Power customer demand evaluation method and system based on fuzzy evaluation
Through the electricity customer demand assessment method based on fuzzy evaluation, deep learning and fuzzy logic are combined to achieve high-order feature adaptive fuzzification and dynamic optimization of rules, which solves the problems of insufficient evaluation accuracy and flexibility in existing methods and improves the adaptability and recognition accuracy of electricity customer demand assessment.
Patent Information
- Application Number
- CN202510645065.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-20
- Publication Date
- 2025-09-23
AI Technical Summary
Existing electricity customer demand assessment methods are difficult to adapt to the dynamics of load characteristics and environmental changes, have insufficient multi-source feature fusion capabilities, and lack adaptive optimization of fuzzy inference rules, resulting in insufficient assessment accuracy and flexibility.
A fuzzy evaluation-based method is adopted. High-order feature vectors are received through the deep learning feature extraction layer. Fuzzy processing is performed using the fuzzy membership function of normalized deformation and nonlinear mapping. Rule matching and reasoning are performed in combination with the fuzzy rule base, and quantitative scores or discrete grade results are generated through weighted integration.
The adaptability and accuracy of the model in environments with diverse load characteristics have been improved, supporting flexible adaptation to different application scenarios and improving recognition accuracy and applicability.
Smart Images
Figure CN120688770A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power demand assessment, and in particular to a method and system for assessing power customer demand based on fuzzy evaluation. Background Art
[0002] As the level of informatization and intelligence in power systems continues to improve, demand-side management of power customers has become a crucial component of building smart grids. Traditional load forecasting methods have gradually evolved from single-time series modeling to intelligent assessment systems that incorporate multi-source heterogeneous data and multi-factor comprehensive analysis. In recent years, deep learning has demonstrated superior performance in power load feature extraction and complex pattern recognition. Fuzzy logic, owing to its excellent uncertainty handling and language expression capabilities, has gained widespread application in demand-side response decision-making and customer behavior analysis. Existing research generally combines deep learning with fuzzy reasoning to achieve accurate forecasting and interpretable assessment of power customer demand.
[0003] However, existing electricity customer demand assessment methods based on the integration of deep learning and fuzzy reasoning still have significant shortcomings in practical applications. First, current methods generally rely on static feature partitioning and fixed fuzzy membership function settings, making them difficult to adapt to the dynamic changes in electricity customer load behavior due to multiple environmental factors, resulting in insufficient accuracy and flexibility in the assessment model. Second, the multi-source data feature fusion process lacks a systematic and robust deep fusion mechanism. Heterogeneous information such as time series features, discrete features, and environmental variables is often simply spliced together within the network, failing to fully explore their potential correlations and importance weights, which affects the comprehensive expressive power of feature extraction. Furthermore, existing fuzzy reasoning layers often adopt a fixed rule and a single activation mechanism, failing to dynamically optimize rule contributions based on actual customer response results. This leads to poor stability and adaptability in the inference results. Especially in the context of increasing demand response management requirements, traditional static fuzzy reasoning methods cannot effectively support the dynamic exploration and prediction of customer potential. Therefore, there is an urgent need for a new electricity customer demand assessment method that can achieve adaptive fuzzification of high-order features, dynamic rule optimization reasoning, and comprehensive quantitative output to improve the accuracy, flexibility, and interpretability of demand identification. Summary of the Invention
[0004] In view of the above-mentioned problems, the present invention is proposed.
[0005] Therefore, the technical problems solved by the present invention are: the existing electricity customer demand assessment methods have the problem of being difficult to adapt to the dynamics of load characteristics and environmental changes, the problem of insufficient multi-source feature fusion capabilities, the problem of lack of adaptive optimization of fuzzy inference rules, and the problem of how to achieve accurate assessment of customer demand potential based on the fusion of deep feature extraction and fuzzy logic reasoning.
