Power supply service strategy optimization method based on enterprise power supply service portrait

Abstract

The invention relates to a power supply service strategy optimization method based on an enterprise power supply service portrait. The method comprises the following steps: collecting and preprocessing enterprise historical load data, historical operation data and enterprise equipment data; constructing an enterprise load prediction model, and predicting future load data of the enterprise based on historical load data of the enterprise; calculating production intensity characteristics of the enterprise based on the historical operation data of the enterprise; based on the enterprise equipment data and the historical load data, enterprise equipment and load relevance characteristics are extracted, and enterprise future load data are corrected according to the relevance characteristics; identifying a load mode of the enterprise according to the corrected future load data of the enterprise; constructing an enterprise power supply service portrait based on the production intensity characteristics, the equipment and load relevance characteristics and the load mode of the enterprise; and constructing an enterprise power consumption demand elastic response model to calculate an enterprise power consumption elastic demand curve, and optimizing a power supply service strategy according to the enterprise power consumption elastic demand curve.

Description

Technical Field

[0001] The present invention relates to a method for optimizing power supply service strategies based on an enterprise power supply service profile, belonging to the technical field of energy management. Background Art

[0002] In the fields of power system planning and enterprise power consumption management, load forecasting and optimization of power supply service strategies have always been key links to achieve efficient, safe, and economic power supply. Traditional methods mainly include statistical methods and time series models based on historical data, such as autoregressive moving average model (ARMA), exponential smoothing method, and simple regression analysis. These methods usually have certain prediction capabilities when the data volume is small. However, with the increasing complexity of enterprise production, equipment operation, and market environment, traditional technologies have limitations in many aspects: Existing technologies often only focus on using enterprise historical load data for load forecasting, while ignoring the roles of multi-dimensional data such as enterprise production intensity, order fulfillment status, and equipment operation status. The use of a single data source or feature limits the model's ability to capture non-linear and multi-modal change laws; Traditional load forecasting models mainly rely on simple statistical models or shallow neural networks, and it is difficult to capture the long-term dependencies and complex non-linear features in enterprise power consumption data. Especially when facing the time series fluctuations of load data, changing production rhythms, and equipment status changes, traditional models are prone to problems such as underfitting or large prediction deviations; Previous power supply service strategies mainly focused on static cost analysis, did not fully consider the response elasticity of enterprises to price fluctuations, and lacked a comprehensive model that incorporates the synergistic effects of enterprise actual production, equipment operation, and market electricity prices into the decision-making process, making it difficult for power supply service strategies to cope with complex market environments and rapidly changing power consumption demands. Summary of the Invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a method for optimizing power supply service strategies based on an enterprise power supply service profile.

[0004] The technical solution of the present invention is as follows: On the one hand, the present invention provides a method for optimizing power supply service strategies based on an enterprise power supply service profile, including the following steps: Collect enterprise historical load data, historical operation data, and enterprise equipment data, and preprocess the historical load data, historical operation data, and enterprise equipment data; Construct an enterprise load forecasting model to predict the future load data of the enterprise based on the enterprise historical load data; Calculate the production intensity characteristics of the enterprise based on the enterprise historical operation data; Extract the correlation characteristics between enterprise equipment and load based on enterprise equipment data and historical load data, and correct the enterprise's future load data according to the correlation characteristics; Identify the load pattern of the enterprise according to the corrected future load data of the enterprise; Construct an enterprise power supply service portrait based on the enterprise's production intensity characteristics, equipment and load correlation characteristics, and load pattern; Construct an enterprise electricity demand elastic response model, input the enterprise power supply service portrait and the corrected future load data into the enterprise electricity demand elastic response model to obtain the enterprise electricity elastic demand curve, and optimize the power supply service strategy according to the enterprise electricity elastic demand curve.

