Method and system for generating hour-by-hour load sequence based on month-by-month peak and valley electricity utilization records

Abstract

The invention discloses an hour-by-hour load sequence generation method and system based on a month-by-month peak and valley power utilization record, and the method comprises the steps: mapping reference sample data into a high-dimensional vector data space of the month-by-month peak and valley power utilization record of a to-be-constructed target through a mapping function; normalizing all vectors in the high-dimensional vector data space to obtain a normalized high-dimensional vector data space; calculating the distance between the monthly peak-valley electricity utilization record and each reference sample data, selecting a neighbor set according to a threshold value, and converting the distance in the neighbor set into a normalized distance weight value; normalizing the load data in the neighbor set to obtain a load data sequence, decomposing the load data sequence into the sum of a plurality of subsequences, and dividing the subsequences into random subsequences and stable subsequences; and adding the two types of subsequences into a normalized hour-by-hour load sequence of the to-be-constructed target, and carrying out reverse normalization on the normalized hour-by-hour load sequence of the to-be-constructed target to obtain an hour-by-hour load sequence of the to-be-constructed target.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy microgrid construction, and specifically relates to a method and system for generating hourly load sequences based on monthly peak and valley electricity consumption records. Background Art

[0002] In modern power system management and energy planning, accurately grasping the power load curve is of great significance for improving energy efficiency, optimizing resource allocation and realizing intelligent dispatching. Especially in the fields of energy storage equipment investment, intelligent energy system planning and power dispatching, the importance of hourly demand load data throughout the year is self-evident. However, most power consumption organizations currently face a common problem: lack of detailed hourly power consumption record data.

[0003] Currently, most electricity-consuming organizations can only obtain monthly electricity bill information, which usually only contains the total electricity consumption of the month under different electricity price tiers. Although this coarse-grained data meets basic billing needs, it cannot provide sufficient support for more in-depth energy analysis and decision-making. In particular, the limitations of this data are particularly prominent when evaluating energy storage equipment investments, planning smart energy systems, or formulating refined power dispatch strategies.

[0004] On the other hand, some large electricity-consuming organizations or power companies may have more detailed electricity consumption data, including hourly load records, which enable accurate analysis and decision-making, but due to various reasons such as privacy protection or commercial confidentiality, these detailed data are usually not disclosed or shared.

[0005] Therefore, the industry urgently needs a method that can construct a relatively accurate hourly demand load curve for the whole year under limited data conditions. This method should be able to make full use of the existing monthly electricity consumption data, refer to the load patterns of other similar organizations, and derive the hourly load curve of the target electricity consumption organization through reasonable mathematical models and algorithms. Summary of the invention

[0006] The purpose of the present invention is to overcome the problem that a simple construction method lacks data fluctuation information when the hourly electricity consumption record data is insufficient, and proposes a method and system for generating hourly load sequences based on monthly peak and valley electricity consumption records.

[0007] In order to achieve the above object, the present invention adopts the following technical scheme: The reference sample data is mapped to the high-dimensional vector data space of the monthly peak-valley electricity consumption records of the target to be constructed by using a mapping function, and all vectors in the high-dimensional vector data space are normalized to obtain a normalized high-dimensional vector data space; The distance between the monthly peak and valley electricity consumption records and each reference sample data is calculated in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, the distance in the nearest neighbor set is converted into a normalized distance weight value; After normalizing each load data in the neighbor set, the load data sequence is obtained, and the load data sequence is decomposed and converted into the sum of several subsequences. The subsequences are divided into random subsequences and stable subsequences based on information entropy; After summarizing the random subsequences and the stable subsequences, the two types of subsequences are added together to form the normalized hourly load sequence of the target to be constructed, and the normalized hourly load sequence of the target to be constructed is denormalized to obtain the hourly load sequence of the target to be constructed.

[0008] Furthermore, the use of a mapping function to map the reference sample data to the high-dimensional vector data space of the monthly peak and valley electricity consumption records of the target to be constructed includes: The monthly peak-valley electricity consumption records of the target to be constructed are represented as high-dimensional vectors, and the reference sample data are mapped to the data space of the high-dimensional vector representing the monthly peak-valley electricity consumption records of the target to be constructed through a mapping function using the annual tiered electricity price pricing information. The reference sample data and the monthly peak-valley electricity consumption records of the target to be constructed are in the same dimension; The annual tiered electricity price pricing information and the monthly peak and valley electricity consumption records are as follows:

[0009]

[0010] in, To record the peaks and valleys of each month, Subscript i Indicates that the element is in the first i Hour, The element x represents a certain attribute value corresponding to the hour, and there are 8760 hours in a year; Subscript i Indicates the month, Superscript price(k) Indicates the first k Class electricity price, L represents the total electricity consumption corresponding to the electricity price in that month, and there are 12 months in a year; The mapping function is shown below:

[0011] in, To record the peaks and valleys of each month, The annual electricity price is For the whole year load; The mapping is specifically as follows: traversal month by month , get thei The countable set elements in the monthly electricity price sequence { price}; Based on countable set elements { price} i Filter the monthly electricity price sequence to obtain the i Monthly electricity price sequence and specific elements in countable sets price(k) Equal sequence indexes ;from Get the index with sequence index The load data of the same position are summed up to get , iteratively execute the above process to obtain Different, each load sequence The corresponding monthly peak and valley records V。

[0012] Furthermore, the normalization method is as follows:

[0013] in, is the normalization method, For vector V The scalar value of the element with the largest value in , The maximum value of is 1. The minimum value of is not necessarily 0; , With function The interchangeability between them is as follows: .

