the variable is not lower than the preset evaluation coefficient threshold corresponding to each variable, extracting the variable value of each variable in the user's electricity consumption history data except the maximum value, the minimum value and the intermediate value;

[0098] Extracting the ratio between the evaluation coefficient of the variable and its corresponding preset evaluation coefficient threshold;

[0099] The value range compensation amount corresponding to the variable is obtained by using the ratio between the evaluation coefficient of the variable and its corresponding preset evaluation coefficient threshold value in combination with the variable value other than the maximum value, the minimum value and the intermediate value; wherein the value range compensation amount is obtained by the following formula:

[0100]

[0101] Among them, Q represents the compensation value of the value range; P represents the adjustment coefficient; P y represents the preset evaluation coefficient threshold; n represents the number of variables except the maximum value, minimum value and intermediate value; X up and X down Respectively represent the upper and lower limits of the initial reasonable value range; X max 、X min and X z Represents the maximum value, minimum value and median value of the variable respectively; X i Indicates the variable value corresponding to the i-th variable except the maximum value, minimum value and intermediate value;

[0102] Using the value range compensation amount to adjust the upper limit and lower limit of the initial reasonable value range to obtain adjusted upper limit and lower limit, and using the adjusted upper limit and lower limit to integrate into a reasonable value range;

[0103] The adjusted upper and lower limits are obtained by the following formula:

[0104]

[0105] Among them, X upt and X downt They represent the upper and lower limits after adjustment respectively; Q represents the compensation amount of the value range; X up and X down They represent the upper and lower limits of the initial reasonable value range respectively.

[0106] The technical effect of the above-mentioned technical solution is that, by introducing an evaluation coefficient and comparing it with a preset threshold, it enables more precise adjustment of the initial reasonable value range. The initial range is adjusted only when the variable's evaluation coefficient is at least the preset evaluation coefficient threshold, ensuring that the range is optimized only when the variable data demonstrates high importance or relevance. Range adjustment utilizes more comprehensive data information. During the adjustment process, not only the maximum, minimum, and median values ​​of the variable (which are typically extreme or representative values) are considered, but also more data points beyond these three values. This provides a more comprehensive reflection of the variable's distribution, resulting in a more reasonable value range. By calculating the ratio of the variable's evaluation coefficient to the preset threshold and determining the value range compensation based on this ratio, dynamic adjustment of the initial reasonable value range is achieved. This mechanism automatically adjusts the range based on data changes and variable importance, improving the system's adaptability and flexibility. When calculating the value range compensation, the maximum, minimum, and median values ​​are excluded, which helps reduce the undue influence of extreme values ​​on the range adjustment results, making the adjusted range more consistent with actual conditions and the overall data distribution. Improved decision-making support: By optimizing the appropriate value range, this technical solution can provide a more accurate and reliable basis for subsequent decision-making. Whether in energy management, user behavior analysis, or other related fields, more precise value ranges can help decision-makers make more scientific and reasonable decisions. In summary, by introducing evaluation coefficients, utilizing comprehensive data information, and a dynamic adjustment mechanism, this technical solution significantly improves the accuracy and adaptability of the appropriate value range, providing strong support for decision-making in related fields.

[0107] Step S2: Model construction and optimization:

[0108] Based on the precise analysis needs of electricity-cost-sensitive users based on AI algorithm analysis, a precise analysis model for electricity-cost-sensitive users based on AI algorithm analysis is constructed. This model is evaluated and optimized to determine the optimal precise analysis model for electricity-cost-sensitive users.

