A method and apparatus for constructing electricity price samples based on cross-weighting of bidding space

Abstract

This invention provides a method and apparatus for constructing electricity price samples based on cross-weighting of bidding space, belonging to the field of electricity spot market price forecasting technology. The method acquires multi-dimensional electricity operation characteristic data for the date to be predicted and multiple historical days, and calculates the distance between basic multi-dimensional features. For each historical day, a bidding space cross-feature sequence is constructed based on the time-period deviation between the electricity price feature sequence and the bidding space feature sequence, and a bidding space cross-weighting factor is determined based on the dispersion of each cross-feature value. The basic multi-dimensional feature distance is corrected using the factor to obtain a comprehensive distance index. Target historical days are selected based on the comprehensive distance index, and a price forecasting sample set is constructed using the multi-dimensional electricity operation characteristic data and electricity price data of the target historical days. This invention can reflect the linkage between bidding space and electricity price during the sample selection process, improving the representativeness of the price forecasting samples.

Description

Technical Field

[0001] This invention belongs to the field of electricity spot market price forecasting technology, and in particular relates to a method and apparatus for constructing electricity price samples based on the cross-weighting of bidding space. Background Technology

[0002] In the process of forecasting electricity prices in the electricity spot market, it is usually necessary to select historical days similar to the date to be predicted from historical operating data, and then construct a prediction sample using the multidimensional power operation characteristics and price data of the selected historical days. Existing similar day selection methods generally select power operation characteristics such as system load, renewable energy output, and tie-line load, normalize the corresponding characteristics of the date to be predicted and each historical day, calculate the distance, and then select similar historical days based on the obtained distance.

[0003] However, electricity prices in the spot market are not only related to the similarity of various electricity operation characteristics, but are also influenced by factors such as market supply and demand, generator participation, and bidding behavior. Even when the basic electricity operation characteristics on different historical days are relatively similar, their actual electricity price changes can still differ significantly. Existing methods mainly rely on the distance between multi-dimensional electricity operation characteristics to select historical days, making it difficult to effectively distinguish between historical days with similar operation characteristics but different electricity price response characteristics. This results in the constructed electricity price forecast sample set being insufficiently representative of the market state on the forecast date. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for constructing electricity price samples based on the cross-weight of the bidding space, which can reflect the linkage between the bidding space and the electricity price during the sample screening process and improve the representativeness of the electricity price prediction samples.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for constructing electricity price samples based on cross-weighting of bidding space, comprising: Acquire multidimensional power operation characteristic data for the day to be predicted and multiple historical days, and calculate the basic multidimensional feature distance between the day to be predicted and each of the historical days respectively; For each historical day, the electricity price feature sequence and the bidding space feature sequence corresponding to that historical day are determined. The deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence is calculated for each time period to obtain the cross feature value corresponding to each time period. The cross feature values ​​corresponding to each time period are used to form the bidding space cross feature sequence corresponding to that historical day. Based on the degree of dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean, the bidding space cross weight factor corresponding to the historical day is determined; By using the cross-weighting factor of the bidding space corresponding to each historical day to correct the corresponding basic multidimensional feature distance, the comprehensive distance index corresponding to each historical day is obtained. Based on the comprehensive distance index corresponding to each of the historical days, a target historical day is selected from multiple historical days, and based on the multidimensional power operation characteristic data and electricity price data of the target historical day, an electricity price prediction sample set corresponding to the day to be predicted is constructed.

[0006] Furthermore, the step of calculating the deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence on a time-period basis includes: For each historical day, the electricity price feature sequence and the bidding space feature sequence are processed using the maximum-minimum normalization method to obtain the normalized electricity price feature value and the normalized bidding space feature value corresponding to each time period. The cross-feature values ​​corresponding to each time period in the historical day are determined according to the following formula:

[0007] The steps for determining the bidding space cross weight factor corresponding to the historical day based on the dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean include: Based on the crossover characteristic values ​​corresponding to each time period in the historical day, the sequence mean and the bidding space crossover weight factor are determined according to the following formula:

[0008]

[0009] in, Indicates the first The first historical day in China The normalized electricity price characteristic value corresponding to the time period, Indicates the first The first historical day in China Normalized auction space eigenvalues ​​corresponding to the time period Indicates the first The first historical day in China Cross-feature values ​​corresponding to time periods Indicates the first The series mean corresponding to each historical day This indicates the total number of time periods in a single day. Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day.