[0015] Furthermore, the distance between the monthly peak and valley electricity consumption records and each reference sample data is calculated in the normalized high-dimensional vector data space, and after selecting a neighbor set according to a threshold, the distance in the neighbor set is converted into a normalized distance weight value, including: Select reference samples whose distance to the target to be constructed is less than the threshold to form a sample vector set, as shown below:

[0016] The distance set corresponding to the sample vector set is as follows:

[0017] The neighbor set corresponding to the distance set is as follows:

[0018] The distance in the neighbor set is converted into a normalized distance weight value. Use the radial basis function rbf and normalize it as follows:

[0019] in, is a sample vector set, is the distance set of the sample vector set, is the normalized weight at the target to be constructed, Represents two load sequences i,j The distance between them is the distance between their normalized high-dimensional vectors. The peak and valley electricity consumption records of the target to be constructed, is the high-dimensional vector of each reference sample data, is the distance set between the target to be constructed and each reference sample, and threshold is the threshold.

[0020] Furthermore, the load data sequence is obtained by normalizing each load data in the neighboring set, and the load data sequence is decomposed into the sum of several subsequences as follows: for any neighboring set The corresponding normalized load sequence is decomposed into the sum of several subsequences using the empirical mode decomposition method; The decomposition using the empirical mode decomposition method is as follows:

[0021]

[0022] in, is the normalized load sequence, EMD is the empirical mode decomposition method, For vector V The scalar value of the element with the largest value in , For the whole year load; The method of distinguishing subsequences into random subsequences and stable subsequences based on information entropy is specifically as follows: Perform discrete Fourier transform on each subsequence, calculate the information entropy of the positive half frequency sequence obtained after each discrete Fourier transform, sort each subsequence in ascending order according to the information entropy of the positive half frequency sequence, and then divide the subsequence set into random subsequences according to the specified threshold With stable subsequence ; Random Subsequence is the subset of subsequences whose information entropy is less than the threshold sum; Stable subsequence is the subset of subsequences containing information entropy greater than the threshold sum.

[0023] Furthermore, after aggregating the random subsequence and the stable subsequence, the two types of subsequences are added together to form a normalized hourly load sequence of the target to be constructed as follows:

[0024] in, is the normalized hourly load sequence of the target to be constructed, is a random subsequence, is the stable subsequence of the target to be constructed; The stable subsequence of the target to be constructed is as follows:

[0025] The random subsequence of the target to be constructed is as follows:

[0026]

[0027] in, The function is a random sequence construction method, and param represents the statistical characteristic parameters.

[0028] Furthermore, the normalized hourly load sequence of the target to be constructed is denormalized to obtain the hourly load sequence of the target to be constructed as shown in the following formula:

[0029] in, is the hourly load sequence of the target to be built, is the normalized hourly load sequence of the target to be constructed, It is the inverse operation of the normalization method.

[0030] The hourly load sequence generation system based on monthly peak and valley power consumption records includes: A mapping module is used to map the reference sample data to the high-dimensional vector data space of the monthly peak and valley electricity consumption records of the target to be constructed by using a mapping function, and normalize all vectors in the high-dimensional vector data space to obtain a normalized high-dimensional vector data space; The distance weight normalization module is used to calculate the distance between the monthly peak and valley electricity consumption records and each reference sample data in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, convert the distance in the nearest neighbor set into a normalized distance weight value; A subsequence distinguishing module is used to obtain a load data sequence after normalizing each load data in the neighbor set, decompose the load data sequence into the sum of several subsequences, and distinguish the subsequences into random subsequences and stable subsequences based on information entropy; An hourly load sequence module is obtained, which is used to summarize the random subsequences and the stable subsequences and then add the two types of subsequences to form a normalized hourly load sequence of the target to be constructed, and denormalize the normalized hourly load sequence of the target to be constructed to obtain the hourly load sequence of the target to be constructed.

[0031] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for generating hourly load sequences based on monthly peak and valley electricity consumption records is implemented.

[0032] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for generating hourly load sequences based on monthly peak and valley electricity consumption records is implemented.

[0033] Compared with the prior art, the present invention has the following beneficial technical effects: The hourly load sequence generation method based on monthly peak and valley electricity consumption records proposed in the present invention can construct a relatively accurate hourly demand load curve for the whole year under limited data conditions, and can make full use of the existing monthly electricity consumption data. At the same time, it refers to the load patterns of other similar organizations, and derives the hourly load curve of the target electricity consumption organization through reasonable mathematical models and algorithms. The present invention has strong practicality. The completed load sequence generation method takes into account both reference sample information based on similar statistical characteristics and load fluctuation information required for smart grid equipment investment and scheduling evaluation, and therefore has high practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are for explanation purposes only and are not intended to limit the scope of the present invention in any way. In addition, the shapes and proportional dimensions of the components in the drawings are only for illustration purposes to help understand the present invention, and are not intended to specifically limit the shapes and proportional dimensions of the components of the present invention. In the drawings: Figure 1 The present invention is a flow chart of a method for generating hourly load sequences based on monthly peak and valley electricity consumption records.

