Distributed photovoltaic local balance consumption method based on curve texture similarity matching

Abstract

The present invention discloses a distributed photovoltaic local balance absorption method based on curve texture similarity matching, which belongs to the field of photovoltaic power generation technology and includes the following steps: S1, converting the load curve of distributed photovoltaic and the load curve of flexible regulation resources into load images and distributed photovoltaic images respectively; S2, extracting high-frequency components and low-frequency components; S3, obtaining core vectors; S4, judging the similarity of distributed photovoltaic images and load images, and taking the time periods of distributed photovoltaic images and load images with similarities higher than a set threshold as the optimal absorption period; S5, adjusting the aggregation of load resources based on the optimal absorption period. The present invention adopts the above-mentioned distributed photovoltaic local balance absorption method based on curve texture similarity matching, and through data-driven modeling, it can realize the dynamic correction and continuous iterative update of the constructed model, which is not affected by external incentive conditions, economic factors, and climate factors, and has obvious advantages in the universality of the model.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to a distributed photovoltaic local balance consumption method based on curve texture similarity matching. Background Art

[0002] With the vigorous promotion of distributed photovoltaic construction, some areas have already experienced oversupply, resulting in frequent overloads in local areas. To solve this problem, existing technologies can effectively alleviate the problem of local photovoltaic reverse transmission by regulating the local consumption of resources on the demand side. For ordinary electricity, there are usually self-built photovoltaic systems. On the one hand, they can meet their own electricity needs by self-generation and self-use. At the same time, when the load is low, they can sell excess electricity online to obtain profits, realize local distributed photovoltaic point-to-point consumption, and promote local autonomous balance.

[0003] In view of the above situation, most of the existing technical solutions focus on how to accurately predict the output of distributed photovoltaics, improve the accuracy of local distributed photovoltaic predictions, and absorb excess electricity through refined dispatch management of the upper power grid. At the same time, considering the simultaneous existence of multiple roles such as distributed photovoltaics, power users, and aggregators, different subjects usually have independent demands. Most of the existing solutions rely on game theory to carry out source-load interaction optimization decisions, or provide quotation decision-making basis for different subjects. In addition, there are also existing technical solutions that consider guiding the energy flow of distributed photovoltaics through prices, so as to achieve local self-balancing of distributed photovoltaics.

[0004] It can be seen that most of the existing distributed photovoltaic absorption solutions rely on the uncertainty information of multi-party games to carry out relevant research, thereby building a competitive mechanism among multiple subjects, reducing the imbalance between distributed photovoltaic producers and consumers, and thus alleviating the pressure of distributed photovoltaic absorption.

[0005] At the same time, it can be seen that demand-side resources, as a flexible and adjustable resource, have the ability to absorb distributed photovoltaics locally, but this requires accurate prediction and matching of distributed photovoltaic output and the interactive characteristics of adjustable loads. However, the current randomness of distributed photovoltaic output coupled with the uncertainty of the operation of flexible and adjustable resources makes it very difficult to match the two, and it is difficult to accurately call on demand-side resources to achieve a precise balance between distributed photovoltaics and loads in the local area. Summary of the invention

[0006] To solve the above problems, the present invention provides a distributed photovoltaic local balance absorption method based on curve texture similarity matching. Through data-driven modeling, the constructed model can be dynamically corrected and continuously iteratively updated. It is not affected by external incentive conditions, economic factors, and climate factors, and has obvious advantages in the universality of the model.

[0007] To achieve the above object, the present invention provides a distributed photovoltaic local balance consumption method based on curve texture similarity matching, comprising the following steps:

[0008] S1. Collect the load curve of distributed photovoltaics and the load curve of flexible regulation resources, and convert the load curve of distributed photovoltaics and the load curve of flexible regulation resources into a load image and a distributed photovoltaic image respectively;

[0009] S2, using image layering calculation to perform texture analysis on load images and distributed photovoltaic images, and extract high-frequency components and low-frequency components;

[0010] S3, extracting the load change trend based on the historical operation statistical data of distributed photovoltaics, combining the extracted high-frequency components, comparing the texture similarity of the distributed photovoltaic image and the load image, and obtaining the core vector;

[0011] S4, judging the similarity between the distributed photovoltaic image and the load image based on the core vector, and taking the time period of the distributed photovoltaic image and the load image with a similarity higher than a set threshold as the optimal consumption time period;

[0012] S5. Adjust load resource aggregation based on the optimal consumption period.

