Double-sequence prediction method and system for power generation and consumption

Abstract

The invention discloses a double-sequence prediction method and system for power generation and consumption. The method comprises the following steps: acquiring first data of a power system; extracting features of the first data by using a convolutional neural network to obtain a first feature map, and performing weighted attention processing on the first feature map to obtain a second feature map; and performing transformation activation processing on the second feature map to obtain a first predicted value and a second predicted value, and evaluating the difference between the first predicted value and the second predicted value so as to adjust the operation strategy of the power system. By combining the space-time convolutional neural network and the space attention mechanism, the method can more effectively capture the complex space-time characteristics in the data, so that the prediction accuracy of power generation and consumption is remarkably improved; according to the double-sequence prediction capability, power generation and consumption can be predicted at the same time, so that the dynamic power grid load requirement is better met, and the operation efficiency of the micro-grid is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power forecasting, and in particular to a dual sequence forecasting method and system for power generation and consumption. Background Art

[0002] In the field of power generation and consumption forecasting, improving the accuracy and efficiency of forecasting models is the core of current research. With the rapid growth of energy demand and the instability of renewable energy, researchers have proposed a variety of forecasting models based on statistics, machine learning, and deep learning, among which hybrid models have performed well in extracting spatiotemporal features and improving forecasting accuracy. However, this field still faces important challenges.

[0003] Existing technologies have significant deficiencies in power generation and consumption forecasting. First, there is a lack of dual-sequence forecasting methods that can simultaneously predict power generation and consumption, which limits the model's ability to cope with dynamic grid load demands. Modern microgrids require systems that can accurately predict power generation and consumption to ensure grid balance and efficiency. However, most current studies focus on the prediction of one of these aspects, ignoring the interaction between the two. Second, some forecasting models have limitations in capturing spatial dependencies in complex data. Spatial attention mechanism is an effective method that can enhance the model's ability to perceive the importance of different locations in the input data. However, existing models fail to fully utilize this mechanism, resulting in limited ability to capture spatial dependencies in practical applications, thus affecting the accuracy and performance of the model. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a dual sequence prediction method for power generation and consumption to solve the problems of insufficient dual sequence prediction ability and insufficient spatial dependency capture ability in power generation and consumption prediction.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a dual sequence prediction method for power generation and consumption, comprising:

[0008] Obtaining the first data of the power system;

[0009] Extracting features of the first data using a convolutional neural network to obtain a first feature map, and performing weighted attention processing on the first feature map to obtain a second feature map;

[0010] The second characteristic graph is transformed and activated to obtain a first prediction value and a second prediction value, and the difference between the first prediction value and the second prediction value is evaluated to adjust the power system operation strategy.

[0011] As a preferred solution of the dual sequence prediction method for power generation and consumption described in the present invention, wherein: extracting the features of the first data using a convolutional neural network includes:

[0012] Design the convolution kernels of the convolutional neural network, each convolution kernel slides in time and space;

[0013] Calculating the dot product of the first data and the local area of ​​the convolution kernel to generate an original feature map;

[0014] A pooling operation is performed on the original feature map to reduce the spatial size and time length of the original feature map, and a weight value is calculated to obtain a first feature map.

[0015] As a preferred solution of the dual sequence prediction method for power generation and consumption described in the present invention, wherein: performing weighted attention processing on the first feature map includes:

[0016] The first feature map is input into the attention module, and the first feature map is weighted by using the weight value to adjust the importance of different positions in the first feature map to obtain a second feature map.

[0017] As a preferred embodiment of the dual sequence prediction method for power generation and consumption of the present invention, evaluating the difference between the first prediction value and the second prediction value comprises:

[0018] Performing data alignment on the first predicted value and the second predicted value;

[0019] An intelligent algorithm is used to quantify the degree of match between the first prediction value and the second prediction value, and the power system operation strategy is adjusted according to the degree of match.