[0035] Figure 2 This is a structural diagram of the hourly load sequence generation system based on monthly peak and valley electricity consumption records of the present invention.

[0036] Figure 3 This is an electronic device diagram of the method for generating hourly load sequences based on monthly peak and valley electricity consumption records of the present invention.

[0037] Figure 4 A flow chart of a method for generating hourly load sequences based on monthly peak and valley electricity consumption records provided in an embodiment of the present invention.

[0038] Figure 5 A schematic diagram of a load data in a reference sample data set provided by an embodiment of the present invention.

[0039] Figure 6 This is a schematic diagram of some components obtained by decomposing load data using empirical mode decomposition according to an embodiment of the present invention.

[0040] Figure 7 This is an example diagram of data obtained by summarizing stable subsequences in an embodiment of the present invention.

[0041] Figure 8 A one-day electricity price fluctuation chart is provided for obtaining electricity price data corresponding to a target load in an embodiment of the present invention.

[0042] Fig. 9 A vector sample diagram of each high-dimensional vector of sample data obtained in an embodiment of the present invention.

[0043] Fig.10 A schematic diagram of obtaining monthly peak and valley power consumption records of a target load in an embodiment of the present invention.

[0044] Fig.11 This is a scatter plot of the distances from some reference samples in the embodiments of the present invention to the monthly peak and valley power consumption data of the target load.

[0045] Fig.12 A schematic diagram of a target load sequence specification according to an embodiment of the present invention.

[0046] Fig.13 This is a schematic diagram of the annual load data manually generated based on the traditional monthly peak and valley electricity consumption data. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0048] Embodiment 1 See also Figure 1 , a method for generating hourly load sequences based on monthly peak and valley electricity consumption records, comprising the following steps: The reference sample data is mapped to the high-dimensional vector data space of the monthly peak-valley electricity consumption records of the target to be constructed by using a mapping function, and all vectors in the high-dimensional vector data space are normalized to obtain a normalized high-dimensional vector data space; The distance between the monthly peak and valley electricity consumption records and each reference sample data is calculated in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, the distance in the nearest neighbor set is converted into a normalized distance weight value; After normalizing each load data in the neighbor set, the load data sequence is obtained, and the load data sequence is decomposed and converted into the sum of several subsequences. The subsequences are divided into random subsequences and stable subsequences based on information entropy; After summarizing the random subsequences and the stable subsequences, the two types of subsequences are added together to form the normalized hourly load sequence of the target to be constructed, and the normalized hourly load sequence of the target to be constructed is denormalized to obtain the hourly load sequence of the target to be constructed.

[0049] This embodiment maps the reference sample data to a high-dimensional vector data space through a mapping function and normalization processing, so that the relationship and characteristics between the data are clearer, which helps to improve the accuracy of the prediction. By calculating the distance and selecting the nearest neighbor set, the reference sample closest to the target to be constructed can be further screened out, thereby further improving the prediction accuracy. The load data sequence is decomposed into a random subsequence and a stable subsequence. This decomposition method helps the model to better understand and process load data of different properties. By processing the two types of subsequences separately, the model can more flexibly adapt to the power demand in different scenarios, thereby enhancing the adaptability of the model. The generation of hourly load sequences enables the power system to more accurately understand the power demand in each time period, thereby optimizing the allocation of power resources. It can reduce power waste, improve the stability and reliability of power supply, and also help reduce the operating costs of power companies. The generated hourly load sequence can provide an important basis for power dispatching, energy management, and electricity price setting. By accurately predicting power demand, the method of this embodiment helps to promote power companies and users to adopt more energy-saving and environmentally friendly power consumption methods.

[0050] The mapping function is used to map the reference sample data to the high-dimensional vector data space of the monthly peak and valley electricity consumption records of the target to be constructed, including: The monthly peak-valley electricity consumption records of the target to be constructed are represented as high-dimensional vectors, and the reference sample data are mapped to the data space of the high-dimensional vector representing the monthly peak-valley electricity consumption records of the target to be constructed through a mapping function using the annual tiered electricity price pricing information. The reference sample data and the monthly peak-valley electricity consumption records of the target to be constructed are in the same dimension; The annual tiered electricity price pricing information and monthly peak and valley electricity consumption records are as follows:

[0051]

[0052] in, To record the peaks and valleys of each month, Subscript i Indicates that the element is in the first i Hour, The element x represents a certain attribute value corresponding to the hour, and there are 8760 hours in a year; Subscript i Indicates the month, Superscript price(k) Indicates the first k Class electricity price, L represents the total electricity consumption corresponding to the electricity price in that month, and there are 12 months in a year; The mapping function is shown below:

[0053] in, To record the peaks and valleys of each month, The annual electricity price is For the whole year load; The mapping is as follows: traversal month by month , get the i The countable set elements in the monthly electricity price sequence { price}; Based on countable set elements { price} i Filter the monthly electricity price sequence to obtain the i Monthly electricity price sequence and specific elements in countable sets price(k) Equal sequence indexes ;from Get the index with sequence index The load data of the same position are summed up to get , iteratively execute the above process to obtain Different, each load sequence The corresponding monthly peak and valley records V。

[0054] Through the mapping function, the reference sample data and the monthly peak and valley electricity consumption records of the target to be constructed are unified into the same high-dimensional vector data space, ensuring the consistency of the data in terms of dimension, making subsequent analysis and comparison more accurate and reliable. The high-dimensional vector data space can accommodate more information, including the peak and valley characteristics of electricity consumption records, electricity price pricing information, etc. The mapping process is actually a feature extraction and enhancement process, which helps the model better capture and utilize key information in the data. After mapping the data to a high-dimensional space, the model can learn more complex and detailed data features, thereby improving the accuracy of prediction and classification.

[0055] Through the mapping function, data from different sources and formats can be unified into the same high-dimensional space, thereby achieving data fusion and integration, which is of great significance for building a cross-domain and cross-system integrated energy management system. After mapping the data into a high-dimensional space, the model can learn richer feature information, thereby enhancing the generalization ability of the model, which means that the model can make more accurate and reliable predictions when faced with new and unseen data.

[0056] In high-dimensional space, the distance and similarity between data points can be calculated and evaluated more accurately, which helps the power system to understand the electricity demand in different time periods more accurately, thereby optimizing the allocation of power resources, reducing waste, and improving the stability and reliability of power supply. High-dimensional vector data space makes complex analysis possible, helps to discover potential laws and patterns in electricity consumption data, and provides a more scientific basis for power companies' decision-making.

[0057] The normalization method is as follows:

[0058] in, is the normalization method, For vector V The scalar value of the element with the largest value in , The maximum value of is 1. The minimum value of is not necessarily 0; , With function The interchangeability between them is as follows: .

[0060] The distance between the monthly peak and valley electricity consumption records and each reference sample data is calculated in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, the distance in the nearest neighbor set is converted into a normalized distance weight value, including: Select reference samples whose distance to the target to be constructed is less than the threshold to form a sample vector set, as shown below:

[0061] The distance set corresponding to the sample vector set is as follows:

[0062] The neighbor set corresponding to the distance set is as follows:

[0063] When converting the distance in the distance neighbor set to a normalized distance weight value Use the radial basis function rbf and normalize it as follows:

[0064] in, is a sample vector set, is the distance set of the sample vector set, is the normalized weight at the target to be constructed, Represents two load sequences i,j The distance between them is the distance between their normalized high-dimensional vectors. The peak and valley electricity consumption records of the target to be constructed, is the high-dimensional vector of each reference sample data, is the distance set between the target to be constructed and each reference sample, and threshold is the threshold.

[0065] By calculating the distance between the monthly peak and valley electricity consumption records and each reference sample data in the high-dimensional vector data space, the similarities and differences between them can be accurately evaluated. The setting of the threshold helps to screen out the reference samples closest to the target to be constructed, that is, the neighbor set, so as to ensure the data quality of subsequent analysis. The distance in the neighbor set is converted into a normalized distance weight value, which can quantify the influence of different reference samples on the target to be constructed. The normalization process ensures that the weight value ranges between 0 and 1, which is convenient for subsequent calculations and analysis. By considering the weighted average influence of multiple neighbor samples, the model can reduce excessive dependence on a single sample data, thereby improving the robustness and stability of the model. Even in the case of noise or anomalies in some sample data, the model can weaken the influence of these noise or anomalies through the weighted average of the neighbor set.

[0066] When using the nearest neighbor set and the normalized distance weight value for prediction, the electricity consumption of the target to be constructed can be reflected more accurately. The weighted average influence of multiple similar samples is taken into account, thereby improving the accuracy and reliability of the prediction. The normalized distance weight value can be used as the basis for subsequent complex analysis. By considering the nearest neighbor set and the normalized distance weight value, the calculation process of the algorithm can be simplified, the efficiency of the algorithm can be improved, and it is helpful to discover the potential laws and patterns in the electricity consumption data, providing a more scientific basis for the decision-making of power companies.

[0067] After normalizing the load data in the neighbor set, the load data sequence is obtained. The load data sequence is decomposed into the sum of several subsequences. Specifically, for any neighbor set The corresponding normalized load sequence is decomposed into the sum of several subsequences using the empirical mode decomposition method; The decomposition using the empirical mode decomposition method is as follows:

[0068]

[0069] in, is the normalized load sequence, EMD is the empirical mode decomposition method, For vector V The scalar value of the element with the largest value in , For the whole year load; Based on information entropy, subsequences are divided into random subsequences and stable subsequences as follows: Perform discrete Fourier transform on each subsequence, calculate the information entropy of the positive half frequency sequence obtained after each discrete Fourier transform, sort each subsequence in ascending order according to the information entropy of the positive half frequency sequence, and then divide the subsequence set into random subsequences according to the specified threshold With stable subsequence ; Random Subsequence is the subset of subsequences whose information entropy is less than the threshold sum; Stable subsequence is the subset of subsequences containing information entropy greater than the threshold sum.