[0013] Preferably, the distributed photovoltaic image conversion step described in step S1 is as follows: reading meter data of the distributed photovoltaic gateway meter, obtaining the output curve of the distributed photovoltaic 15-minute frequency, and performing vector processing to obtain the distributed photovoltaic output image;

[0014] The load image conversion steps described in step S1 are as follows: reading the meter installed on the demand side or the user's gateway meter, extracting the load curve with a frequency of 15 minutes, and performing vectorization processing to obtain a load image.

[0015] Preferably, step S2 specifically includes the following steps:

[0016] S21, unifying the number of pixels and dimensions of the load image and the distributed photovoltaic image;

[0017] S22, converting the load image and the distributed photovoltaic image into a two-dimensional spectrum matrix through downconvolution operation to obtain a load spectrum;

[0018] S23, performing sampling and smoothing processing on the load spectrum to stratify the load spectrum;

[0019] S24. Perform texture analysis on the load image and the distributed photovoltaic image to extract high-frequency components and low-frequency components.

[0020] Preferably, the downconvolution operation formula described in step S22 is as follows:

[0021]

[0022] Where, L g (x, y, σ) represents the load image or distributed photovoltaic image; x, y represent the position number of the xth row and yth column in the two-dimensional atlas matrix respectively; σ represents the scale parameter of the load image or distributed photovoltaic image, and G(x, y, σ) represents the Gaussian kernel; represents the convolution operation; P(x,y) represents the two-dimensional graph matrix;

[0023] Among them, the calculation formula of Gaussian kernel G(x,y,σ) is as follows:

[0024]

[0025] Wherein, m represents the texture extraction range of the load image or distributed photovoltaic image on the x-axis; n represents the texture extraction range of the load image or distributed photovoltaic image on the y-axis.

[0026] Preferably, step S24 specifically includes the following steps:

[0027] S241. Set the number of groups O corresponding to the load image or distributed photovoltaic image to:

[0028] O=[Log2min(M,N)]-3

[0029] Where M and N represent the maximum number of rows and columns of the two-dimensional atlas matrix respectively;

[0030] S242, threshold difference processing:

[0031] Assuming that the load image or distributed photovoltaic image of the i-th layer is the S layer, the differential load map is the S-1 layer, and the S layers of each group of load maps are numbered as the 0th layer, the 1st layer, the nth layer, the n+1th layer, and the n+2th layer in sequence; and the fuzzy coefficient σ corresponding to the convolution kernel G(x, y, σ) is selected (i,j) (o,r)=σ02 o+r / n , where o represents the grouping number of each layer of the corresponding load spectrum, o∈[0,1,...,o-1]; r represents the layering number of the corresponding load spectrum, r∈[0,1,...,n+2];

[0032] Perform differential calculation on the two-dimensional atlas matrix between different layers in each group to obtain the differential image of the i-th layer and the j-th group. as follows:

[0033]

[0034] In the formula, Represents the elements in the two-dimensional atlas matrix of the i-th layer and the j-th group; Represents the element in the two-dimensional atlas matrix of the i-th layer and the j-1-th group;

[0035] S243, will Considered as the difference between different convolution kernels, select the i-th layer and the j-th group of two-dimensional atlas matrix The element f(x,y) in , calculates its perturbation component f(x+Δx,y+Δy) as follows:

[0036] When x=x in the two-dimensional atlas matrix i ,y=y i When the position is specific, select equally spaced microelements Δx=Δy=h;

[0037]

[0038] Where f(x,y) represents the element in the x-th row and y-th column of the two-dimensional map matrix; f(x+Δx,y+Δy) represents the perturbation component of f(x,y) at the point (x,y); Δx and Δy represent the discretized microelement in the two-dimensional map matrix respectively; f(x i ,y i ) means that in the xth i row, y i The two-dimensional map matrix element of the column; f(x i +h,y i ), f(x i -h,y i ) represent f(x i ,y i ) deviates from +h and -h in the x direction; f(x i ,y i +h), f(x i ,y i -h) respectively represent f(x i ,y i ) deviates from +h and -h in the y direction; h represents x=x in the two-dimensional atlas matrix i ,y=y i Equally spaced microelements at positions;