[0005] Preferably, the enterprise load prediction model is constructed based on a deep hybrid attention time series network model, including a short-term feature extraction layer, a multi-head self-attention mechanism layer, and a multi-layer perceptron layer; After normalizing the load data of the enterprise at different historical times, construct an input sample set, as shown in the following formula: ; Where: Indicates up to The input sample set at the moment; Indicates The normalized enterprise load data at the moment; Indicates the length of the input sample set; Indicates the feature dimension; Indicates the real number field; The short-term feature extraction layer is constructed based on a convolutional neural network. Input the input sample set into the short-term feature extraction layer to extract short-term features, as shown in the following formula: ; Where: Indicates The output of the short-term feature extraction layer at the moment; Indicates the activation function; Indicates the number of convolutional layers in the short-term feature extraction layer; Indicates the Weight matrix of the th convolutional layer; Indicates the Bias of the th convolutional layer; After splicing the short-term features, input them into the multi-head self-attention mechanism layer to extract long-term features, as shown in the following formula: ; ; Where: Indicates the long-term features extracted by the multi-head self-attention layer; 、 , respectively represent the query vector, key vector, and value vector for splicing short-term features; represents the splicing operation; represents the output of the th attention head; represents the weight matrix of the linear projection of the output of the multi-head attention; represents the normalization function; , , respectively represent the parameter weight matrices corresponding to the query vector, key vector, and value vector of the th attention head; represents the transpose operation; represents the dimensionality of the th attention head;

[0006] Preferably, the calculation formula for the production intensity feature of the enterprise is: ; where: represents the production intensity feature of the enterprise at time represents the normalization factor; represents the actual production capacity of the enterprise at time represents the historical maximum production capacity of the enterprise; represents the urgency of order fulfillment of the enterprise at time represents the new order volume of the enterprise at time represents the non-linear control parameter; , , represent the weight coefficients of the production intensity feature; represents the natural constant.

[0007] Preferably, the feature of the correlation between the enterprise equipment and the load is extracted based on the enterprise equipment data and the historical load data as shown in the following formula: ; where: represents the correlation feature between the enterprise equipment and the enterprise historical load at time represents the historical time interval traced back from time; represents the maximum backtracking window length; Represents the enterprise equipment The set of spatial neighborhood equipment; Represents the enterprise equipment And the spatial weight of the th equipment in its spatial neighborhood equipment set; Represents the enterprise equipment The th equipment in the spatial neighborhood equipment set of the enterprise equipment; Represents the historical time weight decay coefficient; Represents The historical load data of the equipment at time ; Represents The overall historical load data of the enterprise at time Represents the correlation function; The future load data of the enterprise is corrected according to the correlation characteristics, as shown in the following formula: ; Where: Represents The corrected future load data of the enterprise at time Represents The future load data of the enterprise at time Represents the prediction span time; Represents the enterprise equipment The importance weight; Represents The correlation characteristics between the enterprise equipment at time and the enterprise historical load; Represents The load data of the enterprise equipment at time ;

[0008] Preferably, the specific steps for identifying the load pattern of the enterprise according to the corrected future load data of the enterprise are as follows: Construct typical load pattern time series of the same length for different types; Align the future load data of the enterprise with any typical load pattern time series through an improved dynamic time warping algorithm, as shown in the following formula: ; ; Where: Represents the output of the improved dynamic time warping algorithm, that is, aligning the future load data of the enterprise with the typical load pattern time series ; Represents aligning the future load data of the enterprise The set of alignment methods between the typical load pattern time series ; denote the th data point in denote the th data point in denote the time weight function; denote the time offset penalty factor; denote the distance between and Based on the aligned future load data of the enterprise, the load pattern closest to it is identified through a Gaussian mixture model, as shown in the following formula: ; ; where: denote the aligned future load data of the enterprise; denote the set of parameters of the Gaussian mixture model, ; denote the number of typical load patterns; denote that under the condition of the given set of parameters , the probability density of the occurrence of denote the prior probability of the th load pattern; denote the determinant of the covariance matrix constructed based on the typical load pattern time series corresponding to the th load pattern; denote the mean of the typical load pattern time series corresponding to the th load pattern; denote the load pattern corresponding to the future load data of the enterprise at time denote the data point of the aligned future load data of the enterprise at time ; denote the set of parameters of the Gaussian mixture model corresponding to the load pattern under which the probability density of its occurrence is the largest.

