High-energy-consumption industrial user load regulation and control value evaluation method and system and medium

Abstract

The invention discloses a high-energy-consumption industrial user load regulation and control value evaluation method and system and a medium, and the method comprises the steps: forming a candidate scheme through a preset regulation and control strategy generation rule based on an operation plan and a constraint condition of a high-energy-consumption industrial user; inputting the candidate scheme into a pre-constructed multi-target similarity evaluation model to obtain a correlation coefficient of the candidate scheme relative to the positive / negative reference sequence, and determining a deviation amount of the candidate scheme relative to the positive / negative reference sequence based on the correlation coefficient; and carrying out risk value assessment based on the deviation of the candidate scheme and the positive / negative reference sequence in combination with a pre-constructed risk sensitive perception model. According to the multi-target similarity evaluation model, effective information can be integrated through correlation coefficients under the scene that information is incomplete and a system is uncertain, and evaluation failure caused by data missing of a traditional model is avoided; and through a perception value function and a decision weight function of the risk-sensitive perception model, subjective perception of a decision maker on regulation and control income and regulation and control loss is accurately described.

Description

Technical Field

[0001] This invention relates to the field of power system control, specifically to a method, system, and medium for evaluating the value of load regulation for high-energy-consuming industrial users. Background Technology

[0002] As the power system transitions to a new type of power system, high-energy-consuming industrial users (such as those in the steel, chemical, and non-ferrous metals industries) are the core load entities of the power system. The controllability of their loads is of great significance for peak shaving and valley filling, renewable energy consumption, and stable system operation. Currently, the power system's load control for high-energy-consuming industrial users has upgraded from "passive peak shaving" to "active optimization." This requires accurately quantifying the value of different control methods (maintenance, rotation, staggered operation, and peak avoidance) to determine the optimal control strategy.

[0003] Current key technologies focus on the execution of load control measures and single-dimensional value assessment: On the one hand, existing control measures cover scenarios such as maintenance (load adjustment during equipment maintenance), shift work (load transfer during periodic shutdowns), staggered work hours (shifting production periods), and peak avoidance (load reduction during peak electricity consumption periods), but lack quantitative basis for the "value" of different measures; on the other hand, traditional load value evaluation methods mostly adopt single-objective assessment (such as only considering the load reduction amount or economic cost), or rely on complete and certain operating data, which is difficult to adapt to the actual scenarios of fragmented production data and uncertain equipment status of high-energy-consuming industrial users (such as load fluctuations caused by sudden failures).

[0004] The existing technology has significant drawbacks: Poor information adaptability: Traditional evaluation models rely on complete indicator data. In scenarios where information on high-energy-consuming industrial users is incomplete (such as missing production data) or the system status is uncertain (such as fluctuations in raw material supply affecting the load), the evaluation results are biased and cannot support reliable decision-making. Lack of risk perception: Ignoring the risk preferences of decision-makers in load regulation (such as a stronger tendency to avoid "regulation leading to production losses" than to pursue "regulation to obtain subsidy benefits"), the evaluation results are out of touch with actual decision-making needs, which can easily lead to great resistance to the implementation of regulation plans; The reference system is too simple: it often uses a single standard sequence (such as "optimal load reduction") as the evaluation benchmark, without considering the positive and negative effects of the indicators (such as "load reduction" can reduce electricity costs, but may also reduce production capacity), and it does not solve the problem of inconsistent dimensions of different indicators (such as economic cost, production efficiency, and grid contribution), resulting in low evaluation accuracy. Incomplete quantitative logic: There is a lack of a unified quantitative framework for different control measures such as maintenance and rotation, making it impossible to compare the value of different measures horizontally, resulting in unreasonable allocation of control resources. Summary of the Invention

[0005] To address the problems of poor information adaptability, lack of risk perception, single reference system, and incomplete quantitative logic in existing technologies, this invention proposes a method for evaluating the value of load regulation for high-energy-consuming industrial users, including: Based on the operation plans and constraints of high-energy-consuming industrial users, candidate solutions are generated by pre-set control strategies. The constraints include: equipment start-up and shutdown constraints, production interruptibility, and load transfer capability constraints. The candidate scheme is input into a pre-built multi-objective similarity evaluation model to obtain the correlation coefficient of the candidate scheme relative to the positive / negative reference sequence, and the deviation of the candidate scheme relative to the positive / negative reference sequence is determined based on the correlation coefficient. Risk value assessment is performed based on the deviation between candidate solutions and positive / negative reference sequences, combined with a pre-built risk sensitivity perception model. The constructed multi-objective similarity evaluation model is built by determining the ideal reference vector based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users, and combining it with similarity measurement methods.

[0006] Optionally, the construction of the multi-objective similarity evaluation model includes: Based on the index types in the load control scenario of high energy-consuming industrial users, positive / negative reference coefficients are determined according to the calculation formulas of each type, and positive / negative reference sequences are constructed based on the positive / negative reference coefficients in chronological order. Using the positive / negative reference sequences as the basis for decision-making, a multi-objective similarity evaluation model is constructed by combining similarity measurement methods.

[0007] Optionally, the step of inputting the candidate solution into a pre-constructed multi-objective similarity evaluation model to obtain the correlation coefficient between the candidate solution and the positive / negative reference sequence, and determining the deviation of the candidate solution relative to the positive / negative reference sequence based on the correlation coefficient, includes: A similarity metric method is used to compare the numerical values ​​of each candidate solution in the multidimensional space with the positive / negative reference sequences in the multi-objective similarity evaluation model dimension by dimension, so as to obtain the correlation coefficient of each candidate solution relative to the positive / negative reference sequences in the multidimensional space. The correlation coefficient of each candidate solution relative to the positive / negative reference sequence in the multidimensional space is converted into a standardized similarity coefficient; The standardized similarity coefficient is used as the deviation of the candidate scheme from the positive / negative reference sequence.

[0008] Optionally, the step of inputting the candidate solution into a pre-constructed multi-objective similarity evaluation model to obtain the correlation coefficient between the candidate solution and the positive / negative reference sequence, and determining the deviation of the candidate solution relative to the positive / negative reference sequence based on the correlation coefficient, further includes: By integrating the standardized similarity coefficients of each candidate solution in each dimension, a comprehensive correlation coefficient of the candidate solution relative to the ideal reference vector is obtained.

[0009] Optionally, the standardized similarity coefficient is calculated using the following formula:

[0010] in, Let i be the correlation coefficient of scheme i under index k. It is the value of scheme i under index k. It is the value of the ideal reference vector under index k. It is the resolution coefficient.

[0011] Optionally, the comprehensive correlation coefficient of the candidate solution relative to the ideal reference vector is calculated using the following formula:

[0012] Where n is the total number of indicators, It is the correlation coefficient of scheme i under index k. To achieve a comprehensive correlation coefficient, Let i be the number of indicators and i be the number of enterprises.