[0070] After summarizing the random subsequences and the stable subsequences, the two types of subsequences are added together to form the normalized hourly load sequence of the target to be constructed as follows:

[0071] in, is the normalized hourly load sequence of the target to be constructed, is a random subsequence, is the stable subsequence of the target to be constructed; The stable subsequence of the target to be built is as follows:

[0072] The random subsequence of the target to be built is as follows:

[0073]

[0074] in, The function is a random sequence construction method, and param represents the statistical characteristic parameters.

[0075] The normalized hourly load sequence of the target to be built is denormalized to obtain the hourly load sequence of the target to be built as shown in the following formula:

[0076] in, is the hourly load sequence of the target to be built, is the normalized hourly load sequence of the target to be constructed, It is the inverse operation of the normalization method.

[0077] Denormalizing the normalized hourly load series of the target to be built can not only restore the original scale of the data and improve the readability of the model output, but also support subsequent analysis and decision-making, maintain data consistency, and improve the practicality of the model.

[0078] The successful development of this embodiment will provide valuable decision-making support tools for many organizations that lack detailed electricity consumption data, help improve energy efficiency, reduce operating costs, and pave the way for large-scale application of smart grids and renewable energy. At the same time, it can also provide important data support for government departments to formulate relevant policies and industry standards, and promote the entire power industry to develop in a more intelligent, efficient and sustainable direction.

[0079] Embodiment 2 See also Figure 2 , an hourly load sequence generation system based on monthly peak and valley power consumption records, including: A mapping module is used to map the reference sample data to the high-dimensional vector data space of the monthly peak and valley electricity consumption records of the target to be constructed by using a mapping function, and normalize all vectors in the high-dimensional vector data space to obtain a normalized high-dimensional vector data space; The distance weight normalization module is used to calculate the distance between the monthly peak and valley electricity consumption records and each reference sample data in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, convert the distance in the nearest neighbor set into a normalized distance weight value; A subsequence distinguishing module is used to obtain a load data sequence after normalizing each load data in the neighbor set, decompose the load data sequence into the sum of several subsequences, and distinguish the subsequences into random subsequences and stable subsequences based on information entropy; An hourly load sequence module is obtained, which is used to summarize the random subsequences and the stable subsequences and then add the two types of subsequences to form a normalized hourly load sequence of the target to be constructed, and denormalize the normalized hourly load sequence of the target to be constructed to obtain the hourly load sequence of the target to be constructed.

[0080] In this embodiment, by mapping the reference sample data to a high-dimensional vector data space and performing normalization processing, the system can more accurately capture the inherent laws and characteristics of the power consumption data. Using the distance weight normalization module, the system can screen out the reference sample closest to the target to be constructed, thereby further improving the accuracy of the prediction. The subsequence differentiation module decomposes the load data sequence into random subsequences and stable subsequences through information entropy. This decomposition helps the model to better understand and process load data of different properties. By processing the two types of subsequences separately, the system can more flexibly adapt to the power demand in different scenarios, thereby enhancing the generalization ability of the model. The generation of hourly load sequences enables the power system to more accurately understand the power demand in each time period, thereby optimizing the allocation of power resources. The hourly load sequences generated by the system can provide an important basis for power dispatching, energy management, and electricity price setting. It helps power companies and relevant departments make more scientific and reasonable decisions and improve overall operational efficiency. For users, the generation of hourly load sequences helps to better understand the power consumption situation, so as to adopt a more energy-saving and environmentally friendly way of using power.

[0081] Embodiment 3 See also Figure 3 An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for generating hourly load sequences based on monthly peak and valley electricity consumption records described in Embodiment 1 is implemented.

[0082] Embodiment 4 A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for generating hourly load sequences based on monthly peak and valley electricity consumption records described in Embodiment 2 is implemented.

[0083] Embodiment 5 The method for generating hourly load sequences based on monthly peak and valley electricity consumption records has the following specific steps: S1. Represent the monthly peak-valley electricity consumption record of the target to be constructed as a high-dimensional vector, and use the annual tiered electricity price information to map the reference sample data to the data space of the vector so that it has the same dimension as the monthly peak-valley electricity consumption record of the target to be constructed; Specifically, S1 includes a sequence with a defined length of 8760:

[0084] The subscript i indicates that the element is in the i-th hour of the year, and the element x indicates a certain attribute value corresponding to the hour. Therefore, the annual electricity price is expressed as , the annual load is expressed as ,etc.

[0085] The monthly peak and valley records are also a sequence, expressed as:

[0086] Wherein, the subscript i represents the month, the superscript price(k) represents the k-th electricity price in the month, and L represents the total electricity consumption corresponding to the electricity price in the month. If there is an electricity bill, the sequence can be directly obtained.

[0087] Without having an electricity bill, and When is known, V can be calculated from these two. Different, each load sequence The corresponding V is different. The calculation process is: 1) Month by month traversal , get the countable set element {price} in the electricity price sequence in the i-th month.