[0009] Preferably, an elastic response model for the electricity demand of the enterprise is constructed, and the portrait of the enterprise's power supply service is input into the elastic response model for the electricity demand of the enterprise to obtain the elastic demand curve for the electricity of the enterprise, as shown in the following formula: ; where: denote the elastic electricity demand of the enterprise at time Indicates the total number of time-of-use electricity price tiers; Indicates the price elasticity coefficient of the th time-of-use electricity price tier; Indicates the electricity price of the th time-of-use electricity price tier; Indicates the base electricity price; Indicates the enterprise power supply service portrait at time

[0010] Preferably, the power supply service strategy is optimized according to the enterprise's elastic electricity demand curve, as shown in the following formula: ; Where: Indicates the optimized power supply service strategy of the enterprise; Indicates the electricity price at time Indicates the balance coefficient; Indicates the variance of the enterprise's elastic electricity demand.

[0011] On the other hand, the present invention also provides a power supply service strategy optimization system based on the enterprise power supply service portrait, including a data acquisition module, an enterprise load forecasting module, a production intensity feature generation module, a correlation feature production module, a load pattern recognition module, an enterprise power supply service portrait construction module, and a power supply service strategy optimization module; The data acquisition module is used to collect the enterprise's historical load data, historical operation data, and enterprise equipment data, and preprocess the historical load data, historical operation data, and enterprise equipment data; The enterprise load forecasting module is used to construct an enterprise load forecasting model and forecast the enterprise's future load data based on the enterprise's historical load data; The production intensity feature generation module is used to calculate the enterprise's production intensity feature based on the enterprise's historical operation data; The correlation feature production module is used to extract the correlation features between the enterprise equipment and the load based on the enterprise equipment data and the historical load data, and correct the enterprise's future load data according to the correlation features; The load pattern recognition module is used to identify the enterprise's load pattern according to the corrected enterprise's future load data; The enterprise power supply service portrait construction module is used to construct an enterprise power supply service portrait based on the enterprise's production intensity feature, the correlation feature between the equipment and the load, and the load pattern; The power supply service strategy optimization module is used to construct an elastic response model for the electricity demand of enterprises. The enterprise power supply service portrait and the corrected future load data are input into the elastic response model for the electricity demand of enterprises to obtain the elastic demand curve for the electricity consumption of enterprises, and the power supply service strategy is optimized according to the elastic demand curve for the electricity consumption of enterprises.

[0012] In another aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method as described in the present invention is implemented.

[0013] In another aspect, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method as described in the present invention is implemented.

[0014] The present invention has the following beneficial effects: 1. By collecting the historical load data, historical operation data, and enterprise equipment data of enterprises, the present invention constructs a comprehensive enterprise power supply service portrait. This portrait integrates production intensity characteristics, the correlation between equipment and load, and typical load patterns, and can comprehensively reflect the actual situation of each link of enterprises from production to power supply, improving the adaptability and accuracy of load forecasting and power consumption response.

[0015] 2. The present invention proposes a load forecasting model based on a deep hybrid attention time series network. By introducing a convolutional neural network for short-term feature extraction, a multi-head self-attention mechanism to capture long-term dependencies, and a multi-layer perceptron for mapping, the modeling ability of the model for complex time series features and non-linear relationships is greatly improved, ensuring the robustness and accuracy of the forecasting results.

[0016] 3. The present invention constructs an elastic response model for electricity demand, combines the enterprise power supply service portrait with the predicted load data, considers the influence of electricity price, price elasticity coefficient, and comprehensive enterprise characteristics, and through the analysis of the elastic demand curve, realizes cost minimization on the premise of ensuring stable power consumption. Combined with the optimization objective function of the power supply service strategy, while reducing the power supply cost, the volatility of the power consumption load is further balanced, thereby improving the economy and stability of the power supply system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] It should be understood that the step numbers used in the text are only for convenient description and do not limit the execution order of the steps.