[0039] S244, filtering processing:

[0040]

[0041] In the formula, H(x i ,y i ) represents the two-dimensional map matrix at the xth i row, y i The second-order derivative operator matrix of the columns; f(x i ,yi ) represents the two-dimensional map matrix at the xth i row, y i The value of the column;

[0042] And determine whether to keep it according to the filtering result: if [H(x i ,y i ) 1,1 +H(x i ,y i ) 2,2 ] 2 / {[H(x i ,y i ) 1,1 H(x i ,y i ) 2,2 ]-[H(x i ,y i ) 1,2 ] 2}<(γ+1) 2 / γ is retained, otherwise it is discarded; where H(x,y) 1,1 ,H(x,y) 2,2 ,H(x,y) 1,2 Respectively represent H(x i ,y i )The elements of the 1st row and 1st column, the 2nd row and 2nd column, and the 1st row and 2nd column of the matrix; γ represents the decision threshold.

[0043] Preferably, in step S244, the maximum load P is selected max 1‰ is used as the judgment threshold, and the data below the judgment threshold is filtered out as disturbance data.

[0044] Preferably, in step S5, in order to obtain profits from the response projects participating in photovoltaic consumption, the load aggregator quotes to the power company to participate in the response. After the power company has cleared the payment, it signs a corresponding contract with the load aggregator. For the response amount within the scope of the contract, the power company conducts compensation transactions according to the response amount in accordance with the contract. The power company will not compensate for the excess response that exceeds the response requirements, and will punish the response deviation that does not meet the requirements.

[0045] Preferably, in step S5, the incentives provided by the power company to each response node of the load aggregator are set as:

[0046]

[0047] In the formula, C earn represents the revenue of the load aggregator at the response node; t1 and t2 represent the scheduling start time and scheduling end time respectively; P a Indicates the settlement power; represents the clearing price of the load aggregator of the i-th response node; ΔH represents the response time interval; represents the actual response value of the aggregator at the response node at time t; P b Indicates the clearing power in response to the contract signed by this node.

[0048] Preferably, the penalty in step S5 is a response deviation penalty or a response delay penalty;

[0049] Where, the response deviation penalty C s The calculation formula is as follows:

[0050]

[0051] In the formula, α s Represents the response deviation penalty factor; M1 represents the deviation penalty factor when the response amount is lower than the set lower limit; M2 represents the deviation penalty factor when the response amount is higher than the set upper limit; Indicates the agreed response completion amount; ν1 represents the negative deviation adjustment coefficient, and ν2 represents the positive deviation adjustment coefficient; according to the Implementation Rules of Jiangsu Province Electricity Demand Response, currently ν1=0.8, ν2=1.2, M1 is recommended to be 0.9, and M2 is recommended to be 0.95.

[0052] Response delay penalty C t The calculation formula is as follows:

[0053] C t =β(t ε -t1)

[0054] Where: β represents the delay penalty coefficient; t ε Indicates that the response power reaches The time taken, ε represents the delay coefficient.

[0055] The present invention has the following beneficial effects:

[0056] It weakens the need for traditional physical modeling, does not require accurate construction of user behavior models, and is not affected by complex equipment working mechanisms. It has better versatility, and through data-driven modeling, it can achieve dynamic correction and continuous iterative updating of the built model. It is not affected by external incentive conditions, economic factors, and climate factors, and has obvious advantages in the universality of the model.

[0057] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flow chart of the distributed photovoltaic local balance consumption method based on curve texture similarity matching of the present invention;

[0059] Figure 2 It is a graph matching result diagram of the simulation experiment of the present invention;

[0060] Figure 3 This is a diagram showing the matching results of the load curve of the simulation experiment of the present invention and the local photovoltaic power generation. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical scheme and advantages disclosed in the embodiments of the present invention clearer, the embodiments of the present invention are further described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not intended to limit the embodiments of the present invention. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions.

[0062] It should be noted that the terms "include", "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or devices.