[0020] As a preferred solution of the dual sequence prediction method for power generation and consumption described in the present invention, wherein: the dot product of the first data and the local area of ​​the convolution kernel is calculated, and the calculation formula is expressed as:

[0021]

[0022] Where X is the first input data, is the weight of the convolution kernel, k t and k sare the indices of the time dimension and space dimension respectively, and STConv(X) represents the original feature map.

[0023] As a preferred solution of the dual sequence prediction method for power generation and consumption described in the present invention, the calculation of the weight value includes:

[0024] First, the original feature map is pooled, which is expressed as:

[0025] Pooling(X) = agg(X t:t+T,s:s+S )

[0026] Where agg represents the aggregation function in the pooling operation, X is the input feature, T and S represent the time range and spatial range of the feature respectively;

[0027] Then calculate the weight value of each feature, expressed as:

[0028] Attention Weights=σ(FC(Pooling(X)))

[0029] Among them, Pooling represents the global average pooling output, FC represents the fully connected layer in the pooling operation, σ represents the activation function in the pooling operation, and Attention Weights represents the generated weights.

[0030] As a preferred solution of the dual sequence prediction method for power generation and consumption described in the present invention, the intelligent algorithm includes:

[0031] The KLD algorithm is used to evaluate the difference between the first predicted value and the second predicted value, expressed as:

[0032]

[0033] Among them, P(i) represents the probability value of the distribution at the i-th position, Q(i) represents the probability value of the distribution at the i-th position, and D KL (P||Q) is the difference measure with respect to .

[0034] In a second aspect, the present invention provides a system for dual sequence prediction of power generation and consumption, comprising:

[0035] A data acquisition module, used for acquiring first data of the power system;

[0036] A feature extraction module, configured to extract features of the first data using a convolutional neural network to obtain a first feature map, and perform weighted attention processing on the first feature map to obtain a second feature map;

[0037] The prediction module is used to perform transformation activation processing on the second feature map to obtain a first prediction value and a second prediction value, evaluate the difference between the first prediction value and the second prediction value, and then adjust the power system operation strategy.

[0038] In a third aspect, the present invention provides a computing device, comprising:

[0039] Memory and processor;

[0040] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the dual sequence prediction method of power generation and consumption are implemented.

[0041] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the dual sequence prediction method for power generation and consumption.

[0042] Compared with the prior art, the present invention has the following advantages: by combining the spatiotemporal convolutional neural network with the spatial attention mechanism, the method can more effectively capture the complex spatiotemporal features in the data, thereby significantly improving the prediction accuracy of power generation and consumption; the dual-sequence prediction capability of the present invention enables it to simultaneously predict power generation and consumption, thereby better responding to dynamic grid load demands and improving the operating efficiency of the microgrid. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0044] Figure 1 A schematic diagram of a general flow chart of a dual sequence prediction method for power generation and consumption according to an embodiment of the present invention;

[0045] Figure 2 The figure is a schematic diagram of the overall process of a dual sequence prediction method for power generation and consumption according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0049] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0050] At the same time, in the description of the present invention, it should be noted that the orientations or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the system or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0051] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0052] Example 1

[0053] Reference Figure 1-2, as an embodiment of the present invention, provides a dual sequence prediction method for power generation and consumption, comprising:

[0054] S100: Acquire first power system data;

[0055] S102: extracting features of the first data using a convolutional neural network to obtain a first feature map, and performing weighted attention processing on the first feature map to obtain a second feature map;

[0056] S104: performing transformation activation processing on the second characteristic graph to obtain a first prediction value and a second prediction value, evaluating the difference between the first prediction value and the second prediction value, and then adjusting the power system operation strategy;

[0057] It should be noted that the present invention first obtains the basic data of the power system, and then uses advanced convolutional neural network technology to deeply mine the features of these data, and further highlights the key features through the weighted attention mechanism, suppresses irrelevant information, and on this basis, transforms and activates the feature data to accurately predict the first prediction value and the second prediction value, that is, the two key indicator values ​​of power generation and consumption; by carefully comparing the differences between the two prediction values, potential problems in the operation of the power system can be discovered in time, and the operation strategy can be adjusted accordingly. The method of the present invention not only significantly improves the efficiency and accuracy of data analysis, but also greatly enhances the accuracy of prediction, optimizes the structure of the power system, and improves energy utilization efficiency.