[0020] It should be understood that the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless otherwise clearly specified in the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0021] The terms "comprising" and "including" indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0022] The term " / or" refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0023] Embodiment 1: See Figure 1 , a method for optimizing power supply service strategies based on the power supply service portrait of an enterprise, including the following steps: Collect the historical load data, historical operation data, and enterprise equipment data of the enterprise, and preprocess the historical load data, historical operation data, and enterprise equipment data, including missing value imputation, outlier removal, and normalization; Construct an enterprise load prediction model and predict the future load data of the enterprise based on the historical load data of the enterprise; Calculate the production intensity characteristics of the enterprise based on the historical operation data of the enterprise; Extract the correlation characteristics between the enterprise equipment and the load based on the enterprise equipment data and the historical load data, and correct the future load data of the enterprise according to the correlation characteristics; Identify the load pattern of the enterprise according to the corrected future load data of the enterprise; Construct an enterprise power supply service portrait based on the production intensity characteristics, equipment-load correlation characteristics, and load pattern of the enterprise; Construct an enterprise electricity demand elasticity response model, input the enterprise power supply service portrait and the corrected future load data into the enterprise electricity demand elasticity response model to obtain the enterprise electricity elasticity demand curve, and optimize the power supply service strategy according to the enterprise electricity elasticity demand curve.

[0024] As a preferred implementation manner of this embodiment, the enterprise load forecasting model is constructed based on a deep hybrid attention time series network model, and includes a short-term feature extraction layer, a multi-head self-attention mechanism layer, and a multi-layer perceptron layer; Normalize the load data of the enterprise at different historical times and construct an input sample set, as shown in the following formula: ; Where: Represents the input sample set up to Time; Represents The normalized enterprise load data at time; Represents the length of the input sample set; Represents the feature dimension; Represents the real number field; The short-term feature extraction layer is constructed based on a convolutional neural network. Input the input sample set into the short-term feature extraction layer to extract short-term features, as shown in the following formula: ; Where: Represents The output of the short-term feature extraction layer at time; Represents the activation function; Represents the number of convolutional layers of the short-term feature extraction layer; Represents the Weight matrix of the th convolutional layer; Represents the Bias of the th convolutional layer; Concatenate the short-term features and input them into the multi-head self-attention mechanism layer to extract long-term features, as shown in the following formula: ; ; Where: Represents the long-term features extracted by the multi-head self-attention layer; , , Represents the query vector, key vector, and value vector for concatenating the short-term features respectively; Represents the concatenation operation; Represents the Output of the th attention head; Represents the total number of attention heads (usually 8, 12, 16, etc.); Represents the output linear projection weight matrix of the multi-head attention, which is used to transform the concatenated head outputs into the desired dimension; Represents the normalization function; 、 、 Represent the parameter weight matrices of the query vector, key vector, and value vector corresponding to the -th attention head respectively; Represents the transpose operation; Represents the -th dimension of the attention head; Then, the long-term features are input into a multi-layer perceptron layer to map and generate the load prediction result.

[0025] As a preferred implementation manner of this embodiment, the calculation formula for the production intensity feature of the enterprise is: ; Where: Represents the production intensity feature of the enterprise at the -th moment; Represents the normalization factor; Represents the -th moment of the actual production capacity of the enterprise; Represents the historical maximum production capacity of the enterprise; Represents the -th moment of the order fulfillment urgency of the enterprise; Represents the -th moment of the new order volume of the enterprise; Represents the non-linear control parameter; 、 、 Represent the production intensity feature weight coefficients; Represents the natural constant.