[0010] Furthermore, the step of correcting the basic multidimensional feature distance corresponding to each of the historical days using the bidding space cross-weighting factor includes: Using the cross-weighting factor of the bidding space corresponding to the same historical day, the distance of the basic multidimensional features corresponding to that historical day is corrected according to the following formula:

[0011] in, Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day Indicates the first The basic multidimensional feature distance corresponding to each historical day Indicates the first The comprehensive distance index corresponding to each historical day; The steps of selecting a target historical day from multiple historical days based on the comprehensive distance index corresponding to each historical day, and constructing a sample set for electricity price prediction corresponding to the day to be predicted based on the multidimensional power operation characteristic data and electricity price data of the target historical day include: The historical days are sorted in ascending order of the comprehensive distance index, and the historical days whose comprehensive distance index is less than the similarity index threshold are determined as the target historical days. Based on the multidimensional power operation characteristic data and electricity price data of each target historical day, the electricity price prediction sample set is obtained.

[0012] Furthermore, the step of determining the bidding space feature sequence corresponding to the historical day includes: Regarding the first The first historical day For each time period, the corresponding bidding space characteristic value is determined according to the following formula:

[0013] in, Indicates the first The first historical day in China The characteristic value of the bidding space corresponding to the time period, Indicates the first The first historical day in China System load forecast for the time period Indicates the first The first historical day in China The power load of the tie line during the time period, Indicates the first The first historical day in China Wind power output forecast for the period Indicates the first The first historical day in China Forecasted photovoltaic output for the period This indicates the proportion of new energy vehicles that did not submit bids. Indicates the first The first historical day in China Total output of non-market-based generating units during the specified time period; Arrange the bidding space feature values ​​corresponding to each time period of the historical day in chronological order to obtain the bidding space feature sequence corresponding to the historical day.

[0014] In a second aspect, the present invention provides an apparatus for constructing electricity price samples based on cross-weighted bidding space, comprising: The basic distance calculation module is used to acquire multi-dimensional power operation characteristic data of the day to be predicted and multiple historical days, and to calculate the basic multi-dimensional characteristic distance between the day to be predicted and each of the historical days respectively. The cross-feature sequence construction module is used to determine the electricity price feature sequence and the bidding space feature sequence corresponding to each historical day, calculate the deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence for each time period, obtain the cross-feature value corresponding to each time period, and form the bidding space cross-feature sequence corresponding to the historical day by the cross-feature values ​​corresponding to each time period. The cross weight determination module is used to determine the cross weight factor of the bidding space corresponding to the historical day based on the degree of dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean. The distance correction module is used to correct the corresponding basic multidimensional feature distance using the bidding space cross-weighting factor corresponding to each historical day, so as to obtain the comprehensive distance index corresponding to each historical day. The sample set construction module is used to select target historical days from multiple historical days based on the comprehensive distance index corresponding to each historical day, and to construct the electricity price prediction sample set corresponding to the day to be predicted based on the multidimensional power operation characteristic data and electricity price data of the target historical days.

[0015] Furthermore, the cross-feature sequence construction module is specifically used to process the electricity price feature sequence and the bidding space feature sequence for each historical day using a maximum-minimum normalization method to obtain the normalized electricity price feature value and the normalized bidding space feature value corresponding to each time period, and to determine the cross-feature value corresponding to each time period in the historical day according to the following formula:

[0016] The cross-weight determination module is specifically used to determine the sequence mean and the bidding space cross-weight factor according to the cross-feature values ​​corresponding to each time period in the historical day, using the following formula:

[0017]

[0018] in, Indicates the first The first historical day in China The normalized electricity price characteristic value corresponding to the time period, Indicates the first The first historical day in China Normalized auction space eigenvalues ​​corresponding to the time period Indicates the first The first historical day in China Cross-feature values ​​corresponding to time periods Indicates the first The series mean corresponding to each historical day This indicates the total number of time periods in a single day. Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day.