[0013] Optionally, the risk value assessment based on the deviation between the candidate scheme and the positive / negative reference sequence combined with a pre-built risk sensitivity perception model includes: By substituting the deviation of the candidate scheme from the positive / negative reference sequence into the positive / negative risk value function in the pre-constructed risk sensitivity perception model, the positive / negative risk value of the indicator relative to high energy-consuming industrial users under the maintenance and control value is obtained. The decision weights when facing gains / losses are determined by combining the weights of indicators under the value of maintenance and control with the subjective weight function of a pre-built risk sensitivity perception model. Based on the positive / negative risk value combined with the decision weight when facing gains / losses, the maintenance and control value of the candidate solutions is determined. Risk value assessment is conducted based on the maintenance and control value of the candidate solutions.

[0014] Optionally, the positive / negative risk value function is as follows:

[0015] In the formula, For the positive risk value function of indicator j relative to high energy-consuming industrial user i under the maintenance and control value, Let index j be the negative risk value function relative to high-energy-consuming industrial user i under the maintenance and control value. The negative reference coefficient for the j-th indicator of high-energy-consuming industrial user i under the maintenance and control value. The positive reference coefficient for the j-th indicator of high-energy-consuming industrial user i under the maintenance and control value. and These represent the risk preference coefficient and the risk aversion coefficient, respectively; i represents the high-energy-consuming industrial user serial number; and j represents the indicator serial number. This represents the sensitivity coefficient of decision-makers to gains and losses.

[0016] Optionally, the subjective weight function is shown in the following equation:

[0017] In the formula, The decision weight for the benefit, As the decision weight when losses occur, The weight of the j-th indicator under the maintenance and control value; and These are the risk attitude coefficients of decision-makers when facing "gains" and "losses," respectively. , It characterizes the sensitivity of decision-makers to residual weights under gain and loss scenarios.

[0018] Optionally, the maintenance and control value is calculated using the following formula:

[0019] In the formula, For the maintenance and control value of high-energy-consuming industrial users i, For the positive risk value function of indicator j relative to high energy-consuming industrial user i under the maintenance and control value, Let index j be the negative risk value function relative to high-energy-consuming industrial user i under the maintenance and control value. As a decision weight when faced with benefits, As the decision weight when facing losses.

[0020] Furthermore, this invention also provides a load control value assessment system for high-energy-consuming industrial users, comprising: The scheme generation module is used to generate candidate schemes based on the operation plans and constraints of high-energy-consuming industrial users, and generate rules through pre-set control strategies. The constraints include: equipment start-up and shutdown constraints, production interruptibility, and load transfer capability constraints. The similarity assessment module is used to input the candidate scheme into a pre-built multi-objective similarity assessment model to obtain the correlation coefficient of the candidate scheme relative to the positive / negative reference sequence, and to determine the deviation of the candidate scheme relative to the positive / negative reference sequence based on the correlation coefficient. The value assessment module is used to perform risk value assessment based on the deviation of candidate solutions from positive / negative reference sequences, combined with a pre-built risk sensitivity perception model. The constructed multi-objective similarity evaluation model is built by determining the ideal reference vector based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users, and combining it with similarity measurement methods.

[0021] Optionally, the similarity assessment module includes: The calculation submodule is used to compare the numerical value of each candidate solution in the multidimensional space with the positive / negative reference sequence in the multi-objective similarity evaluation model dimension by dimension using a similarity measurement method, so as to obtain the deviation of each candidate solution from the positive / negative reference sequence in the multidimensional space. The standardization submodule is used to convert the deviation of each candidate solution from the positive / negative reference sequence in multidimensional space into a standardized correlation coefficient.

[0022] Optionally, the value assessment module is specifically used for: By substituting the deviation of the candidate scheme from the positive / negative reference sequence into the positive / negative risk value function in the pre-constructed risk sensitivity perception model, the positive / negative risk value of the indicator relative to high energy-consuming industrial users under the maintenance and control value is obtained. The decision weights when facing gains / losses are determined by combining the weights of indicators under the value of maintenance and control with the subjective weight function of a pre-built risk sensitivity perception model. Based on the positive / negative risk value combined with the decision weight when facing gains / losses, the maintenance and control value of the candidate solutions is determined. Risk value assessment is conducted based on the maintenance and control value of the candidate solutions.

[0023] Optionally, the subjective weight function is shown in the following equation:

[0024] In the formula, The decision weight for the benefit, As the decision weight when losses occur, The weight of the j-th indicator under the maintenance and control value; and These are the risk attitude coefficients of decision-makers towards gains and losses, respectively. , It characterizes the sensitivity of decision-makers to residual weights under gain and loss scenarios.

[0025] In another aspect, this application also provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a method for evaluating the load regulation value of high-energy-consuming industrial users as described above is implemented.

[0026] Furthermore, this application also provides a readable storage medium on which an executable program is stored, which, when executed, implements the above-described method for evaluating the load regulation value of high-energy-consuming industrial users.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method for evaluating the value of load regulation for high-energy-consuming industrial users, comprising: generating candidate schemes based on the operation plan and constraints of high-energy-consuming industrial users through pre-set regulation strategy generation rules, wherein the constraints include: equipment start-up and shutdown constraints, production interruptibility constraints, and load transfer capability constraints; inputting the candidate schemes into a pre-constructed multi-objective similarity evaluation model to obtain the correlation coefficients of the candidate schemes relative to positive / negative reference sequences, and determining the deviation of the candidate schemes relative to the positive / negative reference sequences based on the correlation coefficients; and conducting risk value assessment based on the deviation of the candidate schemes from the positive / negative reference sequences combined with a pre-constructed risk sensitivity perception model; wherein the constructed multi-objective similarity evaluation model is constructed by determining an ideal reference vector based on the target attributes and industry physical laws under the load regulation scenario of high-energy-consuming industrial users, combined with a similarity measurement method. The multi-objective similarity assessment model of this invention can integrate effective information through similarity coefficients in scenarios with incomplete information (such as missing some production indicators) and system uncertainty (such as equipment status fluctuations), avoiding the evaluation failure caused by data deficiency in traditional models. The risk-sensitive perception model of this invention, with its perception value function and decision weight function, accurately portrays the decision-maker's subjective perception of "control benefits" and "control losses". The combination of the multi-objective similarity assessment model and the risk-sensitive perception model increases universality and improves the quantitative logic. Attached Figure Description

[0028] Figure 1 This is a flowchart of a method for evaluating the value of load regulation for high-energy-consuming industrial users according to the present invention; Figure 2 This is a schematic diagram of the risk value function in the risk sensitivity perception model of the present invention; Figure 3 This is a schematic diagram of an electronic device structure according to the present invention. Detailed Implementation

[0029] This invention proposes a method for evaluating the controllable value of loads of high-energy-consuming industrial users that integrates multi-objective similarity assessment and risk sensitivity perception. This method can be used to support the power system in making optimal decisions on control measures such as maintenance, rotation, staggered operation, and peak avoidance for high-energy-consuming industrial users.

[0030] To better understand the present invention, the following description, in conjunction with the accompanying drawings and embodiments, will further illustrate the content of the present invention.