[0026] As a preferred implementation manner of this embodiment, the extraction of the enterprise equipment and load correlation feature based on the enterprise equipment data and historical load data is specifically shown in the following formula: ; Where: Represents the correlation feature between the enterprise equipment at the -th moment and the enterprise historical load; Represents the historical time interval traced back from the -th moment; Represents the maximum backtracking window length; Represents the set of spatial neighboring equipment of the enterprise equipment ; Represents the enterprise equipment and the The spatial weight of a device is usually calculated based on the physical distance between devices or the similarity of device functions, reflecting the strength of spatial association; Indicates the enterprise device The th device in the set of spatial neighborhood devices; Indicates the historical time weight decay coefficient; Indicates The historical load data of device at time Indicates The overall historical load data of the enterprise at time Indicates the correlation function; The specific form of the correlation function is as follows: ; ; Where: Indicates the maximum length of the short-term dynamic window, controlling the depth of tracing the dynamic change characteristics of the load; Indicates the time decay weight function of the short-term dynamic window with length , , Indicates the time weight decay rate; Indicates the historical load data of enterprise device at time; Indicates at time the overall historical load data of the enterprise; Indicates the non-linear coupling correlation function, where , ; Indicates the mean value of the historical load data of enterprise device within the short-term dynamic window with length ; Indicates the mean value of the overall historical load data of the enterprise within the short-term dynamic window with length ; Indicates the standard deviation of the historical load data of enterprise device within the short-term dynamic window with length ; Indicates the standard deviation of the overall historical load data of the enterprise within the short-term dynamic window with length ; Indicates a constant; The future load data of the enterprise is corrected according to the association characteristics, and the specific form is as follows: ; Where: Indicates The enterprise future load data corrected at the moment; denote the enterprise future load data at the moment; denote the prediction span moment; denote the enterprise equipment importance weight; denote the relevance feature between the enterprise equipment at the moment and the enterprise historical load; denote the load data of the enterprise equipment at the moment .

[0027] As a preferred implementation manner of this embodiment, the specific steps for identifying the load pattern of the enterprise according to the corrected enterprise future load data are as follows: Construct typical load pattern time series of the same length with different types; Align the enterprise future load data with any typical load pattern time series through an improved dynamic time warping algorithm, as shown in the following formula: ; ; where: denote the output of the improved dynamic time warping algorithm, that is, align the enterprise future load data with the typical load pattern time series ; denote the set of alignment methods between the enterprise future load data and the typical load pattern time series , that is, all alignment methods that meet certain constraint conditions (boundary conditions, continuity, and monotonicity) between two time series data; The boundary condition means that the path must start from the starting points of the two sequences and finally reach the end points of the two sequences; Continuity means that the index change between adjacent points in the path is restricted, and the usually allowed change is: moving one grid forward or moving diagonally forward; Monotonicity means that the index in the path must be strictly monotonically increasing, that is, there cannot be a situation of going back, ensuring that the time order of the time series is not disrupted; denote the th data point in denote the th data point in denote the time weight function; denote the time offset penalty factor; denote The distance between ; Based on the aligned future load data of the enterprise, the load pattern closest to it is identified through a Gaussian mixture model, as shown in the following formula: ; ; Where: represents the aligned future load data of the enterprise; represents the parameter set of the Gaussian mixture model, ; represents the number of typical load patterns; represents that under the condition of the given parameter set ; is the probability density of the occurrence of represents the th prior probability of the load pattern, reflecting the proportion of the th load pattern in the overall data; represents the determinant of the covariance matrix constructed based on the typical load pattern time series corresponding to the th load pattern; represents the mean of the typical load pattern time series corresponding to the th load pattern; represents the load pattern corresponding to the future load data of the enterprise at time represents the data point of the aligned future load data of the enterprise at time ; represents the parameter set of the Gaussian mixture model corresponding to the load pattern with the maximum probability density of its occurrence ; represents a part of the normalization constant in the normal distribution density function; The specific load patterns include: 1. Stable load pattern Its characteristic is that the power consumption changes relatively smoothly, with small fluctuations, and the overall load level remains basically stable.

[0028] 2. Peak-valley alternating pattern Its characteristic is that the electricity demand has obvious periodic fluctuations, usually showing a sharp increase in load during certain periods (such as daytime, production peak periods), and a significant decrease in load at night or during rest periods.

[0029] 3. Step-shaped load pattern Its characteristic is that the load shows segmented changes with different stages of the production process or equipment status, and obvious jumps may be caused by equipment start-stop, process conversion, etc.

[0030] 4. Sudden load pattern Its characteristic is that under specific events or abnormal situations (such as equipment failures, emergency dispatching), the load will sharply rise or fall within a short period of time, showing sudden fluctuations.

[0031] 5. Increasing or decreasing trend pattern Its characteristic is that during the period of enterprise production expansion or reduction, the overall load shows a gradually increasing or decreasing trend.

[0032] 6. Seasonal or periodic change pattern Its characteristic is that the load is affected by seasons, climate, or market demand cycles, showing a long-term fluctuating trend.