[0019] Furthermore, the distance correction module is specifically used to correct the basic multidimensional feature distance corresponding to the same historical day using the bidding space cross-weight factor, according to the following formula:

[0020] in, Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day Indicates the first The basic multidimensional feature distance corresponding to each historical day Indicates the first The comprehensive distance index corresponding to each historical day; The sample set construction module is specifically used to sort multiple historical days in ascending order of the comprehensive distance index, determine the historical days whose comprehensive distance index is less than the similarity index threshold as the target historical days, and obtain the electricity price prediction sample set based on the multidimensional power operation characteristic data and electricity price data of each target historical day.

[0021] Furthermore, the cross-feature sequence construction module is specifically used for: Regarding the first The first historical day For each time period, the corresponding bidding space characteristic value is determined according to the following formula:

[0022] in, Indicates the first The first historical day in China The characteristic value of the bidding space corresponding to the time period, Indicates the first The first historical day in China System load forecast for the time period Indicates the first The first historical day in China The power load of the tie line during the time period, Indicates the first The first historical day in China Wind power output forecast for the period Indicates the first The first historical day in China Forecasted photovoltaic output for the period This indicates the proportion of new energy vehicles that did not submit bids. Indicates the first The first historical day in China Total output of non-market-based generating units during the specified time period; Arrange the bidding space feature values ​​corresponding to each time period of the historical day in chronological order to obtain the bidding space feature sequence corresponding to the historical day.

[0023] In a third aspect, the present invention provides an electronic device, characterized in that it includes a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement a method for constructing electricity price samples based on bidding space cross weights.

[0024] In a fourth aspect, the present invention provides a computer-readable storage medium storing at least one instruction that, when executed by a processor, implements a method for constructing electricity price samples based on bidding space cross-weights.

[0025] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention constructs a time-by-time cross-feature sequence between electricity price characteristics and bidding space characteristics, and determines the cross-weight factor of bidding space based on the dispersion of cross-feature values. It corrects the distance of basic multi-dimensional features, which can reflect the linkage between bidding space and electricity price in the historical day screening process, distinguish historical days with similar basic power operation characteristics but different electricity price response characteristics, and improve the representativeness of electricity price prediction samples.

[0026] 2. This invention reduces the impact of differences in dimensions and value ranges between different features by normalizing the electricity price feature sequence and the bidding space feature sequence. At the same time, it determines the bidding space feature value based on the system load forecast, tie line load, new energy forecast output, and total output of non-market units, and uses the dispersion of cross feature values ​​to form a weighting factor, which can improve the comparability and stability of the quantitative results of the linkage between bidding space and electricity price.

[0027] 3. This invention uses the cross-weighting factor of the bidding space to correct the distance of the basic multidimensional features, and sorts and filters multiple historical days according to the obtained comprehensive distance index. It can clarify the screening criteria of the target historical days while taking into account the differences in multidimensional power operation characteristics and the linkage between bidding space and electricity price, thereby improving the consistency and repeatability of the electricity price prediction sample set construction results. Attached Figure Description

[0028] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart illustrating a method for constructing electricity price samples based on cross-weighted bidding space according to an embodiment of the present invention. Figure 2 This is a structural block diagram of an electricity price sample construction device based on bidding space cross weights according to an embodiment of the present invention; Figure 3 This is a structural block diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0030] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0031] Example 1 This embodiment provides a method for constructing electricity price samples based on the cross-weighting of bidding space. For example... Figure 1 As shown, the method specifically includes the following steps S1 to S5.

[0032] Step S1: Obtain multidimensional power operation characteristic data for the day to be predicted and multiple historical days, and calculate the basic multidimensional feature distance between the day to be predicted and each historical day.