[0031] Example 1: A method for evaluating the value of load regulation for high-energy-consuming industrial users, such as Figure 1 As shown, it includes: Step 1: Based on the operation plan and constraints of high-energy-consuming industrial users, generate rules to form candidate solutions through pre-set control strategies. The constraints include: equipment start-up and shutdown constraints, production interruptibility, and load transfer capability constraints. Step 2: Input the candidate solution into the pre-built multi-objective similarity evaluation model to obtain the correlation coefficient of the candidate solution relative to the positive / negative reference sequence, and determine the deviation of the candidate solution relative to the positive / negative reference sequence based on the correlation coefficient; Step 3: Conduct a risk value assessment based on the deviation between the candidate solution and the positive / negative reference sequence, combined with a pre-built risk sensitivity perception model; The constructed multi-objective similarity evaluation model is built by determining the ideal reference vector based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users, and combining it with similarity measurement methods.

[0032] The present invention will be further described below: After constructing a load controllability value index system for high-energy-consuming industrial users, it is necessary to quantify the controllability value of high-energy-consuming industrial users, such as maintenance, rotation, staggered operation, and peak avoidance, in order to determine appropriate control measures. This invention proposes a load controllability value quantification method (MOSE-RSPM) that integrates a multi-objective similarity assessment model and a risk sensitivity perception mechanism.

[0033] Step 1: Based on the operation plans and constraints of high-energy-consuming industrial users, candidate solutions are generated through pre-set control strategies and rules. Specifically, this includes: In this invention, the control candidate scheme refers to a set of feasible load control methods generated based on the production conditions, equipment characteristics, and control requirements of high-energy-consuming industrial users. Candidate schemes can be obtained directly from external sources, or they can be formed based on information such as the high-energy-consuming industrial user's operation plan, equipment start-up and shutdown constraints, production interruptibility, and load transfer capability, through pre-set control strategy generation rules. In this embodiment, the candidate scheme is formed based on information such as the high-energy-consuming industrial user's operation plan, equipment start-up and shutdown constraints, production interruptibility, and load transfer capability, through pre-set control strategy generation rules. For example, for typical users in industries such as steel, electrolytic aluminum, and cement, the candidate solutions may include: (1) Maintenance solution A: Arrange a 2-hour shutdown for maintenance during non-critical production periods, which is expected to reduce the load by 5 MW; (2) Shifting the time solution B: Move the high-load process from the peak period (14:00–17:00) to the valley period (22:00–01:00), which reduces the load by 3 MW; (3) Peak avoidance solution C: Reduce the load by 4 MW after the power grid issues the peak shaving instruction, which lasts for 1 hour; (4) Rotation solution D: Rotate some production lines for half a day to achieve a 2 MW load transfer.

[0034] Therefore, the "candidate schemes" in this application are all multiple optional control modes constructed to meet the load control needs of high-energy-consuming industrial users, and are used as evaluation inputs for the MOSE model and RSPM model.

[0035] Before step 2, we will further introduce the construction of the multi-objective similarity evaluation model: S1 Multi-objective Similarity Evaluation Model Construction Multi-objective similarity evaluation is a decision analysis method suitable for environments with incomplete information and system uncertainty. It is widely used in multi-objective decision-making, comprehensive evaluation, and optimization problems. The core concept of this model is to decompose complex problems into several observable index dimensions. Given limited known data, it judges the merits of each candidate solution by evaluating the degree of similarity between each candidate solution and an idealized reference state.

[0036] The basic idea of ​​a multi-objective similarity evaluation model is to define an optimal reference state across all evaluation dimensions, called the ideal reference vector, and use this as the benchmark for decision-making. All alternative solutions can be considered as the "proximity" to this ideal state, and this proximity can be modeled using similarity measurement methods in multi-dimensional space.

[0037] Ideal reference vector: This refers to the set of optimal values ​​pursued in each indicator dimension, representing the optimal level that the decision-making objective hopes to achieve. It serves as the central coordinate for evaluation in the model, used to measure the performance of each scheme on different indicators. The ideal reference vector in this invention is not arbitrarily set subjectively, but determined based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users. The ideal reference vector represents the optimal operating state achievable in each evaluation indicator dimension, and its construction follows these principles: (1) Based on physical upper bounds or optimal technical capabilities. For benefit-type indicators (such as load reduction, grid contribution, control response speed, etc.), the ideal reference value is the optimal level of the indicator within the physical or technically achievable range, such as the maximum load that the equipment can reduce, the fastest response time of the user's control equipment, etc.

[0038] (2) Setting based on industry benchmark data. For different industrial sectors (steel, cement, electrolytic aluminum, etc.), ideal values ​​for this dimension can be constructed based on industry typical values, historical best operating conditions, or national standards (such as unit energy consumption, equipment rated load utilization rate) to make the reference vector industry-adaptable.

[0039] (3) Setting based on control objectives. For grid dispatch-type control (such as peak shaving and time shifting), the ideal reference can be set according to the grid side's requirements for peak shaving amplitude, peak shaving duration, energy saving benefits, etc., to make it more in line with the grid objectives. Therefore, the ideal reference vector is not simply set by humans, but is a multi-dimensional optimal vector formed based on industry rules, equipment capacity constraints, user controllability and control objectives, used to quantify "the degree of deviation of candidate schemes from the achievable optimal state".

[0040] Multidimensional deviation (similarity distance): The value of each candidate solution under each evaluation index is compared with the ideal reference vector dimension by dimension, forming its "deviation trajectory" in multidimensional space. These deviations are further converted into standardized similarity coefficients to reflect the degree of closeness between the solution and the ideal state.

[0041] Similarity tolerance range (similarity radius): The model also introduces the concept of "acceptable deviation," representing the maximum tolerable offset of the solution in each dimension. This tolerance range reflects the inclusiveness of the solution's deviation from the ideal target under realistic conditions. A smaller deviation radius means that the solution is closer to the ideal state, while a larger radius indicates that the solution has a large deviation in multiple dimensions and poor overall performance.

[0042] Finally, the model integrates and analyzes the similarity coefficients of each solution across different indicator dimensions to determine the overall degree of closeness of each solution to the ideal state. This model effectively supports decision-making and provides quantitative evidence for optimal selection, especially in situations with numerous evaluation indicators and uncertain or partially missing data.

[0043] Correlation coefficient: (1) in, Let i be the correlation coefficient of scheme i under index k. It is the value of scheme i under index k. It is the value of the ideal reference vector under index k. It is the resolution coefficient, used to adjust the sensitivity to differences, and is usually set to 0.5.

[0044] Overall correlation coefficient: (2) Where n is the total number of indicators, It is the correlation coefficient of scheme i under index k. To achieve a comprehensive correlation coefficient, Let i be the number of indicators and i be the number of enterprises.