[0033] As a preferred implementation manner of this embodiment, an elastic response model for enterprise electricity demand is constructed. The enterprise power supply service portrait is input into the elastic response model for enterprise electricity demand to obtain the enterprise electricity elastic demand curve, as shown in the following formula:[[]] ; Where:[[]] represents the enterprise electricity elastic demand at time represents the total number of time-of-use electricity price tiers; represents the th price elasticity coefficient of the time-of-use electricity price tier; represents the th electricity price of the time-of-use electricity price tier; represents the basic electricity price; represents the service portrait influence function; represents the enterprise power supply service portrait at time The specific form of the service portrait influence function is as shown in the following formula:[[]] ; Where:[[]] represents the th weight vector of the time-of-use electricity price tier; represents the th bias of the time-of-use electricity price tier; represents the Sigmoid activation function; The specific form of the enterprise power supply service portrait is as shown in the following formula:[[]] ; As a preferred implementation manner of this embodiment, the power supply service strategy is optimized according to the enterprise electricity elastic demand curve, as shown in the following formula:[[]] ; Where:[[]] Represents the optimized power supply service strategy of the enterprise; Represents The electricity price at time Represents the balance coefficient, which is used to balance the trade-off between cost minimization and load smoothness. When is larger, more attention will be paid to the stability of the load (reducing fluctuations) during optimization. When is smaller, the main focus is on reducing the power supply cost; Represents the variance of the elastic demand for the enterprise's electricity consumption; By minimizing to find the optimal power supply strategy, so as to achieve the goal of the lowest cost or the best load smoothness on the basis of meeting the electricity demand. The actual optimization variables in the overall decision-making include: the enterprise's active load management strategy (such as load peak shaving and peak shifting measures), the start-stop and charge-discharge strategies of energy storage devices, and the reasonable adjustment of the load time sequence in the production process.

[0034] Embodiment 2: A power supply service strategy optimization system based on the enterprise power supply service portrait, including a data acquisition module, an enterprise load prediction module, a production intensity feature generation module, a correlation feature production module, a load pattern recognition module, an enterprise power supply service portrait construction module, and a power supply service strategy optimization module; The data acquisition module is used to collect the enterprise's historical load data, historical operation data, and enterprise equipment data, and preprocess the historical load data, historical operation data, and enterprise equipment data; The enterprise load prediction module is used to construct an enterprise load prediction model and predict the enterprise's future load data based on the enterprise's historical load data; The production intensity feature generation module is used to calculate the production intensity feature of the enterprise based on the enterprise's historical operation data; The correlation feature production module is used to extract the correlation features between the enterprise equipment and the load based on the enterprise equipment data and the historical load data, and correct the enterprise's future load data according to the correlation features; The load pattern recognition module is used to identify the load pattern of the enterprise according to the corrected enterprise's future load data; The enterprise power supply service portrait construction module is used to construct an enterprise power supply service portrait based on the enterprise's production intensity feature, equipment and load correlation feature, and load pattern; The power supply service strategy optimization module is used to construct an enterprise electricity demand elastic response model, input the enterprise power supply service portrait and the corrected future load data into the enterprise electricity demand elastic response model to obtain the enterprise electricity elastic demand curve, and optimize the power supply service strategy according to the enterprise electricity elastic demand curve.

[0035] This system is used to implement the method in Embodiment 1, which will not be elaborated here.

[0036] Embodiment 3: This embodiment provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method described in any embodiment of the present invention is implemented.

[0037] Embodiment 4: This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method described in any embodiment of the present invention is implemented.

[0038] In the embodiments of the present application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent the case where A exists alone, A and B exist simultaneously, or B exists alone. Where A and B may be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after. "At least one of the following" and its similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, and c may represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c may be single or multiple.

[0039] Those of ordinary skill in the art can realize that the units and algorithm steps described in the embodiments disclosed herein can be implemented by a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0040] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be elaborated here.