[0109] As a preferred solution of this embodiment, a precise analysis model for electricity-cost-sensitive users based on AI algorithm analysis is constructed, and the following operations are performed:

[0110] Obtain user electricity consumption characteristic data based on big data, divide the user electricity consumption characteristic data based on big data, and determine the precise analysis training set and precise analysis test set of electricity cost-sensitive users;

[0111] According to the demand for precise analysis of electricity bill sensitive users based on AI algorithm analysis, a convolutional neural network model suitable for precise analysis of electricity bill sensitive users is selected. Based on deep learning technology, the convolutional neural network model suitable for precise analysis of electricity bill sensitive users is trained using a precise analysis training set for electricity bill sensitive users. The convolutional neural network model suitable for precise analysis of electricity bill sensitive users can autonomously learn the precise analysis behavior of electricity bill sensitive users from the precise analysis training set for electricity bill sensitive users, and accurately identify electricity bill sensitive users, thereby determining a precise analysis model for electricity bill sensitive users based on AI algorithm analysis.

[0112] As a preferred solution of this embodiment, the precise analysis model of electricity cost-sensitive users based on AI algorithm analysis is evaluated and optimized, and the following operations are performed:

[0113] Obtain an accurate analysis model for electricity-cost-sensitive users based on AI algorithm analysis;

[0114] Based on the electricity cost-sensitive user precision analysis test set, the electricity cost-sensitive user precision analysis model based on AI algorithm analysis is tested and its performance is evaluated to determine whether the electricity cost-sensitive user precision analysis model based on AI algorithm analysis can achieve the expected effect of accurately analyzing and identifying electricity cost-sensitive users;

[0115] The precise analysis model for electricity-cost-sensitive users is repeatedly iterated, and the model performance indicators within each iteration cycle, including accuracy, precision, and recall, are calculated. Based on the model performance indicators, the evaluation test results based on the precise analysis model for electricity-cost-sensitive users are determined.

[0116] Specifically, accuracy represents the ratio of the number of samples correctly predicted by the model to the total number of samples. The accuracy is obtained by the following formula:

[0117] ;

[0118] Among them, TP is a true positive example, FP is a false positive example, TN is a true negative example, and FN is a false negative example.

[0119] Specifically, precision represents the proportion of samples that are actually positive among all samples predicted to be positive samples. The precision is obtained by the following formula:

[0120] ;

[0121] Among them, TP is a true positive example and FP is a false positive example.

[0122] Specifically, the recall rate represents the proportion of samples that are correctly predicted by the model among all samples that are actually positive samples. The recall rate is obtained by the following formula:

[0123] ;

[0124] Among them, TP is a true positive example and FP is a false positive example.

[0125] As a preferred solution of this embodiment, in data analysis, "true positives", "false positives", "true negatives" and "false negatives" are important concepts for evaluating the performance of classification models, and their definition based on the confusion matrix is ​​used to describe the relationship between the model prediction results and the actual results.

[0126] Among them, True Positive (TP) is the sample that the model predicts to be positive and is actually positive. It means that the model correctly identifies the positive samples and is a reflection of good model performance. For example, in disease detection, the true positive refers to the number of patients correctly diagnosed.

[0127] Among them, false positives (FP) are samples that the model predicts to be positive but are actually negative. This means that the model mistakenly predicts negative samples as positive, that is, "false positives". For example, in spam filtering, false positives may mean that normal emails are mistakenly marked as spam.

[0128] Among them, True Negative (TN) is a sample that the model predicts to be negative and is actually negative. It means that the model correctly identifies the negative sample, that is, "correctly rejected". For example, in fraud detection, a true negative example means that the model correctly identifies a non-fraudulent transaction as a normal transaction.

[0129] Among them, false negatives (FN) are samples that the model predicts to be negative but are actually positive. This means that the model mistakenly predicts positive samples as negative, that is, "missing reports". This is particularly important in some high-risk scenarios, such as medical diagnosis, where missed diagnoses may have serious consequences.

[0130] Therefore, the accuracy, precision and recall rates can be used to accurately evaluate the analysis and recognition performance of the electricity cost-sensitive user precision analysis model based on AI algorithm analysis, which facilitates better analysis and identification of electricity cost-sensitive users.