[0045] Step 2: Input the candidate solution into a pre-built multi-objective similarity evaluation model to obtain the correlation coefficient of the candidate solution relative to the positive / negative reference sequence, and determine the deviation of the candidate solution relative to the positive / negative reference sequence based on the correlation coefficient, including: A similarity metric method is used to compare the numerical values ​​of each candidate solution in the multidimensional space with the positive / negative reference sequences in the multi-objective similarity evaluation model dimension by dimension, so as to obtain the correlation coefficient of each candidate solution relative to the positive / negative reference sequences in the multidimensional space. The correlation coefficient of each candidate solution relative to the positive / negative reference sequence in the multidimensional space is converted into a standardized similarity coefficient; The standardized similarity coefficient is used as the deviation of the candidate scheme from the positive / negative reference sequence.

[0046] Furthermore, the standardized similarity coefficient is calculated using the following formula:

[0047] in, Let i be the correlation coefficient of scheme i under index k. It is the value of scheme i under index k. It is the value of the ideal reference vector under index k. It is the resolution coefficient.

[0048] The candidate scheme is input into a pre-built multi-objective similarity evaluation model to obtain the deviation of the candidate scheme from the positive / negative reference sequence, and the method further includes: By integrating the standardized similarity coefficients of each candidate solution in each dimension, a comprehensive correlation coefficient of the candidate solution relative to the ideal reference vector is obtained.

[0049] Furthermore, the comprehensive correlation coefficient of the candidate solution relative to the ideal reference vector is calculated using the following formula:

[0050] Where n is the total number of indicators, It is the correlation coefficient of scheme i under index k. To achieve a comprehensive correlation coefficient, Let i be the number of indicators and i be the number of enterprises.

[0051] Step 3 also includes the construction of a risk sensitivity perception model.

[0052] S2 Risk Sensitivity Perception Model Construction.

[0053] The Risk-Sensitive Perception Model (RSP) is a theoretical tool used to characterize individuals' decision-making preferences in uncertain environments. It evolved from the psychological mechanisms of decision-making in behavioral economics. This model breaks away from the assumption of "maximizing expected utility" in traditional rational decision-making models, emphasizing the differences in subjective perceptions when individuals weigh gains and losses, and particularly highlighting the behavioral characteristics of risk aversion and loss aversion.

[0054] Unlike the absolute value maximization in the classic model, individuals often make relative judgments around a certain psychological benchmark during the decision-making process, and make action choices based on the perceived "good" or "bad" or "gain or loss" of the outcome based on that benchmark.

[0055] Reference Benchmark: The decision-making process in the risk sensitivity perception model revolves around a reference benchmark. This benchmark may originate from an individual's current state, historical experience, or expected goals. Decision-makers do not simply assess the absolute value of an outcome but focus on its change relative to this benchmark. The same outcome may be perceived as a gain or a loss under different reference conditions. This reference-based perception mechanism is one of the core logics of the model. In this application, the reference benchmark is not independently set but is directly provided by the calculation results of the MOSE model, enabling the risk sensitivity perception model to construct corresponding positive and negative risk value functions based on the offset of candidate solutions relative to positive / negative reference sequences, thereby achieving quantitative modeling of perceived gains and perceived losses.

[0056] Loss aversion bias: Another key assumption of the model is that loss sensitivity is greater than gain sensitivity. In other words, individuals typically react more strongly to potential losses than to equivalent gains. This "negative sensitivity bias" means that the dissatisfaction arising from losing a certain value far outweighs the satisfaction derived from gaining the same value. This mechanism explains the widespread phenomenon in reality of "seeking benefits and avoiding harm," but with a greater tendency towards "harm avoidance."

[0057] Under risk conditions, this loss-avoidance behavior causes individuals to tend to avoid options that may result in losses when faced with choices of the same expected value, even if this sacrifices the opportunity to gain more benefits.

[0058] Perceived Value Function: In the risk-sensitive perception model, the function used to describe the decision-maker's subjective reaction is called the perceived value function, which has an asymmetric concave-convex structure. Within the payoff range, the perceived value function exhibits concavity, reflecting the phenomenon of "diminishing marginal satisfaction," meaning that the sensitivity to additional payoffs decreases as the payoff increases, demonstrating a more conservative risk attitude.

[0059] Within the loss interval, the perceived value function exhibits convexity, indicating that as the loss increases, the individual's sense of distress grows more rapidly, and they are more prone to aggressive behaviors, such as taking risks, to avoid further losses.

[0060] This perceptual structure of "stable returns and risky losses" vividly reflects the psychological dynamics of people when facing risks and uncertain outcomes, which is different from the linear utility structure in traditional rational assumptions.

[0061] Value function: (3) Where x is the offset of the candidate control scheme relative to the reference point (positive / negative reference sequence) output by the MOSE model on a certain evaluation index. and A value less than 1 typically reflects the diminishing returns effect. >1 indicates the degree of loss aversion among people. For value function, , This is the decreasing benefit coefficient.

[0062] Decision weight function: Prospect theory also introduces a decision weighting function to describe how people perceive the probability of an event occurring. People tend to overestimate low-probability events and underestimate high-probability events.

[0063] (4) Where p is the probability of the event occurring. This reflects the decision-makers' distorted perception of probability. Let be the decision weight function. To help decision-makers perceive the probability.

[0064] S3 Risk Sensitivity Perception Model Decision Process The risk sensitivity perception model consists of two steps: (1) Editing stage: Individuals reconstruct the decision problem based on the reference point and determine the gains and losses.

[0065] (2) Evaluation stage: Individuals assess different risks through value functions and decision weight functions, and then make decisions.

[0066] In the risk sensitivity perception model of this application, "different risks" refers to various uncertainties that high-energy-consuming industrial users may face during the implementation of load regulation. Typical risks include, but are not limited to, production loss risk, equipment condition risk, grid-side revenue uncertainty risk, energy price fluctuation risk, load response deviation risk, and other uncertainties.

[0067] Step 3: Conduct a risk value assessment based on the deviation between the candidate solution and the positive / negative reference sequence, combined with a pre-built risk sensitivity perception model, including: By substituting the deviation of the candidate scheme from the positive / negative reference sequence into the positive / negative risk value function in the pre-constructed risk sensitivity perception model, the positive / negative risk value of the indicator relative to high energy-consuming industrial users under the maintenance and control value is obtained. The decision weights when facing gains / losses are determined by combining the weights of indicators under the value of maintenance and control with the subjective weight function of a pre-built risk sensitivity perception model. Based on the positive / negative risk value combined with the decision weight when facing gains / losses, the maintenance and control value of the candidate solutions is determined. Risk value assessment is conducted based on the maintenance and control value of the candidate solutions.

[0068] The maintenance and control value is a comprehensive risk value calculated based on a risk sensitivity perception model. Its value directly reflects the comprehensive value level of the candidate control scheme after considering the asymmetric perception of benefits and losses.

[0069] In this application, different risks are not evaluated using completely independent strategies, but rather through a unified evaluation achieved via the value function and decision weight function in the risk sensitivity perception model. Therefore, this invention does not design different models for each type of risk, but rather distinguishes them by using a "unified framework + different risks falling into different quadrants (gain zone or loss zone)," so that various risks are automatically mapped to the corresponding regions of the value function, achieving a differentiated but unified risk response mechanism.