[0041] In several embodiments provided by the present application, if any 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 such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0042] The above are only the embodiments of the present invention, and thus do not limit the patent scope of the present invention. Any equivalent structural or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for optimizing power supply service strategies based on the enterprise power supply service profile, characterized in that, It includes the following steps: Collect the historical load data, historical operation data and enterprise equipment data of the enterprise, and preprocess the historical load data, historical operation data and enterprise equipment data; Construct an enterprise load forecasting model, and forecast the future load data of the enterprise based on the historical load data of the enterprise; Calculate the production intensity characteristics of the enterprise based on the historical operation data of the enterprise; Extract the correlation characteristics between enterprise equipment and load based on the enterprise equipment data and historical load data, and correct the future load data of the enterprise according to the correlation characteristics; Identify the load pattern of the enterprise according to the corrected future load data of the enterprise; Construct an enterprise power supply service portrait based on the production intensity characteristics, equipment and load correlation characteristics, and load pattern of the enterprise; Construct an enterprise electricity demand elastic response model, input the enterprise power supply service portrait and the corrected future load data into the enterprise electricity demand elastic response model to obtain the enterprise electricity elastic demand curve, and optimize the power supply service strategy according to the enterprise electricity elastic demand curve.

2. The power supply service strategy optimization method based on the enterprise power supply service portrait according to claim 1, wherein The enterprise load forecasting model is constructed based on a deep hybrid attention time series network model, and includes a short-term feature extraction layer, a multi-head self-attention mechanism layer and a multi-layer perceptron layer; Normalize the load data of the enterprise at different historical times and construct an input sample set, as shown in the following formula: ； Wherein: represents the input sample set up to time; represents the normalized enterprise load data at time; represents the length of the input sample set; represents the real number field; The short-term feature extraction layer is constructed based on a convolutional neural network, and the input sample set is input into the short-term feature extraction layer to extract short-term features, as shown in the following formula: ； Wherein: denotes the output of the short-term feature extraction layer at a moment; denotes the activation function; denotes the number of convolutional layers of the short-term feature extraction layer; denotes the weight matrix of the th convolutional layer; bias of the Concatenate the short-term features and input them into the multi-head self-attention mechanism layer to extract long-term features, as shown in the following formula: ； ； Wherein: represents the long-term features extracted by the multi-head self-attention layer; , , respectively represent the query vector, key vector, and value vector for splicing short-term features; represents the splicing operation; represents the output of the th attention head; represents the total number of attention heads; represents the output linear projection weight matrix of the multi-head attention; represents the normalization function; , , respectively represent the parameter weight matrices of the query vector, key vector, and value vector corresponding to the th attention head; represents the transpose operation; represents the th dimension of the attention head; Then input the long-term features into the multi-layer perceptron layer to map and generate the load forecasting result.

3. The power supply service strategy optimization method based on the enterprise power supply service portrait according to claim 2, characterized in that The calculation formula for the production intensity characteristics of the enterprise is: ； Wherein: represents the production intensity characteristic of the enterprise at time the normalization factor; represents the actual production capacity of the enterprise at time represents the historical maximum production capacity of the enterprise; represents the urgency of order fulfillment of the enterprise at time represents the new order volume of the enterprise at time represents the non - linear control parameter; , , represent the weight coefficients of the production intensity characteristics; represents the natural constant.