[0058] In an optional embodiment, the first data of the power system in step S100 may be power generation, power consumption, power load in different time periods, power generation of various power generation facilities, power consumption distribution in various regions, power demand of different industries, power price and power transmission loss, etc.;

[0059] In the embodiment of the present application, after the first data is obtained, the data is preprocessed; the data preprocessing includes data standardization, normalization, processing of missing values ​​and outliers;

[0060] In an optional embodiment, the data standardization method may be minimum-maximum standardization, Z-score standardization, or decimal calibration standardization; the normalization method may be Min-Max normalization, Z-score normalization, maximum value normalization, or vector normalization; the missing values ​​may be filled by mean interpolation, similar mean interpolation, multiple interpolation, or prediction based on a machine learning model; outlier processing may be identification and deletion, replacement with the median, mode, or model-based prediction value, or methods based on statistical rules.

[0061] In a possible embodiment, for missing values, the overall mean interpolation method of the feature is used to fill in the missing values, which is expressed as follows:

[0062]

[0063] in, is the padding value, x i is the observation of the data, and n is the number of observations.

[0064] In a possible embodiment, for outliers, the interquartile range of the box plot is used to first determine and then delete them. The determination of outliers is expressed by the formula:

[0065] Lower Bound=Q1-1.5×IQR

[0066] Upper Bound=Q3+1.5×IQR

[0067] Among them, Q1 and Q3 are the first and third quartiles of the data respectively, and IQR is the interquartile range. The calculation formula is as follows:

[0068] IQR=Q3-Q1

[0069] Based on the above method, the normal range of the data is determined, and the values ​​outside the normal range are identified as outliers, and the outliers are deleted.

[0070] In a possible embodiment, if the Min-Max normalization method is selected, it is expressed as follows:

[0071]

[0072] Among them, x is the original data and x′ is the normalized data.

[0073] In a possible embodiment, if the Z-score standardization method is selected, it is expressed as follows:

[0074]

[0075] Among them, x is the original data, μ is the mean, σ is the standard deviation, and z is the standardized data, that is, the input feature X.

[0076] In the embodiment of the present application, the above step S102 also includes the following sub-steps A1-A3;

[0077] In A1: a convolution kernel of a convolutional neural network is designed, each convolution kernel slides in time and space to calculate the dot product of the first data and the local area of ​​the convolution kernel to generate an original feature map;

[0078] In the embodiment of the present application, the dot product of the first data and the local area of ​​the convolution kernel is calculated, and the calculation formula is expressed as:

[0079]

[0080] Where X is the first input data, is the weight of the convolution kernel, k t and k s are the indices of the time dimension and space dimension respectively, and STConv(X) represents the original feature map.

[0081] In A2: a pooling operation is performed on the original feature map to reduce the spatial size and time length of the original feature map, and a weight value is calculated to obtain a first feature map;

[0082] In the embodiment of the present application, a pooling operation is performed on the original feature map, which is expressed as:

[0083] Pooling(X) = agg(X t:t+T,s:s+S )

[0084] Where agg represents the aggregation function in the pooling operation, X is the input feature, T and S represent the time range and spatial range of the feature respectively;

[0085] In the embodiment of the present application, the weight value of each feature is calculated, which is expressed as:

[0086] Attention Weights=σ(FC(Pooling(X)))

[0087] Among them, Pooling represents the global average pooling output, FC represents the fully connected layer in the pooling operation, σ represents the activation function in the pooling operation, and Attention Weights represents the generated weights.

[0088] In A3: the first feature map is input into the attention module, and the first feature map is weighted by using the weight value to adjust the importance of different positions in the first feature map to obtain the second feature map;

[0089] In an embodiment of the present application, because the importance of data at different positions in the input feature map varies, in order to more effectively capture and utilize these key information, the present invention selects a one-dimensional spatial attention module, and based on the obtained weight values, uses the one-dimensional spatial attention module to apply these weight values ​​to the first feature map, and performs weighted processing on each position in the first feature map; important positions will be given higher weights, so that they will receive more attention in subsequent calculations, while relatively unimportant positions will be weakened, thereby achieving fine adjustment of the importance of different positions in the input feature map.