[0131] Analyze the evaluation test results of the precise analysis model for electricity-cost-sensitive users. When the precise analysis model for electricity-cost-sensitive users based on AI algorithm analysis cannot achieve the expected effect of accurately analyzing and identifying electricity-cost-sensitive users, collect user feedback data from actual usage scenarios, re-evaluate the precise analysis model for electricity-cost-sensitive users based on the user feedback data, analyze the performance of the precise analysis model for electricity-cost-sensitive users, clarify the optimized structure or parameters of the precise analysis model for electricity-cost-sensitive users, and determine an optimization adjustment plan based on the precise analysis model for electricity-cost-sensitive users;

[0132] According to the optimization and adjustment plan based on the precise analysis model of electricity cost sensitive users, the precise analysis model of electricity cost sensitive users is optimized and adjusted. After repeated iterations, the precise analysis model of electricity cost sensitive users based on AI algorithm analysis can achieve the expected effect of accurately analyzing and identifying electricity cost sensitive users, thereby determining the optimal precise analysis model of electricity cost sensitive users.

[0133] Step S3: Accurate analysis and identification:

[0134] Deploy the optimal electricity-cost-sensitive user precision analysis model and deploy it in the actual electricity-cost-sensitive user precision analysis and identification environment;

[0135] Obtain real-time user electricity usage data and input it into the optimal electricity-cost-sensitive user precision analysis model. Based on the optimal electricity-cost-sensitive user precision analysis model, perform precision analysis and pattern recognition on the user's real-time electricity usage data, and determine the precision analysis and identification results of electricity-cost-sensitive users based on AI algorithm analysis;

[0136] Based on the real-time data of users' electricity consumption and the precise analysis and identification results of electricity-sensitive users based on AI algorithm analysis, a precise analysis report on electricity-sensitive users is generated, stored, and displayed in a visual form.

[0137] Step S4: Intelligent marketing management and control:

[0138] Analyze the precise analysis and identification results of electricity-sensitive users based on AI algorithm analysis, formulate intelligent marketing and control plans for electricity-sensitive users, and conduct intelligent marketing and control for electricity-sensitive users based on the intelligent marketing and control plans.

[0139] As a preferred solution of this embodiment, intelligent marketing management and control is performed on electricity-sensitive users by performing the following operations:

[0140] Obtain accurate analysis and identification results of electricity-sensitive users based on AI algorithm analysis;

[0141] Analyze the precise analysis and identification results of electricity-sensitive users based on AI algorithms, find out the personalized electricity usage behavior of electricity-sensitive users, and determine the intelligent marketing and control plan for electricity-sensitive users based on their personalized electricity usage behavior;

[0142] Based on the intelligent marketing and control solution for electricity-sensitive users, intelligent marketing and control are carried out for electricity-sensitive users.

[0143] Among them, measures are taken to optimize electricity usage habits and recommend energy-saving products for users who are highly sensitive to electricity costs;

[0144] Among them, for users who are slightly sensitive to electricity bills, personalized services and point rewards are adopted to attract them to maintain a lower level of electricity consumption.

[0145] Therefore, this embodiment obtains real-time electricity usage data of users, performs precise analysis and identification of electricity cost-sensitive users based on the optimal precise analysis model of electricity cost-sensitive users, determines the precise analysis and identification results of electricity cost-sensitive users based on AI algorithm analysis, mines and analyzes the precise analysis and identification results of electricity cost-sensitive users based on AI algorithm analysis, formulates an intelligent marketing and control plan for electricity cost-sensitive users, and performs intelligent marketing and control on electricity cost-sensitive users based on the intelligent marketing and control plan for electricity cost-sensitive users. It can accurately analyze electricity cost-sensitive users, effectively identify electricity cost-sensitive users, provide users with personalized electricity consumption services, and improve electricity consumption management effects and the efficiency and effectiveness of power services.

[0146] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is used to implement one or more instructions, specifically for loading and executing one or more instructions in a computer storage medium to implement the above method.

[0147] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium having a computer program stored thereon, which, when executed by a processor, performs the above-described method. The storage medium may be any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0148] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present disclosure. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0149] The above shows and describes the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present disclosure. Various changes and improvements may be made to the present disclosure without departing from the spirit and scope of the present disclosure, and such changes and improvements shall fall within the scope of the present disclosure.