[0006] To solve the above technical problems, the present invention provides the following technical solutions: a method for evaluating electricity customer demand based on fuzzy evaluation, comprising receiving high-order feature vectors from a deep learning feature extraction layer, and dividing them into multiple evaluation index dimensions according to load trend characteristics, demand response sensitivity characteristics, customer behavior pattern characteristics, and environmental impact factor characteristics; fuzzifying the feature values using a fuzzy membership function of normalized deformation and nonlinear mapping to generate feature membership, and performing rule matching and reasoning using a fuzzy rule base based on the feature membership to generate a fuzzy reasoning output; and defuzzifying the fuzzy reasoning output through weighted integration to obtain a quantitative score or discrete grade result of the customer demand potential.
[0007] As a preferred solution of the electricity customer demand assessment method based on fuzzy evaluation described in the present invention, the evaluation index dimensions include electricity consumption behavior feature dimensions, environment-related feature dimensions, economic policy feature dimensions and other auxiliary feature dimensions. Fuzzy processing parameters are set separately for each dimension according to the feature type to perform membership calculation and interval division.
[0008] As a preferred solution of the electricity customer demand assessment method based on fuzzy evaluation described in the present invention, the fuzzy membership function is constructed based on a combination of normalization processing, power transformation and exponential mapping, and the fuzzy interval boundaries and mapping sensitivity parameters can be dynamically adjusted and updated based on historical load data and real-time feedback.
[0009] As a preferred solution of the electricity customer demand assessment method based on fuzzy evaluation described in the present invention, the dynamic adjustment of the fuzzy interval boundary is achieved by combining gradient descent based on the loss function with a regularization term, and the fuzzy membership function shape and interval position are updated by minimizing the customer demand prediction deviation.
[0010] As a preferred solution of the electricity customer demand assessment method based on fuzzy evaluation described in the present invention, the fuzzy rule base is divided into sub-rule sets according to customer categories, each sub-rule set contains several feature combination patterns and corresponding demand level labels, and supports dynamic addition, deletion and revision of rule content based on feature changes.
[0011] As a preferred solution of the electricity customer demand assessment method based on fuzzy evaluation described in the present invention, the rule matching and reasoning include performing multiplication and combination operations on the membership of each feature, and realizing weighted fusion output of multiple activation rules by setting feature contribution weights and rule activation indexes.
[0012] As a preferred solution of the electricity customer demand assessment method based on fuzzy evaluation described in the present invention, the weighted integral defuzzification adopts weighted calculation of fuzzy output membership within a set integral interval, the integral weighting factor introduces the demand potential bias index, and the output quantitative score is used to generate a customer demand potential index.
[0013] Another object of the present invention is to provide an electricity customer demand assessment system based on fuzzy evaluation, which can receive high-order feature vectors from a deep learning feature extraction layer and perform fuzzy processing using fuzzy membership functions of normalized deformation and nonlinear mapping, thereby solving the problem of insufficient model adaptability caused by static feature division and fixed fuzzy intervals in current electricity customer demand assessment technology.
[0014] As a preferred solution of the power customer demand assessment system based on fuzzy evaluation described in the present invention, it includes: a high-order feature classification module, a fuzzy rule reasoning module, and a demand potential defuzzification output module;
[0015] The high-order feature classification module is used to receive high-order feature vectors from the deep learning feature extraction layer and divide them into multiple evaluation index dimensions according to load trend characteristics, demand response sensitivity characteristics, customer behavior pattern characteristics and environmental impact factor characteristics;
[0016] The fuzzy rule reasoning module is used to perform fuzzy processing on the eigenvalues by using the fuzzy membership function of normalized deformation and nonlinear mapping to generate characteristic membership, and perform rule matching and reasoning using the fuzzy rule base according to the characteristic membership to generate fuzzy reasoning output;
[0017] The demand potential defuzzification output module is used to defuzzify the fuzzy reasoning output through weighted integration to obtain a quantitative score or discrete grade result of the customer demand potential.
[0018] A computer device includes a memory and a processor. The memory stores a computer program. The processor executes the computer program to implement the steps of a method for evaluating power customer demand based on fuzzy evaluation.
[0019] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for evaluating power customer demand based on fuzzy evaluation.