[0090] In the embodiment of the present application, the above step S104 further includes steps B1-B3;

[0091] In B1: aligning the first predicted value and the second predicted value;

[0092] In B2: using an intelligent algorithm to quantify the matching degree between the first prediction value and the second prediction value, and adjusting the power system operation strategy according to the matching degree;

[0093] In an optional embodiment, the intelligent algorithm may be a support vector machine algorithm, a KLD algorithm, a convolutional neural network, a recurrent neural network, a random forest, or the like;

[0094] In the embodiment of the present application, the intelligent algorithm selects the KLD algorithm, and uses the KLD algorithm to evaluate the difference between the first prediction value and the second prediction value, which is expressed as:

[0095]

[0096] Among them, P(i) represents the probability value of the distribution at the i-th position, Q(i) represents the probability value of the distribution at the i-th position, and D KL (P||Q) is the difference measure with respect to .

[0097] In a possible embodiment, according to the matching degree, the power system operation strategy may be adjusted as follows:

[0098] ① In the case of high matching, the KLD value is less than 0.1:

[0099] In this case, the power system believes that the current power generation plan is sufficient to meet future consumption demand, so no major adjustments are needed. The system may maintain the current power generation level while monitoring the deviation between real-time data and predicted values ​​to make fine adjustments if necessary.

[0100] ② Medium matching, KLD value is between 0.1 and 0.3:

[0101] The power system triggers an early warning mechanism to alert operators to future supply and demand imbalances. Depending on the specific cause of the mismatch, for example, insufficient power generation or higher-than-expected consumption, the power system recommends increasing or decreasing specific power generation units, or adjusting the load on the consumer side through demand response measures.

[0102] ③ Low matching degree, KLD value is greater than 0.3:

[0103] The system automatically starts backup power generation units and adjusts power distribution in different areas in real time through smart grid technology. The system may send emergency notifications to consumers, requiring them to reduce non-essential loads, or initiate demand response plans to reduce power demand during peak hours.

[0104] It should be noted that the present invention adopts an intelligent algorithm to perform differential evaluation on the first predicted value and the second predicted value of power generation and consumption, and adjusts the power system operation strategy according to the matching degree, so as to realize the accurate prediction and efficient management of the power supply and demand balance; in the case of high matching degree, the system can maintain a stable power generation level and reduce unnecessary adjustments; in the case of medium matching degree, the triggering of the early warning mechanism enables the operator to respond to potential supply and demand imbalances in advance, and optimize power distribution by flexibly adjusting the power generation unit or demand response measures; in the case of low matching degree, the system can quickly start the backup resources, adjust the power distribution in real time, and send emergency notifications to the consumer side to ensure the rapid balance of power supply and demand, thereby significantly improving the overall efficiency and stability of the power system and effectively avoiding the adverse effects of power shortage or surplus.

[0105] The above is a schematic scheme of a dual sequence prediction method for power generation and consumption in this embodiment. It should be noted that the technical scheme of the system for dual sequence prediction of power generation and consumption and the technical scheme of the dual sequence prediction method for power generation and consumption belong to the same concept, and the details of the technical scheme of the dual sequence prediction system for power generation and consumption in this embodiment that are not described in detail can all be referred to the description of the technical scheme of the dual sequence prediction method for power generation and consumption.

[0106] The dual sequence prediction system for power generation and consumption in this embodiment includes:

[0107] A data acquisition module, used for acquiring first data of the power system;

[0108] A feature extraction module, configured to extract features of the first data using a convolutional neural network to obtain a first feature map, and perform weighted attention processing on the first feature map to obtain a second feature map;

[0109] The prediction module is used to perform transformation activation processing on the second feature map to obtain a first prediction value and a second prediction value, evaluate the difference between the first prediction value and the second prediction value, and then adjust the power system operation strategy.