[0070] Example 2 Since the controllable value indicators are independent of each other, the following will take the maintenance control value as an example to introduce the quantification method of load controllable value.

[0071] Construct the positive (negative) reference coefficient matrix: The MOSE model generates a positive (negative) reference coefficient matrix based on a unified measure transformation, the construction of positive / negative reference sequences, and the deviation analysis between candidate solutions and the reference sequences. This matrix reflects how close the candidate solutions are to the ideal or worst-case scenario in each indicator. Subsequently, the risk sensitivity perception model uses this matrix as a benchmark to further conduct a risk value assessment.

[0072] The basic idea of ​​a multi-objective similarity evaluation model is to define an optimal reference state across all evaluation dimensions, called the ideal reference vector, and use this as the benchmark for decision-making. All alternative solutions can be considered as the "proximity" to this ideal state, and this proximity can be modeled using similarity measurement methods in multi-dimensional space.

[0073] Now assume the initial evaluation matrix under the maintenance and control value is as follows: , , Let n represent the number of high-energy-consuming industrial users, m represent the number of indicators under maintenance and control value, N=3, M=5. Traditional multi-objective similarity assessment models typically construct standard sequences based on a single pattern. To improve this, the concepts of "positive reference coefficient" and "negative reference coefficient" are proposed. The construction methods for standard sequences vary depending on the indicator type: 1) Benefit-oriented indicators.

[0074] (5) 2) Cost-related indicators.

[0075] (6) In the formula: and Let represent the positive and negative reference coefficients of the j-th indicator under the maintenance and control value, respectively. Let j be the j-th indicator of company i. By calculating various indicators, a standard sequence and a positive reference sequence can be obtained. and negative reference sequence .

[0076] The dimensions of the various indicators under the maintenance and control value are different, and the numerical values ​​of the indicators vary considerably. Therefore, it is necessary to perform a unified measure transformation on the evaluation matrix of maintenance and control value, i.e., a multi-objective similarity transformation. Here, the evaluation matrix of maintenance and control value is the initial evaluation matrix under the maintenance and control scenario, which is constructed from the multi-dimensional operating data of high-energy-consuming industrial users under maintenance and control conditions. After the unified measure transformation, it serves as the input matrix of the multi-objective similarity assessment model. Let... , Then there is The formula for multi-target similarity transformation is as follows: (7) In the formula, This is a multi-target similarity transformation operation. Let be the coefficient of the j-th indicator of value i. Let be the positive reference coefficient for the j-th indicator under the maintenance and control value. Let i be the transformed sequence of the j-th indicator of value i, where i is the value index and j is the indicator index. When i = 0, the standard positive reference sequence can be obtained respectively. and standard negative reference sequence , where T represents the multi-target similarity transformation operation formula.

[0077] By constructing a standard pattern sequence for maintenance and control value and performing a unified measurement transformation on the maintenance evaluation matrix, which is built from multi-dimensional operational data of high-energy-consuming industrial users under maintenance and control scenarios, this data can be obtained through on-site monitoring, control records, enterprise production management systems, or estimation based on industry experience. Each row of the matrix corresponds to the control performance of a user or candidate solution, and each column corresponds to an evaluation index, which is used as input to the MOSE model for subsequent reference coefficient calculation. The calculation formulas for positive and negative reference coefficients can be derived as follows: Record the positive reference coefficient of the j-th indicator for high energy-consuming industrial user i under the maintenance and control value. for ,but: (8) In the formula: This represents the difference in association between the j-th indicator of high-energy-consuming industrial user i and the positive reference coefficient. ; The resolution coefficient is usually set to 0.5. This represents the j-th index in the standard positive reference sequence.

[0078] Similarly, the negative reference coefficient of the j-th indicator for high-energy-consuming industrial user i under the maintenance and control value can be obtained. , recorded as : (9) In the formula: This represents the difference in association between the j-th indicator of high-energy-consuming industrial user i and the negative reference. , Let j represent the j-th index in the standard negative reference sequence.

[0079] By calculating the reference coefficients of each high-energy-consuming industrial user and the standard positive (negative) reference coefficient sequence under the maintenance and control value, and This allows for the establishment of a corresponding positive (negative) reference coefficient matrix.

[0080] Quantifying the adjustability value of load: The risk sensitivity perception model (RSPM) is a theoretical framework for characterizing individuals' stochastic decision-making behavior in uncertain environments. Unlike traditional perfectly rational optimal choice models, RSPM argues that real-world decision-makers typically exhibit bounded rationality, and their choices in the face of risk and uncertainty are often influenced by a combination of factors, including psychological expectations, cognitive biases, and subjective perceptions.

[0081] The basic premise of this model is that individuals evaluate decision outcomes not based on absolute numerical values, but rather on a psychological reference point. This reference point may come from the current state, historical experience, or expected goals. Based on this, decision-makers will divide outcomes into "perceived gains" and "perceived losses," and form an asymmetric risk preference response accordingly.

[0082] In the modeling structure of the risk sensitivity perception model, the perceived value function and the subjective weight function jointly determine an individual's psychological evaluation of different options: the perceived value function describes the intensity of an individual's subjective reaction to gains or losses when the outcome deviates from the reference point. Specifically, different options refer to different control candidate schemes. Suppose a steel company needs to participate in load control during the peak period of the power grid from 14:00 to 17:00. Based on its process structure and equipment characteristics, the following candidate schemes can be formed: Scheme A (Maintenance Scheme): Planned maintenance of the rolling mill from 14:00 to 16:00 can reduce the load by 5 MW. Scheme B (Shifted Operating Hours Scheme): Shifting the operating period of the high-energy-consuming heating furnace from the peak period to 21:00 to 24:00 can reduce the peak period load by 3 MW. Scheme C (Peak Avoidance Scheme): Reducing some fan load during the peak period, reducing 4 MW, lasting for 1 hour. These are multiple executable control methods generated based on enterprise production conditions, equipment status, process interruptibility, and power grid control requirements. Each scheme represents a control strategy that the user can adopt within a specific time period. The subjective weighting function is used to characterize how individuals perceive and weight uncertainty probabilities; that is, under the same objective probability, the importance of different events in their minds may differ significantly. In this application, "event" refers to the uncertain gains or losses that may occur during the implementation of candidate control schemes, and their probabilities, such as whether load reduction targets are met, whether capacity losses exceed expectations, and whether electricity price compensation is obtained as scheduled. The subjective weighting function is used to characterize the decision-maker's psychological perception bias regarding the probability of these uncertain events and to weight and integrate risk value.

[0083] In a gain-oriented situation, decision-makers generally exhibit a risk-averse tendency. They prefer a more conservative option, even if the return is lower, rather than taking greater risks for higher returns. At this stage, the perceived value function shows a concave structure, reflecting the psychological mechanism that the marginal utility of gains gradually decreases with increasing returns. In a loss-oriented situation, however, individual behavior patterns typically reverse, exhibiting a risk-seeking tendency. At this point, although taking risks may lead to greater losses, individuals are more likely to adopt aggressive strategies to avoid or mitigate existing losses. The perceived value function then shows a convex structure, indicating that as the magnitude of the loss increases, the individual's psychological reaction to each additional unit of loss becomes more intense, i.e., the "marginal anguish" rises rapidly.