[0088] 2) Filter the electricity price sequence for the i-th month based on the countable set elements, and obtain the sequence index index in the sequence that is equal to the specific element price(k) in the set .

[0089] 3) From Get the index with the index The load data at the same position are summed up to get , iteratively execute the above process to obtain V.

[0090] Through this method, the reference sample data can be converted to the same dimensional space as the monthly peak and valley power consumption, making it measurable. The above operations are all deterministic logical operations, so they can all be realized through software automation. This conversion method is defined as a function, that is, .

[0091] S2, applying a normalization method that is interchangeable with the aforementioned mapping function to all vectors in the aforementioned high-dimensional data space to make it have better metricity; Specifically, S2 includes the load difference between different reference sample data, which will affect the correct measurement of the similarity between data. Therefore, the data needs to be normalized. After given, each reference sample data has a corresponding V, so each reference sample data can also be expressed as In this representation, a normalization method is needed to measure the data in the high-dimensional space where V is located, and the normalization method and its inverse operation are required to be able to To do this, use the function The normalization method with exchangeability is as follows:

[0092] in The scalar value of the largest element in vector V.

[0093] Compared with the common maximum and minimum normalization methods, this normalization method has the same maximum value of 1, but the minimum value is not necessarily 0. This method is commutative with the original function, that is:

[0094] S3, calculating the distance between the monthly peak and valley electricity consumption records and each reference sample data in the high-dimensional data space, and selecting a neighbor set based on a threshold, and then using a radial basis function to convert the sample-to-use distance in the neighbor set into a normalized distance weight value; Specifically, S3 includes defining the distance between two load sequences i, j as the distance between their respective normalized high-dimensional vectors, that is:

[0095] The peak and valley electricity consumption records of the target to be constructed are recorded as , the high-dimensional vector of each reference sample data is recorded as , then the distance between the target to be constructed and each reference sample is a set In order to avoid excessive smoothing of data caused by using multiple samples, a threshold threashold is introduced to filter the distance set, and reference samples whose distance to the target to be constructed is less than the threshold are selected to form a sample vector set. That is:

[0096] The sample set corresponds to a distance set .

[0097] At the same time, the sample vector set also corresponds to a sample set, namely , this set is called the nearest neighbor set.

[0098] right Use the radial basis function rbf and normalize it to express the normalized weight of the target to be constructed, that is, .

[0099] S4. Use the exchangeability normalization method to process the load data in the neighbor set, then decompose the load data sequence and convert it into the sum of several subsequences, and then use the information entropy as a benchmark to quantitatively evaluate each subsequence and distinguish it into a random subsequence and a stable subsequence; Specifically, S4 includes defining the normalized load sequence as .

[0100] For any , the corresponding normalized load sequence can be decomposed into the sum of several subsequences using the empirical mode decomposition (EMD) method. That is:

[0101] The length of each subsequence is the same as the length of the element sequence itself. It is an empirical mode decomposition method, which is a public technology.

[0102] The N subsequences obtained by decomposing each load sequence are quantitatively evaluated to determine the randomness implied in the sequence. This is mainly achieved through the following steps: 1) Perform discrete Fourier transform on each subsequence and take the positive half of the discrete data as the next step of analysis.

[0103] 2) Calculate the information entropy of the positive half frequency sequence obtained after each discrete Fourier transform, sort each subsequence in ascending order, and then divide the subsequence set into two categories according to the specified threshold.

[0104] 3) The subset containing subsequences whose information entropy is less than the threshold is defined as , the sum of the subsets is called a random subsequence ; The subset of subsequences greater than the threshold is defined as , the sum of the subsets is called a stable subsequence .

[0105] S5. After summarizing the aforementioned stable subsequences and random subsequences using the weighted sum method and the statistical fusion method respectively, the two types of subsequences are added into a normalized hourly load sequence with a construction target, and then the inverse method of the exchangeable normalization method in the second step is used to finally realize the generation of an hourly load sequence with a construction target.

[0106] Specifically, S5 includes any According to method 5, the stable subsequence can be calculated and random subsequences Combined with the normalized weights obtained in method 4, the stable subsequence of the target to be constructed can be expressed as:

[0107] For any For the random subsequence in , first obtain its statistical characteristics such as the first four order distances, one-step autocorrelation coefficient, etc., namely:

[0108] Then, the normalized weights obtained by method 4 are used on the neighbor set to obtain the weighted sum of the above parameters, namely:

[0109] Here param represents various statistical characteristic parameters. On this basis, the general random sequence construction method is used to obtain the random subsequence of the target to be constructed:

[0110] in The function is a random sequence construction method, which is a general public method.