[0063] Like reference numerals and letters denote similar items in the following drawings, and thus, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0064] like Figure 1 As shown, the distributed photovoltaic local balance consumption method based on curve texture similarity matching includes the following steps:

[0065] S1. Collect the load curve of distributed photovoltaics and the load curve of flexible regulation resources, and convert the load curve of distributed photovoltaics and the load curve of flexible regulation resources into load images and distributed photovoltaic images respectively; among which, the distributed photovoltaic output image corresponds to the aggregated output image within the area where it is located, while the load image is based on users with independent metering and control equipment.

[0066] The distributed photovoltaic image conversion steps described in step S1 are as follows: read the meter data of the distributed photovoltaic gateway meter, obtain the output curve of the distributed photovoltaic 15-minute frequency, and perform vector processing to obtain the distributed photovoltaic output image; the load image conversion steps described in step S1 are as follows: read the meter installed on the demand side or the user's gateway meter, extract the load curve with a frequency of 15 minutes, and perform vector processing to obtain the load image. The load image and distributed photovoltaic image of this embodiment are data blocks of 96 points*365 days.

[0067] It should be noted that the present invention is also applicable to the processing method of measurement data with more refined data, and it only needs to expand the image size of distributed photovoltaics; similarly, for loads with higher frequency measurement data, it only needs to expand the load image size to meet the requirements.

[0068] S2, using image layering calculation to perform texture analysis on load images and distributed photovoltaic images, and extract high-frequency components and low-frequency components;

[0069] Step S2 specifically includes the following steps:

[0070] S21, unifying the number of pixels and dimensions of the load image and the distributed photovoltaic image;

[0071] S22, converting the load image and the distributed photovoltaic image into a two-dimensional spectrum matrix through downconvolution operation to obtain a load spectrum;

[0072] The downconvolution operation formula described in step S22 is as follows:

[0073]

[0074] Where, L g (x, y, σ) represents the load image or distributed photovoltaic image; x, y represent the position number of the xth row and yth column in the two-dimensional atlas matrix respectively; σ represents the scale parameter of the load image or distributed photovoltaic image, and G(x, y, σ) represents the Gaussian kernel; represents the convolution operation; P(x,y) represents the two-dimensional graph matrix;

[0075] Among them, the calculation formula of Gaussian kernel G(x,y,σ) is as follows:

[0076]

[0077] Wherein, m represents the texture extraction range of the load image or distributed photovoltaic image on the x-axis; n represents the texture extraction range of the load image or distributed photovoltaic image on the y-axis.

[0078] S23, performing sampling and smoothing processing on the load spectrum to stratify the load spectrum;

[0079] S24. Perform texture analysis on the load image and the distributed photovoltaic image to extract high-frequency components and low-frequency components.

[0080] Step S24 specifically includes the following steps:

[0081] S241. Set the number of groups O corresponding to the load image or distributed photovoltaic image to:

[0082] O=[Log2 min(M,N)]-3

[0083] Where M and N represent the maximum number of rows and columns of the two-dimensional atlas matrix respectively;

[0084] S242, threshold difference processing:

[0085] Assuming that the load image or distributed photovoltaic image of the i-th layer is the S layer, the differential load map is the S-1 layer, and the two edge layers cannot solve the extreme values, so they do not participate in the calculation. Therefore, only the S-3 layer in the differential load map needs to extract features. For the convenience of representation, the S layers of each group of load maps are numbered as the 0th layer, the 1st layer, the nth layer, the n+1th layer, and the n+2th layer in sequence; and the fuzzy coefficient σ corresponding to the convolution kernel G(x, y, σ) is selected. (i,j) (o,r)=σ02 o+r / n , where o represents the grouping number of each layer of the corresponding load spectrum, o∈[0,1,...,o-1]; r represents the layering number of the corresponding load spectrum, r∈[0,1,...,n+2];

[0086] Perform differential calculation on the two-dimensional atlas matrix between different layers in each group to obtain the differential image of the i-th layer and the j-th group. as follows:

[0087]

[0088] In the formula, Represents the elements in the two-dimensional atlas matrix of the i-th layer and the j-th group; Represents the element in the two-dimensional atlas matrix of the i-th layer and the j-1-th group;

[0089] In step S242, the maximum load P is selected. max 1‰ is used as the threshold, and the data below the threshold is filtered out as disturbance data.