[0084] This asymmetry in the perception mechanism is one of the key features of the risk sensitivity perception model. It reveals that an individual's attitude towards risk is highly dependent on their perceived position: in the gain range, they tend to avoid risk, while in the loss range, they may actively pursue risk. This inconsistency in behavior is an important phenomenon that traditional rational models struggle to explain.

[0085] By jointly constructing a nonlinear perception function and a probability weighting mechanism, RSPM can effectively simulate the psychological dynamics and behavioral outcomes of humans in complex decision-making scenarios in reality. The risk value function of the risk-sensitive perception model is as follows: Figure 2 As shown.

[0086] Using the positive (negative) reference sequences obtained from the multi-objective similarity assessment model as reference values, a positive (negative) risk value function for index j relative to high-energy-consuming industrial user i under the maintenance and control value is constructed: (10) In the formula: and These represent the risk preference coefficient and the risk aversion coefficient, respectively, which are the concavity and convexity of the value function; This represents the sensitivity coefficient of decision-makers to gains and losses. According to the risk sensitivity perception model, the decision-making weights for decision-makers when facing gains and losses are as follows: (11) In the formula: The decision weight for the benefit, As the decision weight when losses occur, The weight of the j-th indicator under the maintenance and control value; and These are the risk attitude coefficients of decision-makers when facing "gains" and "losses," respectively. , The sensitivity of decision-makers to residual weights is characterized under gain and loss scenarios, respectively, to reflect the asymmetric distortion of the decision weight function in the high-weight range. If the comprehensive positive (negative) risk value function of maintenance is defined as the maintenance control value of high-energy-consuming industrial users, then the maintenance control value of high-energy-consuming industrial user i is... It is the sum of the positive risk value and the negative risk value, that is: (12) The maintenance and control value is a comprehensive risk value calculated based on a risk sensitivity perception model. Its value directly reflects the comprehensive value level of the candidate control scheme after considering the asymmetric perception of benefits and losses.

[0087] Specifically, the value of maintenance and control is used to compare and rank different candidate control schemes: under the same control scenario, the larger the value of maintenance and control, the higher the overall value of the control scheme after comprehensively considering the grid-side benefits, user-side losses and decision-makers' risk preferences, and the higher its priority.

[0088] In practical applications, based on the power grid operation requirements or control resource constraints, corresponding value thresholds can be set or a value ranking method can be used to select the top few maintenance and control schemes as the preferred execution schemes, thereby maximizing the value of load control under controllable risk conditions.

[0089] This invention, through a fusion of a "Multi-Objective Similarity Assessment Model (MOSE) + Risk Sensitivity Perception Model (RSPM)," achieves the following significant effects in evaluating the adjustable value of loads for high-energy-consuming industrial users, and demonstrates clear advantages over existing technologies: Enhancing information adaptability and making quantification results more reliable: The MOSE model, through its design of "ideal reference vector + multi-dimensional deviation + similarity tolerance range," can integrate effective information through similarity coefficients in scenarios with incomplete information (such as missing production indicators) and system uncertainty (such as equipment status fluctuations), avoiding the evaluation failure caused by data gaps in traditional models. In practical applications, even if the data gap rate of core indicators (such as unit time production capacity) reaches 20%, the error in quantifying the regulatory value can still be controlled within 5% through similarity analysis of other related indicators (such as raw material consumption and equipment operating time), which is far superior to the error rate of over 15% of traditional models. Aligned with practical decision-making, control measures are easier to implement: The RSPM model introduces a "reference benchmark + loss aversion preference," accurately depicting decision-makers' subjective perceptions of "control benefits" and "control losses" through a perceived value function (concaveness of the benefit interval and convexity of the loss interval) and a decision weight function (overestimating low-probability risks and underestimating high-probability benefits). For example, in the evaluation of maintenance and control, the asymmetric impact of "grid peak-shaving benefits from equipment maintenance" and "production losses caused by shutdowns" can be quantified, making the evaluation results more consistent with actual decision-making psychology, and increasing the acceptance rate of control measures by more than 40%. Unified evaluation standards enable more accurate horizontal comparisons: By constructing a "positive / negative reference coefficient matrix" (distinguishing between benefit-type indicators such as grid contribution and cost-type indicators such as production loss) and using unified measurement transformation, the problem of inconsistent dimensions for different indicators is solved, while simultaneously achieving a comprehensive consideration of both "positive effects (benefits)" and "negative effects (losses)". For example, when comparing the value of maintenance and peak shaving regulation, "grid peak shaving subsidies (ten thousand yuan)" and "production losses (tons / hour)" can be converted into dimensionless coefficients through unified measurement transformation, reducing the horizontal comparison error to within 3%, while traditional models, due to the lack of unified dimensions, often have comparison errors exceeding 20%. Covering multiple control scenarios and providing more comprehensive decision support: Taking maintenance control as an example, this invention provides a unified quantitative framework that can be extended to scenarios such as rotational control, time-shifting, and peak-shaving (by replacing the positive / negative reference coefficients of the corresponding indicators). For example, rotational control can use "load transfer duration" as the core indicator, and time-shifting control can use "improvement in grid load rate after time shifting" as the core indicator. Both are calculated using MOSE-RSPM fusion logic to realize the value ranking of control measures in multiple scenarios. This provides the power system with a basis for "allocating control resources according to value priority," increasing the overall grid benefit of load control by 15%-25%, while reducing the average control loss of high-energy-consuming users by 10%-18%.

[0090] This invention addresses the shortcomings of existing technologies for evaluating the adjustable load value of high-energy-consuming industrial users, aiming to solve the following technical problems and ultimately achieve accurate value quantification that aligns with actual decision-making needs: To address the problem of large evaluation bias in traditional models under scenarios with incomplete information and system uncertainty, a value quantification method that can adapt to the data fragmentation and state fluctuations of high-energy-consuming industrial users is provided. To address the problem that traditional methods ignore decision-makers' risk sensitivity perception, this approach ensures that value assessment results match the risk preferences (such as loss aversion tendencies) in actual decision-making, thereby improving the feasibility of control measures. To address the issues of a single evaluation reference system and inconsistent indicator dimensions in the existing system, the accuracy and horizontal comparability of value evaluation of different regulatory measures can be improved through multi-dimensional reference and unified measurement transformation. A unified quantitative framework covering scenarios such as maintenance, shift work, staggered operation, and peak avoidance will be constructed to achieve accurate calculation and comparison of the value of different control measures, providing a basis for the allocation of power system load control resources.