[0111] Combining the random subsequence with the stable subsequence, we can get the general normalized load sequence of the target to be solved:

[0112] Use the inverse operation of normalization mentioned in Method 2 to construct the load sequence data of the target to be solved:

[0113] This embodiment provides a method for generating hourly load sequences based on monthly peak and valley power consumption records in combination with the accompanying drawings. 1) Read a reference sample data set, where a load data such as Figure 4 As shown; 2) Use empirical mode decomposition to decompose the load data, and the resulting components are as follows: Figure 5 These sequences are all of equal length, in this example 8760; 3) After Fourier transforming each subsequence, the information entropy is calculated based on the kernel density estimation method on the frequency domain data, and the information entropy of the remaining subsequences is as follows: array([-4.42176182, -2.27995619, -1.98191449, -2.75489486, -3.19851772, -3.13732574, -3.36194394, -3.39774579, -1.1605304]) 4) Take the segmentation threshold -3, and we get Four pieces of data: Four pieces of data.

[0114] 5) Summarize the stable subsequences. The obtained data is as follows: Figure 6 As shown; 6) Perform statistical feature extraction on one of the random subsequences, and the results are as follows: {'mean':4.6310388007035635e-05, 'variance':1.7889339475518486e-07, 'skewness':0.0025101290201300463, 'kurtosis':-0.07009223848516388, autocorrelation(lag1)':0.6928725827100822} 7) Obtain the electricity price data corresponding to the target load. The daily electricity price fluctuation is as follows: Figure 7 As shown; 8) Based on the electricity price data, the reference sample data is statistically summarized to obtain high-dimensional vectors of each sample data. One vector sample is Figure 8 As shown, in this case, the vector is 40-dimensional; 9) At the same time, obtain the monthly peak and valley power consumption records of the target load, such as Fig. 9 As shown; 10) Calculate the distance between the reference sample and the monthly peak and valley power consumption data of the target load. The corresponding distance scatter plots of some samples are as follows: Fig.10 As shown, the distances from 26 sample data to the target data are shown, along with the distance values; [ 0.7210115336803177, 0.7558998037776984, 0.7959095111181895, 0.8046346878372183, 0.8059061321653479, 0.8156946702346761, 0.8166512234351301, 1.0230950710964608, 1.1286851961371265, 1.1575372167867548, 1.2301921695652929, 1.3844806901672677, 1.3961870410426334, 1.410113406931574, 1.410113406931574, 1.435166726962071, 1.4920752184775667, 1.5486837099469752, 11) Take 0.8 as the threshold and obtain the neighbor set with 3 elements. The corresponding weight coefficient calculated using rbf is as follows: array([3.04483943e-06, 1.19590160e-03, 9.98801054e-01]) 12) All subsequences are aggregated based on weights and reverse normalized to generate a standard target load sequence such as Fig.11 As shown; 13) As a comparison, the traditional annual load data manually generated based on monthly peak and valley power consumption data is as follows Fig.13 shown.

[0115] It can be seen that the data generated by the method provided in this embodiment not only considers the reference sample information based on similar statistical characteristics, but also considers the load fluctuation information required for smart grid equipment investment and dispatch evaluation. Such data is suitable for various organizations to conduct new energy equipment investment and dispatch evaluation.

[0116] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0117] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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.

[0118] 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 1A function specified in one or more boxes.

[0119] 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 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 is described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation of the present invention can still be modified or replaced by equivalents, and any modification or equivalent replacement that does not deviate from the spirit and scope of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for generating hourly load sequences based on monthly peak and valley electricity consumption records, characterized in that: The following steps are involved: The reference sample data is mapped to the high-dimensional vector data space of the monthly peak-valley electricity consumption records of the target to be constructed by using a mapping function, and all vectors in the high-dimensional vector data space are normalized to obtain a normalized high-dimensional vector data space; The distance between the monthly peak and valley electricity consumption records and each reference sample data is calculated in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, the distance in the nearest neighbor set is converted into a normalized distance weight value; After normalizing each load data in the neighbor set, the load data sequence is obtained, and the load data sequence is decomposed and converted into the sum of several subsequences. The subsequences are divided into random subsequences and stable subsequences based on information entropy; After summarizing the random subsequences and the stable subsequences, the two types of subsequences are added together to form the normalized hourly load sequence of the target to be constructed, and the normalized hourly load sequence of the target to be constructed is denormalized to obtain the hourly load sequence of the target to be constructed.

2. The method for generating hourly load sequences based on monthly peak and valley electricity consumption records according to claim 1, characterized in that: The method of mapping the reference sample data to the high-dimensional vector data space of the monthly peak and valley electricity consumption records of the target to be constructed by using a mapping function includes: The monthly peak-valley electricity consumption records of the target to be constructed are represented as high-dimensional vectors, and the reference sample data are mapped to the data space of the high-dimensional vector representing the monthly peak-valley electricity consumption records of the target to be constructed through a mapping function using the annual tiered electricity price pricing information. The reference sample data and the monthly peak-valley electricity consumption records of the target to be constructed are in the same dimension; The annual tiered electricity price pricing information and the monthly peak and valley electricity consumption records are as follows: in, To record the peaks and valleys of each month, Subscript i Indicates that the element is in the first i Hour, The element x represents a certain attribute value corresponding to the hour, and there are 8760 hours in a year; Subscript i Indicates the month, Superscript price (k) Indicates the first k Class electricity price, L represents the total electricity consumption corresponding to the electricity price in that month, and there are 12 months in a year; The mapping function is shown below: in, To record the peaks and valleys of each month, The annual electricity price is For the whole year load; The mapping is specifically as follows: traversal month by month , get the i The countable set elements in the monthly electricity price sequence { price }; Based on countable set elements { price } i Filter the monthly electricity price sequence to obtain the i Monthly electricity price sequence and specific elements in countable sets price(k) Equal sequence indexes ;from Get the index with sequence index The load data with the same position are summed up to get , iteratively execute the above process to obtain Different, each load sequence The corresponding monthly peak and valley records V .