4. The power supply service strategy optimization method based on the enterprise power supply service portrait according to claim 3, characterized in that Extract the correlation characteristics between enterprise equipment and load based on the enterprise equipment data and historical load data, as shown in the following formula: ； in: express Moment Enterprise Equipment Correlation characteristics with the enterprise's historical load; Indicates from A historical time interval that looks back at all times; Indicates the maximum lookback window length; Represents enterprise equipment The set of spatial neighborhood devices; Represents enterprise equipment The first device in its spatial neighborhood The spatial weight of each device; Represents enterprise equipment The first device in the set of spatial neighbors Devices; Represents the historical time weight attenuation coefficient; express Time equipment Historical load data; express The overall historical load data of the enterprise at all times; represents the correlation function; Correct the future load data of the enterprise according to the correlation characteristics, as shown in the following formula: ； Wherein: represents the future load data of the enterprise after moment correction; represents the future load data of the enterprise at moment represents the prediction span moment; represents the enterprise equipment importance weight of represents the correlation characteristic between the enterprise equipment at moment and the historical load of the enterprise; represents the load data of the enterprise equipment at moment ​ 5. The power supply service strategy optimization method based on the enterprise power supply service portrait according to claim 4, wherein The specific steps for identifying the load pattern of the enterprise according to the corrected future load data of the enterprise are: Construct typical load pattern time series of different types with the same length; Align the future load data of the enterprise with any typical load pattern time series through an improved dynamic time warping algorithm, as shown in the following formula: ； ； Wherein: represents the output of the improved dynamic time warping algorithm, that is, aligning the future load data of the enterprise with the time series of the typical load pattern for alignment; represents the set of alignment methods between the future load data of the enterprise and the time series of the typical load pattern ; represents the th data point in ; represents the th data point in ; represents the time weight function; represents the time offset penalty factor; represents the distance between and ; Based on the aligned future load data of the enterprise, identify the load pattern closest to it through a Gaussian mixture model, as shown in the following formula: ； ； Wherein: represents the aligned future load data of the enterprise; represents the parameter set of the Gaussian mixture model, ; represents the number of typical load patterns; represents that under the condition of the given parameter set , the probability density of occurrence; represents the prior probability of the $i$-th load pattern; represents the determinant of the covariance matrix constructed based on the typical load pattern time series corresponding to the $i$-th load pattern; represents the mean of the typical load pattern time series corresponding to the $i$-th load pattern; represents the load pattern corresponding to the future load data of the enterprise at time $t$; represents the data point of the aligned future load data of the enterprise at time $t$; represents the parameter set of the Gaussian mixture model corresponding to the load pattern with the maximum probability density of its occurrence.

6. The power supply service strategy optimization method based on the enterprise power supply service portrait according to claim 5, characterized in that, Construct an enterprise electricity demand elastic response model, input the enterprise power supply service portrait into the enterprise electricity demand elastic response model to obtain the enterprise electricity elastic demand curve, as shown in the following formula: ； Wherein: represents the elastic demand for electricity of the enterprise at a certain moment; represents the total number of time-of-use electricity price tiers; represents the price elasticity coefficient of the th time-of-use electricity price tier; represents the basic electricity price; represents the service portrait influence function; represents the enterprise power supply service portrait at a certain moment.

7. A power supply service strategy optimization method based on the enterprise power supply service portrait according to claim 6, characterized in that, Optimize the power supply service strategy according to the enterprise electricity elastic demand curve, as shown in the following formula: ； Wherein: represents the optimized power supply service strategy of the enterprise; represents the electricity price at time represents the balance coefficient; represents the variance of the elastic demand for electricity of the enterprise.

8. A power supply service strategy optimization system based on an enterprise power supply service profile, characterized in that, It includes a data acquisition module, an enterprise load forecasting module, a production intensity characteristic generation module, a correlation characteristic production module, a load pattern recognition module, an enterprise power supply service portrait construction module and a power supply service strategy optimization module; The data acquisition module is used to collect the historical load data, historical operation data and enterprise equipment data of the enterprise, and preprocess the historical load data, historical operation data and enterprise equipment data; The enterprise load forecasting module is used to build an enterprise load forecasting model and predict the future load data of the enterprise based on the historical load data of the enterprise; The production intensity feature generation module is used to calculate the production intensity feature of the enterprise based on the historical operation data of the enterprise; The correlation feature production module is used to extract the correlation features between the enterprise equipment and the load based on the enterprise equipment data and the historical load data, and correct the future load data of the enterprise according to the correlation features; The load pattern recognition module is used to identify the load pattern of the enterprise according to the corrected future load data of the enterprise; The enterprise power supply service portrait construction module is used to construct an enterprise power supply service portrait based on the production intensity feature, the correlation feature between the equipment and the load, and the load pattern of the enterprise; The power supply service strategy optimization module is used to build an elastic response model for the electricity demand of the enterprise, input the enterprise power supply service portrait and the corrected future load data into the elastic response model for the electricity demand of the enterprise to obtain the elastic demand curve for the electricity of the enterprise, and optimize the power supply service strategy according to the elastic demand curve for the electricity of the enterprise.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that When the processor executes the program, the method described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the method described in any one of claims 1 to 7 is implemented.

Images