[0091] The core of this invention lies in constructing a fusion quantitative logic of "multi-target similarity assessment - risk sensitivity perception", as detailed below: The core construction logic of the MOSE model is: based on the "ideal reference vector", it integrates multi-dimensional indicators in scenarios with incomplete information and uncertain systems through "multi-dimensional deviation calculation - similarity coefficient integration - positive / negative reference coefficient matrix generation", thus solving the limitations of traditional single-objective evaluation. The risk characterization logic of the RSPM model is: through “reference benchmark setting - perceived value function - decision weight function”, it accurately simulates the decision-maker’s asymmetric perception of gains and losses, especially the risk attitude under loss aversion preference, so that the evaluation results match the actual decision-making psychology. The logic of integrating MOSE and RSPM is as follows: the "positive / negative reference sequence" output by the MOSE model is used as the "reference benchmark" of the RSPM model. Positive / negative risk value functions are constructed for each regulatory indicator. Through decision weight integration, the "positive risk value + negative risk value" is finally used as the quantitative result of regulatory value, forming a complete link of "multi-dimensional information integration - risk perception correction - comprehensive value output". Multi-scenario adaptation logic: Taking maintenance and control as the basic example, by replacing the indicator dimensions of "positive / negative reference coefficients" (such as replacing rotation control with "load transfer duration" and "time period matching degree", and replacing peak avoidance control with "peak load reduction amount" and "valley load compensation amount"), full coverage of scenarios such as rotation, staggered operation, and peak avoidance is achieved, forming a unified quantitative framework.

[0092] A method for evaluating the adjustable value of load for high-energy-consuming industrial users that integrates multi-objective similarity assessment and risk sensitivity perception includes the construction steps of a multi-objective similarity assessment model (MOSE) and a risk sensitivity perception model (RSPM). The positive / negative reference sequence output by the MOSE model is used as the reference benchmark of the RSPM model to achieve the integrated application of the two. The construction steps of the MOSE model include: setting an ideal reference vector, calculating the correlation coefficient between each candidate control scheme and the ideal reference vector, integrating them to obtain a comprehensive correlation coefficient, and constructing a positive / negative reference coefficient matrix based on the indicator type (benefit type / cost type), while achieving unified indicator dimensions through unified measurement transformation; The steps for constructing the RSPM model include: setting a decision reference benchmark, characterizing subjective value perception through a perceived value function, characterizing probability perception distortion through a decision weight function, and constructing positive / negative risk value functions based on the positive / negative reference sequences of the MOSE model to quantify the regulatory value of candidate solutions. The integrated application steps include: calculating the decision weight of each control indicator, integrating the positive / negative risk value function according to the weight, obtaining the comprehensive control value of high energy-consuming industrial users, and the method can be adapted to various load control scenarios such as maintenance, shift work, staggered operation, and peak avoidance. The parameter settings of the unified measure transformation satisfy the following: when i=0, the standard positive reference sequence and the standard negative reference sequence can be output respectively, realizing the dimensionless transformation of different dimensional indicators.

[0093] Example 3 Based on the same inventive concept, this invention also provides a load control value assessment system for high-energy-consuming industrial users, characterized in that it includes: The scheme generation module is used to generate candidate schemes based on the operation plans and constraints of high-energy-consuming industrial users, and generate rules through pre-set control strategies. The constraints include: equipment start-up and shutdown constraints, production interruptibility, and load transfer capability constraints. The similarity assessment module is used to input the candidate scheme into a pre-built multi-objective similarity assessment model to obtain the correlation coefficient of the candidate scheme relative to the positive / negative reference sequence, and to determine the deviation of the candidate scheme relative to the positive / negative reference sequence based on the correlation coefficient. The value assessment module is used to perform risk value assessment based on the deviation of candidate solutions from positive / negative reference sequences, combined with a pre-built risk sensitivity perception model. The constructed multi-objective similarity evaluation model is built by determining the ideal reference vector based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users, and combining it with similarity measurement methods.

[0094] Optionally, the similarity assessment module includes: The calculation submodule is used to compare the numerical value of each candidate solution in the multidimensional space with the positive / negative reference sequence in the multi-objective similarity evaluation model dimension by dimension using a similarity measurement method, so as to obtain the deviation of each candidate solution from the positive / negative reference sequence in the multidimensional space. The standardization submodule is used to convert the deviation of each candidate solution from the positive / negative reference sequence in multidimensional space into a standardized correlation coefficient.

[0095] Optionally, the value assessment module is specifically used for: By substituting the deviation of the candidate scheme from the positive / negative reference sequence into the positive / negative risk value function in the pre-constructed risk sensitivity perception model, the positive / negative risk value of the indicator relative to high energy-consuming industrial users under the maintenance and control value is obtained. The decision weights when facing gains / losses are determined by combining the weights of indicators under the value of maintenance and control with the subjective weight function of a pre-built risk sensitivity perception model. Based on the positive / negative risk value combined with the decision weight when facing gains / losses, the maintenance and control value of the candidate solutions is determined. Risk value assessment is conducted based on the maintenance and control value of the candidate solutions.

[0096] Optionally, the subjective weight function is shown in the following equation:

[0097] In the formula, The decision weight for the benefit, As the decision weight when losses occur, The weight of the j-th indicator under the maintenance and control value; and These are the risk attitude coefficients of decision-makers towards gains and losses, respectively. , The sensitivity of decision-makers to residual weights is characterized under gain and loss scenarios, respectively, to reflect the asymmetric distortion of the decision weight function in the high-weight range.

[0098] Example 4 like Figure 3 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0099] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the high-energy-consuming industrial user load regulation value assessment method in the above embodiments.

[0100] Example 5 Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the high-energy-consuming industrial user load control value assessment method described in the above embodiments.

[0101] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0102] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0103] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0104] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0105] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A method for evaluating the value of load regulation for high-energy-consuming industrial users, characterized in that, include: Based on the operation plans and constraints of high-energy-consuming industrial users, candidate solutions are generated by pre-set control strategies. The constraints include: equipment start-up and shutdown constraints, production interruptibility, and load transfer capability constraints. The candidate scheme is input into a pre-built multi-objective similarity evaluation model to obtain the correlation coefficient of the candidate scheme relative to the positive / negative reference sequence, and the deviation of the candidate scheme relative to the positive / negative reference sequence is determined based on the correlation coefficient. Risk value assessment is performed based on the deviation between candidate solutions and positive / negative reference sequences, combined with a pre-built risk sensitivity perception model. The constructed multi-objective similarity evaluation model is built by determining the ideal reference vector based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users, and combining it with similarity measurement methods.

2. The method as described in claim 1, characterized in that, The construction of the multi-objective similarity evaluation model includes: Based on the index types in the load control scenario of high energy-consuming industrial users, positive / negative reference coefficients are determined according to the calculation formulas of each type, and positive / negative reference sequences are constructed based on the positive / negative reference coefficients in chronological order. Using the positive and negative reference sequences as the basis for decision-making, a multi-objective similarity evaluation model is constructed by combining similarity measurement methods.