[0150] The present invention is not limited to the above-mentioned optimal implementation mode. Anyone can derive various other forms of accurate analysis methods for electricity-sensitive users based on AI algorithm analysis under the inspiration of the present invention. All equal changes and modifications made according to the scope of the patent application of the present invention should be covered by the scope of the present invention.

Claims

1. A precise analysis method for electricity cost-sensitive users based on AI algorithm analysis, characterized in that: include: Obtain historical data on user electricity consumption behavior, based on the maximum, minimum, and intermediate values ​​of the variables, and the upper limit of the preset value range X up and the lower limit X down , calculate the adjustment coefficient; When the adjustment coefficient exceeds the preset threshold, the compensation amount is calculated and X is corrected synchronously. up and X down Generate optimized value range; Build an electricity cost-sensitive user analysis model based on the optimized data; The electricity cost-sensitive user analysis model is used to perform user identification on real-time electricity usage behavior data.

2. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 1 is characterized by: The generation of the adjustment coefficient includes: Detecting data distribution anomalies: Dynamically generating a logarithmic calculation benchmark based on the deviation relationship between the maximum, minimum, and median values ​​of the variable and the upper and lower limits of the preset range; Quantify extreme value deviations: The first deviation parameter is used to characterize the abnormal degree of the maximum value exceeding the preset upper limit; The second deviation parameter is used to characterize the abnormal degree of the minimum value exceeding the preset lower limit; Fusion abnormality indicator: Based on the logarithmic calculation benchmark, the first deviation parameter and the second deviation parameter, an adjustment coefficient is output to trigger a range correction.

3. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 1 is characterized by: The generation of the compensation amount includes: Evaluate the regulation status: Generate a range correction intensity factor based on the proportional relationship between the regulation coefficient and the preset threshold; Analyze the density of data distribution: Generate a distribution stability factor based on the concentration of data points relative to the median value except for extreme values; Dynamic compensation control: determining the compensation base value by comparing the difference between the range correction intensity factor and the distribution stability factor; Adaptive discreteness scaling: The compensation base value is calibrated based on the fluctuation characteristics of the data set and the final compensation amount is output.

4. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 3 is characterized by: The calculation of the compensation amount meets the following conditions: Based on the proportional relationship between the adjustment coefficient and the preset threshold, constructing the first square root term; Construct the second square root term based on the distribution deviation of data points from the median value except for the extreme values; generating a dynamic correction factor by the difference between the first square root term and the second square root term; Combined with the discrete degree calculation after data standardization, the compensation amount is output.

5. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 1 is characterized by: After obtaining the historical data of the user's electricity usage behavior, the historical data of the user's electricity usage is cleaned and normalized, and then the adjustment coefficient is calculated.

6. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 1 is characterized by: The electricity cost-sensitive user analysis model is a convolutional neural network model.

7. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 1 is characterized by: The maximum and minimum values ​​of the variables correspond to the peak and valley data of applied electricity, respectively.

8. The method for accurately analyzing electricity cost-sensitive users based on AI algorithm analysis according to claim 7 is characterized by: The synchronous correction X up and X down include: Scale the upper and lower limits of the preset range by multipliers (1+Q) and (1-Q) respectively; Take the maximum value of the scaled value as the correction upper limit X up , the minimum value is used as the correction lower limit X down , generate an optimized value range, where Q is the compensation amount.

9. A precise analysis system for electricity cost-sensitive users, characterized by: include: The data dynamic processing unit is configured to perform: Obtain historical data on user electricity consumption behavior, based on the maximum, minimum, and intermediate values ​​of the variables, and the upper limit of the preset value range X up and the lower limit X down , calculate the adjustment coefficient; when the adjustment coefficient exceeds the preset threshold, calculate the compensation amount and correct X synchronously up and X down Generate optimized value range; The model building unit is configured to: build an electricity cost-sensitive user analysis model based on the optimized data; The real-time identification unit is configured to use the electricity cost-sensitive user analysis model to perform user identification on the real-time electricity usage behavior data.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images