[0033] Specifically, multidimensional power operation characteristic data corresponding to multiple historical days are obtained from historical datasets, and multidimensional power operation characteristic data corresponding to the day to be predicted are also obtained. The multidimensional power operation characteristic data for each day includes the feature values ​​of multiple types of power operation characteristics at each time period of the day.

[0034] The various power operation characteristics of the forecast date and each historical date are normalized separately to eliminate the dimensional differences between different types of power operation characteristics, resulting in multidimensional normalized feature sets corresponding to the forecast date and each historical date. For the same type of power operation characteristic, the same normalization rules and normalization parameters are used to ensure that the normalized feature values ​​of the same category on the forecast date and each historical date are comparable.

[0035] In this embodiment, the bidding space is not included in the basic multidimensional feature distance as a general, independent feature alongside power operation characteristics such as load, wind power, and photovoltaics. Instead, a bidding space feature sequence is determined separately based on the system load forecast and corresponding boundary conditions. A cross-weighting factor for the bidding space is determined based on the time-period deviation between the bidding space feature sequence and the electricity price feature sequence. This cross-weighting factor is then used to correct the basic multidimensional feature distance. This avoids severing the linkage between the bidding space and electricity price by representing the bidding space solely through general feature distances.

[0036] Regarding the first For each historical day, a one-to-one correspondence is established between the normalized power operation characteristic values ​​of the same category and time period in the predicted day. The difference between the power operation characteristic values ​​of the historical day and the predicted day is calculated for each category and time period. Based on the sum of the squares of the differences of each power operation characteristic value, the first historical day is determined. The basic multidimensional feature distance corresponding to each historical day:

[0037] in, Indicates the first The basic multidimensional feature distance corresponding to each historical day The total number of categories representing the characteristics of power operation. Indicates the first The first historical day Class-A power operation characteristics in the first Normalized eigenvalues ​​for the time period Indicates the date to be predicted. Class-A power operation characteristics in the first Normalized eigenvalues ​​for the time period This indicates the total number of time periods in a single day.

[0038] The power operation characteristics do not include the bidding space used to determine the cross-weighting factor of the bidding space. Following the above method, multiple historical days are traversed to obtain the basic multidimensional feature distances corresponding to each historical day.

[0039] In one specific application, a day is divided into 96 time periods, namely... The differences between various power operation characteristic values ​​of each historical day and the day to be predicted in 96 time periods are calculated respectively, and the basic multidimensional characteristic distance of the corresponding historical day is determined based on all the differences.

[0040] Step S2: For each historical day, determine the electricity price feature sequence and the bidding space feature sequence corresponding to that historical day. Calculate the deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence for each time period to obtain the cross feature value corresponding to each time period. The cross feature value corresponding to each time period is used to form the bidding space cross feature sequence corresponding to that historical day.

[0041] Specifically, regarding the first For each historical day, the electricity price feature values ​​corresponding to each time period of the historical dataset are extracted and arranged into an electricity price feature sequence according to the time period order.

[0042] Regarding the first For each time period of a historical day, the system load forecast for that period is subtracted by the tie-line power load, the sum of wind power forecast and photovoltaic power forecast calculated according to the proportion of non-bid-participating renewable energy, and the total output of non-market-based units, to obtain the corresponding bidding space characteristic value for that time period:

[0043] in, Indicates the first The first historical day in China The characteristic value of the bidding space corresponding to the time period, Indicates the first The first historical day in China System load forecast for the time period Indicates the first The first historical day in China The power load of the tie line during the time period, Indicates the first The first historical day in China Wind power output forecast for the period Indicates the first The first historical day in China Forecasted photovoltaic output for the period This indicates that new energy vehicles did not submit bids for a certain percentage of their participation, and , Indicates the first The first historical day in China Total output of non-market-based generating units during a given period.

[0044] Arranged in chronological order The bidding space feature values ​​for each time period within a historical day are used to obtain the bidding space feature sequence corresponding to that historical day. With a single day divided into 96 time periods, the 96 bidding space feature values ​​form a 96-point bidding space feature sequence for the corresponding historical day.