3. The method as described in claim 1, characterized in that, The step of inputting the candidate solution into a pre-constructed multi-objective similarity evaluation model to obtain the correlation coefficient of the candidate solution relative to the positive / negative reference sequence, and determining the deviation of the candidate solution relative to the positive / negative reference sequence based on the correlation coefficient, includes: A similarity metric method is used to compare the numerical values ​​of each candidate solution in the multidimensional space with the positive / negative reference sequences in the multi-objective similarity evaluation model dimension by dimension, so as to obtain the correlation coefficient of each candidate solution relative to the positive / negative reference sequences in the multidimensional space. The correlation coefficient of each candidate solution relative to the positive / negative reference sequences in the multidimensional space is then converted into a standardized similarity coefficient. The standardized similarity coefficient is used as the deviation of the candidate scheme from the positive / negative reference sequence.

4. The method as described in claim 3, characterized in that, The step of inputting the candidate scheme into a pre-constructed multi-objective similarity evaluation model to obtain the correlation coefficient between the candidate scheme and the positive / negative reference sequence, and determining the deviation of the candidate scheme relative to the positive / negative reference sequence based on the correlation coefficient, further includes: By integrating the standardized similarity coefficients of each candidate solution in each dimension, a comprehensive correlation coefficient of the candidate solution relative to the ideal reference vector is obtained.

5. The method as described in claim 3, characterized in that, The standardized similarity coefficient is calculated using the following formula: in, Let i be the correlation coefficient of scheme i under index k. It is the value of scheme i under index k. It is the value of the ideal reference vector under index k. It is the resolution coefficient.

6. The method as described in claim 4, characterized in that, The comprehensive correlation coefficient of the candidate scheme relative to the ideal reference vector is calculated using the following formula: Where n is the total number of indicators, It is the correlation coefficient of scheme i under index k. To achieve a comprehensive correlation coefficient, Let i be the number of indicators and i be the number of enterprises.

7. The method as described in claim 1, characterized in that, The risk value assessment based on the deviation between candidate solutions and positive / negative reference sequences, combined with a pre-built risk sensitivity perception model, includes: By substituting the deviation of the candidate scheme from the positive / negative reference sequence into the positive / negative risk value function in the pre-constructed risk sensitivity perception model, the positive / negative risk value of the indicator relative to high energy-consuming industrial users under the maintenance and control value is obtained. The decision weights when facing gains / losses are determined by combining the weights of indicators under the value of maintenance and control with the subjective weight function of a pre-built risk sensitivity perception model. Based on the positive / negative risk value combined with the decision weight when facing gains / losses, the maintenance and control value of the candidate solutions is determined. Risk value assessment is conducted based on the maintenance and control value of the candidate solutions.

8. The method as described in claim 7, characterized in that, The positive / negative risk value function is shown in the following formula: In the formula, For the positive risk value function of indicator j relative to high energy-consuming industrial user i under the maintenance and control value, Let index j be the negative risk value function relative to high-energy-consuming industrial user i under the maintenance and control value. The negative reference coefficient for the j-th indicator of high-energy-consuming industrial user i under the maintenance and control value. This serves as the positive reference coefficient for the j-th indicator of high-energy-consuming industrial user i under the maintenance and control value. and These represent the risk preference coefficient and the risk aversion coefficient, respectively; i represents the high-energy-consuming industrial user serial number; and j represents the indicator serial number. This represents the sensitivity coefficient of decision-makers to gains and losses.

9. The method as described in claim 7, characterized in that, The subjective weighting function is shown in the following formula: In the formula, The decision weight for the benefit, As the decision weight when losses occur, The weight of the j-th indicator under the maintenance and control value; and These are the risk attitude coefficients of decision-makers towards gains and losses, respectively. , It characterizes the sensitivity of decision-makers to residual weights under gain and loss scenarios.

10. The method as described in claim 7, characterized in that, The maintenance and control value is calculated using the following formula: In the formula, For the maintenance and control value of high-energy-consuming industrial users i, For the positive risk value function of indicator j relative to high energy-consuming industrial user i under the maintenance and control value, Let index j be the negative risk value function relative to high-energy-consuming industrial user i under the maintenance and control value. As a decision weight when faced with benefits, Let m represent the decision weights when facing losses, and m be the number of indicators.

11. A load control value assessment system for high-energy-consuming industrial users, characterized in that, include: The scheme generation module is used to generate candidate schemes based on the operation plans and constraints of high-energy-consuming industrial users, and generate rules through pre-set control strategies. The constraints include: equipment start-up and shutdown constraints, production interruptibility, and load transfer capability constraints. The similarity assessment module is used to input the candidate scheme into a pre-built multi-objective similarity assessment model to obtain the correlation coefficient of the candidate scheme relative to the positive / negative reference sequence, and to determine the deviation of the candidate scheme relative to the positive / negative reference sequence based on the correlation coefficient. The value assessment module is used to perform risk value assessment based on the deviation of candidate solutions from positive / negative reference sequences, combined with a pre-built risk sensitivity perception model. The constructed multi-objective similarity evaluation model is built by determining the ideal reference vector based on the target attributes and industry physical laws under the load control scenario of high-energy-consuming industrial users, and combining it with similarity measurement methods.

12. The system as claimed in claim 11, characterized in that, The similarity assessment module includes: The calculation submodule is used to compare the numerical value of each candidate solution in the multidimensional space with the positive / negative reference sequence in the multi-objective similarity evaluation model dimension by dimension using a similarity measurement method, so as to obtain the correlation coefficient of each candidate solution relative to the positive / negative reference sequence in the multidimensional space; The standardization submodule is used to convert the correlation coefficient of each candidate solution relative to the positive / negative reference sequence in the multidimensional space into a standardized similarity coefficient; the standardized similarity coefficient is used as the deviation of the candidate solution from the positive / negative reference sequence.

13. The system as described in claim 11, characterized in that, The valuation module is specifically used for: By substituting the deviation of the candidate scheme from the positive / negative reference sequence into the positive / negative risk value function in the pre-constructed risk sensitivity perception model, the positive / negative risk value of the indicator relative to high energy-consuming industrial users under the maintenance and control value is obtained. The decision weights when facing gains / losses are determined by combining the weights of indicators under the value of maintenance and control with the subjective weight function of a pre-built risk sensitivity perception model. Based on the positive / negative risk value combined with the decision weight when facing gains / losses, the maintenance and control value of the candidate solutions is determined. Risk value assessment is conducted based on the maintenance and control value of the candidate solutions.

14. The system as claimed in claim 11, characterized in that, The subjective weighting function is shown in the following formula: In the formula, The decision weight for the benefit, As the decision weight when losses occur, The weight of the j-th indicator under the maintenance and control value; and These are the risk attitude coefficients of decision-makers towards gains and losses, respectively. , It characterizes the sensitivity of decision-makers to residual weights under gain and loss scenarios.

15. An electronic device, characterized in that, include: At least one processor and memory; The memory and processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a method for evaluating the load regulation value of high-energy-consuming industrial users as described in any one of claims 1 to 10 is implemented.

16. A readable storage medium, characterized in that, It contains an execution program, which, when executed, implements a method for evaluating the load regulation value of high-energy-consuming industrial users as described in any one of claims 1 to 10.

Images