[0020] Beneficial effects of the present invention: The electricity customer demand assessment method based on fuzzy evaluation provided by the present invention constructs a fuzzy membership function by normalizing and deforming high-order eigenvectors and performing nonlinear mapping to complete the dynamic fuzzification processing of eigenvalues. The adaptability and accuracy of the model in a diversified load characteristic environment are improved. The fuzzy rule base is constructed in a sub-rule set manner divided according to customer type, and the addition, deletion, modification and expansion of rules are supported to achieve flexible adaptation to different application scenarios. The scalability and adaptability of the system in actual application are improved. By introducing the demand potential bias index in the defuzzification calculation, the integral weighted processing of high demand potential areas is strengthened. The present invention achieves better results in both recognition accuracy and applicability. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0022] Figure 1 This is an overall flow chart of a method for evaluating power customer demand based on fuzzy evaluation provided in the first embodiment of the present invention. DETAILED DESCRIPTION
[0023] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.
[0024] Example 1, with reference to Figure 1 , as an embodiment of the present invention, provides a method for evaluating power customer demand based on fuzzy evaluation, comprising:
[0025] S1: Receive high-order feature vectors from the deep learning feature extraction layer and divide them into multiple evaluation index dimensions based on load trend characteristics, demand response sensitivity characteristics, customer behavior pattern characteristics, and environmental impact factor characteristics.
[0026] Furthermore, the evaluation index dimensions include electricity consumption behavior feature dimensions, environment-related feature dimensions, economic policy feature dimensions and other auxiliary feature dimensions. Each dimension sets fuzzy processing parameters separately according to the feature type to perform membership calculation and interval division.
[0027] The high-order feature vectors received from the deep learning feature extraction layer are divided into multiple independent evaluation indicator dimensions according to business logic. Specifically, they include:
[0028] Electricity consumption behavior characteristic dimensions: covering customer load change trends, peak-to-valley characteristics, daily load curve patterns, etc.
[0029] Environment-related characteristic dimensions: covering external environmental factors such as temperature, humidity, weather conditions, and holiday effects.
[0030] Economic policy characteristic dimension: covers macro-influencing factors such as electricity price adjustments, energy-saving incentives, and policy support information.
[0031] Other auxiliary characteristic dimensions: such as equipment health, distributed energy access level, energy storage device participation and other auxiliary indicators.
[0032] For each evaluation index dimension, the fuzzy processing strategy is independently set according to the feature attributes, numerical range and change characteristics, including different normalization intervals, fuzzy membership function shapes and language label division standards, so as to achieve accurate fuzzy representation of heterogeneous features.
[0033] For each specific feature in each evaluation indicator dimension, a corresponding fuzzy membership function is set. The membership function is used to map the numerical range of each feature into a linguistic description, such as "low", "medium", "high" or "weak", "medium", "strong", and then fuzzify the high-order feature values according to their corresponding membership functions to obtain the membership degrees of several fuzzy subsets.
[0034] The design of the fuzzy membership function is based on a data-driven approach that integrates domain knowledge. Historical data and domain expert experience are used to set the initial fuzzy interval divisions and function shapes. K-Means clustering or GMM (Gaussian Mixture Model) clustering methods are used to extract natural segmentation from feature data. Clustering results are then modified based on power industry thresholds or expert recommendations to form reasonable fuzzy intervals ranging from "low-medium-high" to more levels.
[0035] The fuzzy membership function is preferably modeled using a trapezoidal function, and the function shape is described by first-order or second-order parameters (such as center, width, and slope) to adapt to the transition interval and critical value phenomena that frequently occur in power demand scenarios.
[0036] It should be noted that the fuzzy membership function is constructed based on a combination of normalization, power transformation, and exponential mapping. The fuzzy interval boundaries and mapping sensitivity parameters can be dynamically adjusted and updated based on historical load data and real-time feedback, which can be expressed as:
[0037]
[0038] Among them, x is the feature input, and are the fuzzy interval boundaries, β m ,γ,δ are adjustable parameters that control the transition characteristics and sensitivity of the fuzzy interval. The fuzzy interval boundaries and parameter values are initially set based on the historical load statistical characteristics and can be dynamically modified based on real-time customer response data to ensure that the fuzzy membership function continues to adapt to changes in the actual demand environment.
[0039] S2: The fuzzy membership function of normalized deformation and nonlinear mapping is used to fuzzify the eigenvalues to generate feature membership. According to the feature membership, the fuzzy rule base is used to perform rule matching and reasoning to generate fuzzy reasoning output.