[0090] S243, will Considered as the difference between different convolution kernels, select the i-th layer and the j-th group of two-dimensional atlas matrix The element f(x,y) in , calculates its perturbation component f(x+Δx,y+Δy) as follows:

[0091]

[0092] To simplify the operation, when determining x=x in the two-dimensional map matrix i ,y=y i When the position is specific, the equally spaced microelement Δx=Δy=h can be selected;

[0093]

[0094]

[0095] Where f(x,y) represents the element in the x-th row and y-th column of the two-dimensional map matrix; f(x+Δx,y+Δy) represents the perturbation component of f(x,y) at the point (x,y); Δx and Δy represent the discretized microelement in the two-dimensional map matrix respectively; f(x i ,y i ) means that in the xth i row, y i The two-dimensional map matrix element of the column; f(x i +h,y i ), f(x i -h,y i ) represent f(x i ,y i ) deviates from +h and -h in the x direction; f(x i ,y i +h), f(x i ,y i -h) respectively represent f(x i ,y i ) deviates from +h and -h in the y direction; h represents x=x in the two-dimensional atlas matrix i ,y=y i Equally spaced microelements at positions;

[0096] S244, filtering processing:

[0097]

[0098] In the formula, H(x i ,y i ) represents the two-dimensional map matrix at the xth i row, y i The second-order derivative operator matrix of the columns; f(x i ,y i ) represents the two-dimensional map matrix at the xth i row, y i The value of the column;

[0099] And determine whether to keep it according to the filtering result: if [H(xi ,y i ) 1,1 +H(x i ,y i ) 2,2 ] 2 / {[H(x i ,y i ) 1,1 H(x i ,y i ) 2,2 ]-[H(x i ,y i ) 1,2 ] 2}<(γ+1) 2 / γ is retained, otherwise it is discarded; where H(x,y) 1,1 ,H(x,y) 2,2 ,H(x,y) 1,2 Respectively represent H(x i ,y i )The elements of the 1st row and 1st column, the 2nd row and 2nd column, and the 1st row and 2nd column of the matrix; γ represents the decision threshold.

[0100] In step S244, the maximum load P is selected. max 1‰ is used as the judgment threshold, and the data below the judgment threshold is filtered out as disturbance data.

[0101] S3, judging the similarity between the distributed photovoltaic image and the load image based on the core vector, and taking the time period of the distributed photovoltaic image and the load image with a similarity higher than a set threshold as the optimal consumption time period;

[0102] S4. Adjust load resource aggregation based on the optimal consumption period.

[0103] The power company monitors the power grid flow information and other information through the smart devices installed in the power grid, and obtains information such as distributed photovoltaics and loads. When the demand side needs to participate in the distribution network regulation due to sufficient sunlight and the distributed photovoltaic output is excessive, the load aggregator is used as an intermediary to control users to participate in the photovoltaic consumption. The load aggregator collects the information of each air-conditioning user through advanced network information technology, and classifies the information of users who are willing to participate in temperature adjustment and those who are willing to participate in start-up and stop. When the load aggregator participates in the regulation of the power company, the load aggregator achieves the purpose of changing the air-conditioning load aggregation power by adjusting the temperature or starting and stopping the air-conditioning of each air-conditioning user. The power company and the load aggregator, as well as the load aggregator and the user, need to sign a bilateral contract to determine information such as compensation for participating in demand response.

[0104] Therefore, in step S4, in order to obtain profits from the response projects participating in PV consumption, the load aggregators quote to the power companies to participate in the response. After the power companies have cleared their accounts, they sign corresponding contracts with the load aggregators. For the response quantity within the scope of the contract, the power companies conduct compensation transactions according to the response quantity in accordance with the contract. The power companies will not compensate for the excess response that exceeds the response requirements, and will punish the response deviation that does not meet the requirements.

[0105] In step S4, the incentives of the power company to each response node of the load aggregator are set as:

[0106]

[0107] In the formula, C earn represents the revenue of the load aggregator at the response node; t1 and t2 represent the scheduling start time and scheduling end time respectively; P a Indicates the settlement power; represents the clearing price of the load aggregator of the i-th response node; ΔH represents the response time interval; represents the actual response value of the aggregator at the response node at time t; P b Indicates the clearing power in response to the contract signed by this node.