[0110] This embodiment also provides a computing device, which is applicable to the case of dual sequence prediction of power generation and consumption, and includes:

[0111] A memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the dual sequence prediction method for power generation and consumption as proposed in the above embodiment.

[0112] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the dual sequence prediction method for power generation and consumption as proposed in the above embodiment.

[0113] The storage medium proposed in this embodiment and the dual sequence prediction method for realizing power generation and consumption proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0114] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.

Claims

1. A dual sequence forecasting method for power generation and consumption, characterized in that: include: Obtaining the first data of the power system; Extracting features of the first data using a convolutional neural network to obtain a first feature map, and performing weighted attention processing on the first feature map to obtain a second feature map; The second characteristic graph is transformed and activated to obtain a first prediction value and a second prediction value, and the difference between the first prediction value and the second prediction value is evaluated to adjust the power system operation strategy.

2. The method for dual sequence prediction of power generation and consumption according to claim 1, characterized in that: Extracting the features of the first data using a convolutional neural network includes: Design the convolution kernels of the convolutional neural network, each convolution kernel slides in time and space; Calculating the dot product of the first data and the local area of ​​the convolution kernel to generate an original feature map; A pooling operation is performed on the original feature map to reduce the spatial size and time length of the original feature map, and a weight value is calculated to obtain a first feature map.

3. The dual sequence prediction method for power generation and consumption according to claim 1 or 2, characterized in that: Performing weighted attention processing on the first feature map includes: The first feature map is input into the attention module, and the first feature map is weighted by using the weight value to adjust the importance of different positions in the first feature map to obtain a second feature map.

4. The method for dual sequence prediction of power generation and consumption according to claim 3, characterized in that: Evaluating the difference between the first predicted value and the second predicted value comprises, Performing data alignment on the first predicted value and the second predicted value; An intelligent algorithm is used to quantify the degree of match between the first prediction value and the second prediction value, and the power system operation strategy is adjusted according to the degree of match.

5. The dual sequence prediction method for power generation and consumption according to claim 4, characterized in that: The dot product of the first data and the local area of ​​the convolution kernel is calculated, and the calculation formula is expressed as: Where X is the first input data, is the weight of the convolution kernel, k t and k s are the indices of the time dimension and space dimension respectively, and STConv(X) represents the original feature map.

6. The method for dual sequence prediction of power generation and consumption according to claim 5, characterized in that: Calculation of weight values ​​includes, First, the original feature map is pooled, which is expressed as: Pooling(X)=agg(X t:t+T,s:s+S ) Where agg represents the aggregation function in the pooling operation, X is the input feature, T and S represent the time range and spatial range of the feature respectively; Then calculate the weight value of each feature, expressed as: Attention Weights=σ(FC(Pooling(X))) Among them, Pooling represents the global average pooling output, FC represents the fully connected layer in the pooling operation, σ represents the activation function in the pooling operation, and Attention Weights represents the generated weights.

7. The method for dual sequence prediction of power generation and consumption according to claim 6, characterized in that: Intelligent algorithms include: The KLD algorithm is used to evaluate the difference between the first predicted value and the second predicted value, expressed as: Among them, P(i) represents the probability value of the distribution at the i-th position, Q(i) represents the probability value of the distribution at the i-th position, and D KL (P||Q) is the difference measure with respect to .

8. A system for dual sequence prediction of power generation and consumption, characterized in that include, A data acquisition module, used for acquiring first data of the power system; A feature extraction module, configured to extract features of the first data using a convolutional neural network to obtain a first feature map, and perform weighted attention processing on the first feature map to obtain a second feature map; The prediction module is used to perform transformation activation processing on the second feature map to obtain a first prediction value and a second prediction value, evaluate the difference between the first prediction value and the second prediction value, and then adjust the power system operation strategy.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the dual sequence prediction method for power generation and consumption described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the dual sequence prediction method for power generation and consumption as claimed in any one of claims 1 to 7.

Images