[0040] Furthermore, the dynamic adjustment of the fuzzy interval boundaries is achieved by combining gradient descent based on the loss function with a regularization term. By minimizing the customer demand forecast deviation, the fuzzy membership function shape and interval position are updated.
[0041] In order to adapt to the dynamic changes in power customer load behavior, the present invention continuously optimizes the interval boundaries and shape parameters of the fuzzy membership function through a dynamic update mechanism based on loss function feedback. The specific update formula is as follows:
[0042]
[0043] Among them, E is the loss function defined according to the actual load response effect, α m is the learning rate, λ is the regularization coefficient, is the initial reference center value. The prediction error is minimized by the gradient descent method, and the regularization term is combined to limit parameter drift, ensuring that the membership function structure is stable and continuously optimized.
[0044] It should be noted that the fuzzy rule base is divided into sub-rule sets according to customer categories. Each sub-rule set contains several feature combination patterns and corresponding demand level labels, and supports dynamic addition, deletion and revision of rule content based on feature changes.
[0045] In the design of the fuzzy rule base, independent sub-rule sets are established for different types of customers (such as residential customers, industrial and commercial customers, and industrial load customers). Each sub-rule set corresponds to a specific set of feature combination patterns and demand level mapping relationships.
[0046] The sub-rule set has a flexible addition and deletion mechanism. When the system detects changes in the feature structure (such as the addition of new distributed access features or the emergence of new power consumption patterns), rule entries can be dynamically added, modified or deleted through the configuration management interface.
[0047] At the same time, the rule priority can be dynamically adjusted within the sub-rule set based on the rule matching accuracy to improve the stability and adaptability of the reasoning results.
[0048] S3: Defuzzify the fuzzy reasoning output through weighted integration to obtain a quantitative score or discrete grade result of customer demand potential.
[0049] Furthermore, rule matching and reasoning include multiplication and combination operations on the membership of each feature, and by setting the feature contribution weight and rule activation index, a weighted fusion output of multiple activation rules is achieved.
[0050] For the fuzzy membership of the input features, the comprehensive reasoning output is achieved by adjusting the feature contribution weight and controlling the rule activation index, which can be expressed as:
[0051]
[0052] in: is the membership degree of the jth premise feature of the i-th rule. Contribute weight to features and control the strength of each feature in rule activation. i is the initial weight of the rule. i It is the rule activation enhancement index, which is used to adjust the nonlinear amplitude of the rule output activation.
[0053] Through the above processing, multiple activation rules can be weighted and fused to generate the final comprehensive fuzzy output, thereby improving the discriminability and stability of the reasoning results.
[0054] It should be noted that weighted integral defuzzification uses fuzzy output membership to perform weighted calculation within the set integral interval. The integral weighting factor introduces the demand potential bias index, and the output quantitative score is used to generate the customer demand potential index.
[0055] The integral defuzzification method with biased exponential weighting is used to transform the fuzzy output into a continuous numerical demand potential assessment result, which can be expressed as:
[0056]
[0057] Where: μ out (z) is the membership function output by fuzzy inference. z is a continuously defined variable within the integration interval. a and b are the lower and upper bounds of the demand potential integration interval, respectively. γ is the potential bias index, which is used to enhance the contribution of high-potential regions in the integration process. By introducing bias index control, the weighting of high-demand potential customers is increased during the integration process, improving the system's accuracy and sensitivity in identifying the strong and weak load response potential classification.
[0058] Example 2 is the second embodiment of the present invention, which is different from the previous embodiment in that:
[0059] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0060] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0061] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.
[0062] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using a combination of any of the following technologies known in the art: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0063] Example 3 is the third embodiment of the present invention. This embodiment provides a system for an electricity customer demand assessment method based on fuzzy evaluation, including a high-order feature classification module, a fuzzy rule reasoning module, and a demand potential defuzzification output module.
[0064] The high-order feature classification module is used to receive high-order feature vectors from the deep learning feature extraction layer and divide them into multiple evaluation index dimensions according to load trend characteristics, demand response sensitivity characteristics, customer behavior pattern characteristics and environmental impact factor characteristics.