3. The method for generating hourly load sequences based on monthly peak and valley electricity consumption records according to claim 2, characterized in that: The normalization method is as follows: in, is the normalization method, For vector V The scalar value of the element with the largest value in , The maximum value of is 1. The minimum value of is not necessarily 0; , With function The interchangeability between them is as follows: 。 4. The method for generating hourly load sequences based on monthly peak and valley electricity consumption records according to claim 3, characterized in that: The method of calculating the distance between the monthly peak and valley electricity consumption records and each reference sample data in the normalized high-dimensional vector data space, selecting a neighbor set according to a threshold, and converting the distance in the neighbor set into a normalized distance weight value includes: Select reference samples whose distance to the target to be constructed is less than the threshold to form a sample vector set, as shown below: The distance set corresponding to the sample vector set is as follows: The neighbor set corresponding to the distance set is as follows: The distance in the neighbor set is converted into a normalized distance weight value. Use the radial basis function rbf and normalize it as follows: in, is a sample vector set, is the distance set of the sample vector set, is the normalized weight at the target to be constructed, Represents two load sequences i,j The distance between them is the distance between their normalized high-dimensional vectors. The peak and valley electricity consumption records of the target to be constructed, is the high-dimensional vector of each reference sample data, is the distance set between the target to be constructed and each reference sample, and threshold is the threshold.

5. The method for generating hourly load sequences based on monthly peak and valley electricity consumption records according to claim 4, characterized in that: The load data sequence is obtained by normalizing each load data in the neighboring set, and the load data sequence is decomposed into the sum of several subsequences as follows: The corresponding normalized load sequence is decomposed into the sum of several subsequences using the empirical mode decomposition method; The decomposition using the empirical mode decomposition method is as follows: in, is the normalized load sequence, EMD is the empirical mode decomposition method, For vector V The scalar value of the element with the largest value in , For the whole year load; The method of distinguishing subsequences into random subsequences and stable subsequences based on information entropy is specifically as follows: Perform discrete Fourier transform on each subsequence, calculate the information entropy of the positive half frequency sequence obtained after each discrete Fourier transform, sort each subsequence in ascending order according to the information entropy of the positive half frequency sequence, and then divide the subsequence set into random subsequences according to the specified threshold With stable subsequence ; Random Subsequence is the subset of subsequences whose information entropy is less than the threshold sum; Stable subsequence is the subset of subsequences containing information entropy greater than the threshold sum.

6. The method for generating hourly load sequences based on monthly peak and valley electricity consumption records according to claim 5, characterized in that: After aggregating the random subsequence and the stable subsequence, the two types of subsequences are added together to form a normalized hourly load sequence of the target to be constructed as follows: in, is the normalized hourly load sequence of the target to be constructed, is a random subsequence, is the stable subsequence of the target to be constructed; The stable subsequence of the target to be constructed is as follows: The random subsequence of the target to be constructed is as follows: in, The function is a random sequence construction method, and param represents the statistical characteristic parameters.

7. The method for generating hourly load sequences based on monthly peak and valley electricity consumption records according to claim 6, characterized in that: The normalized hourly load sequence of the target to be constructed is denormalized to obtain the hourly load sequence of the target to be constructed as shown in the following formula: in, is the hourly load sequence of the target to be built, is the normalized hourly load sequence of the target to be constructed, It is the inverse operation of the normalization method.

8. An hourly load sequence generation system based on monthly peak and valley electricity consumption records, characterized in that: include: A mapping module is used to map the reference sample data to the high-dimensional vector data space of the monthly peak and valley electricity consumption records of the target to be constructed by using a mapping function, and normalize all vectors in the high-dimensional vector data space to obtain a normalized high-dimensional vector data space; The distance weight normalization module is used to calculate the distance between the monthly peak and valley electricity consumption records and each reference sample data in the normalized high-dimensional vector data space, and after selecting the nearest neighbor set according to the threshold, convert the distance in the nearest neighbor set into a normalized distance weight value; A subsequence distinguishing module is used to obtain a load data sequence after normalizing each load data in the neighbor set, decompose the load data sequence into the sum of several subsequences, and distinguish the subsequences into random subsequences and stable subsequences based on information entropy; An hourly load sequence module is obtained, which is used to summarize the random subsequences and the stable subsequences and then add the two types of subsequences to form a normalized hourly load sequence of the target to be constructed, and denormalize the normalized hourly load sequence of the target to be constructed to obtain the hourly load sequence of the target to be constructed.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for generating hourly load sequences based on monthly peak and valley electricity consumption records as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the method for generating hourly load sequences based on monthly peak and valley electricity consumption records as described in any one of claims 1 to 7.

Images