[0108] The penalty in step S4 is a response deviation penalty (the response deviation penalty is set to ensure that the response power of the load aggregator is stable near the required value during the dispatch period) or a response delay penalty;

[0109] Where, the response deviation penalty C s The calculation formula is as follows:

[0110]

[0111] In the formula, α s Represents the response deviation penalty factor; M1 represents the deviation penalty factor when the response amount is lower than the set lower limit; M2 represents the deviation penalty factor when the response amount is higher than the set upper limit; Indicates the agreed response completion amount; ν1 represents the negative deviation adjustment coefficient, and ν2 represents the positive deviation adjustment coefficient; according to the Implementation Rules of Jiangsu Province Electricity Demand Response, currently ν1=0.8, ν2=1.2, M1 is recommended to be 0.9, and M2 is recommended to be 0.95.

[0112] Response delay penalty C t The calculation formula is as follows:

[0113] C t =β(t ε -t1)

[0114] Where: β represents the delay penalty coefficient; t εIndicates that the response power reaches The time taken and ε represent the delay coefficient, which is determined by the power company's response effect requirements.

[0115] Simulation experiment

[0116] The annual 8760-point load curve and regional distributed photovoltaic data were selected, and the two-dimensional load map and distributed photovoltaic map were constructed by the method of the present invention for matching. The similar operation rules can be obtained by combining the typical day. Figure 2 and Figure 3 The results shown verify the effectiveness of the present invention.

[0117] Table 1 Some simulation experiment results

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] Therefore, the present invention adopts the above-mentioned distributed photovoltaic local balance absorption method based on curve texture similarity matching, which can realize dynamic correction and continuous iterative update of the constructed model through data-driven modeling. It is not affected by external incentive conditions, economic factors, and climate factors, and has obvious advantages in the universality of the model.

[0124] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0125] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0126] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A distributed photovoltaic local balance consumption method based on curve texture similarity matching, characterized by: The following steps are involved: S1. Collect the load curve of distributed photovoltaics and the load curve of flexible regulation resources, and convert the load curve of distributed photovoltaics and the load curve of flexible regulation resources into a load image and a distributed photovoltaic image respectively; S2, using image layering calculation to perform texture analysis on load images and distributed photovoltaic images, and extract high-frequency components and low-frequency components; S3, extracting the load change trend based on the historical operation statistical data of distributed photovoltaics, combining the extracted high-frequency components, comparing the texture similarity of the distributed photovoltaic image and the load image, and obtaining the core vector; S4, judging the similarity between the distributed photovoltaic image and the load image based on the core vector, and taking the time period of the distributed photovoltaic image and the load image with a similarity higher than a set threshold as the optimal consumption time period; S5. Regulate load resource aggregation based on the optimal consumption period; In step S5, the load aggregator bids the power company to participate in the response project in order to obtain profits from participating in the photovoltaic consumption response project. After the power company clears the contract, it signs a corresponding contract with the load aggregator. For the response amount within the scope of the contract, the power company will conduct compensation transactions according to the response amount in accordance with the contract. For the excess response part exceeding the response requirement, the power company will not compensate, and will punish the response deviation that does not meet the requirement; In step S5, the incentives of the power company to each response node of the load aggregator are set as: In the formula, represents the revenue of the load aggregator at the response node; and Respectively represent the scheduling start time and scheduling end time; Indicates the settlement power; Indicates The response node load aggregator clearing price; Indicates the response time interval; Indicates the response node The actual response value of the aggregator at the moment; Indicates the clearing power in response to the contract signed by this node.

2. The distributed photovoltaic local balance consumption method based on curve texture similarity matching according to claim 1 is characterized by: The distributed photovoltaic image conversion steps described in step S1 are as follows: reading the meter data of the distributed photovoltaic gateway meter, obtaining the output curve of the distributed photovoltaic 15-minute frequency, and performing vector processing to obtain the distributed photovoltaic output image; The load image conversion steps described in step S1 are as follows: reading the meter installed on the demand side or the user's gateway meter, extracting the load curve with a frequency of 15 minutes, and performing vectorization processing to obtain a load image.