[0065] The fuzzy rule reasoning module is used to fuzzify the eigenvalues using the fuzzy membership function of normalized deformation and nonlinear mapping to generate feature membership. According to the feature membership, the fuzzy rule base is used to perform rule matching and reasoning to generate fuzzy reasoning output.
[0066] The demand potential defuzzification output module is used to defuzzify the fuzzy inference output through weighted integration to obtain the customer demand potential quantitative score or discrete grade result.
[0067] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for evaluating power customer demand based on fuzzy evaluation, characterized in that: include: Receive high-order feature vectors from the deep learning feature extraction layer and divide them into multiple evaluation indicator dimensions based on load trend characteristics, demand response sensitivity characteristics, customer behavior pattern characteristics, and environmental impact factor characteristics; The fuzzy membership function of normalized deformation and nonlinear mapping is used to fuzzify the eigenvalues to generate characteristic membership. According to the characteristic membership, the fuzzy rule base is used to perform rule matching and reasoning to generate fuzzy reasoning output. The fuzzy inference output is defuzzified through weighted integration to obtain a quantitative score or discrete grade result of customer demand potential.
2. The method for evaluating power customer demand based on fuzzy evaluation according to claim 1, characterized in that: The evaluation index dimensions include electricity consumption behavior feature dimensions, environment-related feature dimensions, economic policy feature dimensions and other auxiliary feature dimensions. Fuzzy processing parameters are set separately for each dimension according to the feature type to perform membership calculation and interval division.
3. The method for evaluating electricity customer demand based on fuzzy evaluation according to claim 2, characterized in that: The fuzzy membership function is constructed based on a combination of normalization, power transformation and exponential mapping. The fuzzy interval boundaries and mapping sensitivity parameters can be dynamically adjusted and updated based on historical load data and real-time feedback.
4. The method for evaluating electricity customer demand based on fuzzy evaluation according to claim 3, characterized in that: The dynamic adjustment of the fuzzy interval boundary is achieved by combining gradient descent based on the loss function with a regularization term, and the fuzzy membership function shape and interval position are updated by minimizing the customer demand prediction deviation.
5. The method for evaluating power customer demand based on fuzzy evaluation according to claim 4, characterized in that: The fuzzy rule base is divided into sub-rule sets according to customer categories. Each sub-rule set contains several feature combination patterns and corresponding demand level labels, and supports dynamic addition, deletion and revision of rule content based on feature changes.
6. The method for evaluating electricity customer demand based on fuzzy evaluation according to claim 5, characterized in that: The rule matching and reasoning includes performing product combination operations on the membership of each feature, and achieving weighted fusion output of multiple activation rules by setting feature contribution weights and rule activation indexes.
7. The method for evaluating power customer demand based on fuzzy evaluation according to claim 6, characterized in that: The weighted integral defuzzification adopts weighted calculation of fuzzy output membership within a set integral interval, the integral weighting factor introduces the demand potential bias index, and the output quantitative score is used to generate the customer demand potential index.
8. A system using the method for evaluating power customer demand based on fuzzy evaluation according to any one of claims 1 to 7, characterized in that: Including high-order feature classification module, fuzzy rule reasoning module, demand potential defuzzification output module; The high-order feature classification module is used to receive high-order feature vectors from the deep learning feature extraction layer and divide them into multiple evaluation index dimensions according to load trend characteristics, demand response sensitivity characteristics, customer behavior pattern characteristics and environmental impact factor characteristics; The fuzzy rule reasoning module is used to perform fuzzy processing on the eigenvalues by using the fuzzy membership function of normalized deformation and nonlinear mapping to generate characteristic membership, and perform rule matching and reasoning using the fuzzy rule base according to the characteristic membership to generate fuzzy reasoning output; The demand potential defuzzification output module is used to defuzzify the fuzzy reasoning output through weighted integration to obtain a quantitative score or discrete grade result of the customer demand potential.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for evaluating electricity customer demand based on fuzzy evaluation according to any one of claims 1 to 7 are implemented.
10. A 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 for evaluating electricity customer demand based on fuzzy evaluation according to any one of claims 1 to 7 are implemented.