3. The distributed photovoltaic local balance consumption method based on curve texture similarity matching according to claim 1 is characterized by: Step S2 specifically includes the following steps: S21, unifying the number of pixels and dimensions of the load image and the distributed photovoltaic image; S22, converting the load image and the distributed photovoltaic image into a two-dimensional spectrum matrix through downconvolution operation to obtain a load spectrum; S23, performing sampling and smoothing processing on the load spectrum to stratify the load spectrum; S24. Perform texture analysis on the load image and the distributed photovoltaic image to extract high-frequency components and low-frequency components.

4. The distributed photovoltaic local balance consumption method based on curve texture similarity matching according to claim 3 is characterized by: The downconvolution operation formula described in step S22 is as follows: In the formula, Represents load images or distributed photovoltaic images; The two-dimensional spectrum matrix represents the load or distributed photovoltaic Row, No. The position number of the column; Represents the scale parameter of the load image or distributed photovoltaic image, represents Gaussian kernel; Represents the convolution operation; Represents a two-dimensional map or distributed photovoltaic matrix; Among them, the Gaussian kernel The calculation formula is as follows: In the formula, Indicates the texture extraction range of the load image or distributed photovoltaic image on the x-axis; Indicates the texture extraction range of the load image or distributed photovoltaic image on the y-axis.

5. The distributed photovoltaic local balance consumption method based on curve texture similarity matching according to claim 4 is characterized in that: Step S24 specifically includes the following steps: S241. Setting the number of groups corresponding to the load image or distributed photovoltaic image for: In the formula, and Respectively represent the maximum serial number of the row and column of the two-dimensional atlas matrix; S242, threshold difference processing: Assume The load image of the layer or the distributed photovoltaic image is layer, the difference spectrum is layer, and each set of atlas The layers are numbered as layer 0, layer 1, layer n, layer n+1, layer n+2, and the convolution kernel is selected. The corresponding fuzzy coefficient ,in Indicates the grouping sequence number of each layer of the corresponding graph, ; Indicates the layer number of the corresponding atlas, ; The difference calculation is performed on the two-dimensional atlas matrix between different layers in each group, and the first Layer, Difference image of the group as follows: In the formula, Indicates Layer, A two-dimensional map of a group or an element in a distributed photovoltaic matrix; Indicates Layer, Elements in the two-dimensional atlas matrix of the group; S243, will is regarded as the difference between different convolution kernels, and the Layer, Group 2D Spectrum Matrix Elements in , calculate its disturbance component as follows: In the formula, is the order corresponding to the disturbance component; Take N=3 to get When determining the two-dimensional map matrix , When the specific location is selected, select the equally spaced microelements ; In the formula, Indicates Row, No. Column elements in the two-dimensional atlas matrix; express The disturbance component at the point (x,y); , They represent the discretized elements in the two-dimensional atlas matrix respectively; Indicated in Row, No. Columns of the two-dimensional atlas matrix elements; , Respectively The values ​​of the deviations +h and -h in the x direction; , Respectively The values ​​of deviations from +h and -h in the y direction; Represented in a two-dimensional atlas matrix , Equally spaced microelements at positions; S244, filtering processing: In the formula, Represents the two-dimensional atlas matrix in Row, No. The second-order derivative operator matrix of the columns; Represents the two-dimensional atlas matrix in Row, No. The value of the column; And determine whether to keep it based on the filtering result: if If it is not, it is retained; otherwise, it is discarded; , , Respectively The elements in row 1, column 1, row 2, column 2, and row 1, column 2 of the matrix; Indicates the decision threshold.

6. The distributed photovoltaic local balance consumption method based on curve texture similarity matching according to claim 5 is characterized by: In step S244, the maximum load is selected 1‰ is used as the judgment threshold, and the data below the judgment threshold is filtered out as disturbance data.

7. The distributed photovoltaic local balance consumption method based on curve texture similarity matching according to claim 1 is characterized by: The penalty in step S5 is a response deviation penalty or a response delay penalty; Among them, the response deviation penalty The calculation formula is as follows: In the formula, represents the response deviation penalty factor; Indicates the deviation penalty factor when the response amount is lower than the set lower limit; Indicates the deviation penalty factor when the response amount is higher than the set upper limit; Indicates the agreed response completion amount; represents the negative deviation adjustment coefficient, represents the positive deviation adjustment factor; Response delay penalty The calculation formula is as follows: Where: represents the delay penalty coefficient; Indicates that the response power reaches Time taken, Represents the delay factor.

Images