[0045] Since the electricity price characteristic value and the bidding space characteristic value have different dimensions and ranges, for each historical day, the electricity price characteristic sequence and the bidding space characteristic sequence of that historical day are processed by the maximum-minimum normalization method to obtain the normalized electricity price characteristic value and the normalized bidding space characteristic value corresponding to each time period.

[0046] The following normalization process uses the current processing order as an example. The data is categorized into historical days. To differentiate between electricity price characteristic values ​​and bidding space characteristic values ​​corresponding to different historical days, historical day subscripts are set in the relevant parameters. . No. The first historical day in China The normalized electricity price characteristic value and the normalized bidding space characteristic value corresponding to the time period are as follows:

[0047]

[0048] in, Indicates the first The first historical day in China Electricity price characteristic value for a given time period and They represent the first The minimum and maximum electricity price characteristic values ​​in the electricity price characteristic sequence corresponding to each historical day. Indicates the first The first historical day in China The normalized electricity price characteristic value corresponding to the time period; Indicates the first The first historical day in China The characteristic value of the bidding space during the time period, and They represent the first The minimum and maximum bid space feature values ​​in the bid space feature sequence corresponding to each historical day. Indicates the first The first historical day in China The normalized bidding space characteristic value corresponding to the time period.

[0049] when At that time, the normalized electricity price characteristic value for each period of that historical day is determined according to a preset fixed value; when At that time, the normalized bidding space characteristic value for each time period of that historical day is determined according to the preset fixed value.

[0050] After normalization, the deviation between the normalized electricity price characteristic value and the normalized bidding space characteristic value is calculated one-to-one according to time periods within the same historical day. The first historical day in China The cross-feature values ​​corresponding to the time periods are:

[0051] Calculate the number in sequence The cross-feature values ​​corresponding to each time period in the historical days are arranged in chronological order to obtain the first... The bidding space crossover feature sequence corresponds to each historical day. The bidding space crossover feature sequence reflects the correspondence between electricity price and bidding space in each time period within that historical day, and serves as the basis for determining the bidding space crossover weight factor.

[0052] Step S3: Determine the bidding space cross weight factor corresponding to the historical day based on the degree of dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean.

[0053] First, regarding the first The arithmetic mean of the sequence corresponding to a given historical day is obtained by taking the arithmetic mean of all the cross feature values ​​in the cross feature sequence of the bidding space corresponding to a given historical day.

[0054] Then, the difference between the crossover eigenvalue and the sequence mean for each time period is calculated. The squares of each difference are accumulated and processed according to the total number of time periods per day to obtain the dispersion of the crossover eigenvalue relative to the sequence mean for each time period, thereby determining the degree of dispersion of the crossover eigenvalue relative to the sequence mean for each time period. Cross-weighting factor for bidding space corresponding to each historical day:

[0055] in, Indicates the first The first historical day in China Cross-feature values ​​corresponding to time periods Indicates the first The series mean corresponding to each historical day This indicates the total number of time periods in a single day. Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day.

[0056] When the cross-feature values ​​of each time period in the bidding space cross-feature sequence are relatively concentrated, the bidding space cross-weight factor is close to 1; when the dispersion of the cross-feature values ​​of each time period relative to the sequence mean increases, the bidding space cross-weight factor increases accordingly. In the subsequent distance correction process, the basic multidimensional feature distance corresponding to historical days with a large degree of dispersion will be significantly corrected, thereby reducing the probability that the historical day will be selected as the target historical day.

[0057] By iterating through multiple historical days in the same way, the cross-weighting factor of the bidding space corresponding to each historical day is obtained.

[0058] Step S4: Use the cross-weighting factor of the bidding space corresponding to each historical day to correct the corresponding basic multidimensional feature distance, and obtain the comprehensive distance index corresponding to each historical day.

[0059] Specifically, regarding the first On the first historical day, the cross-weighting factor of the bidding space corresponding to that historical day is used to correct the distance of the basic multidimensional features corresponding to that historical day, thus obtaining the first historical day. The comprehensive distance index corresponding to each historical day:

[0060] in, Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day Indicates the first The basic multidimensional feature distance corresponding to each historical day Indicates the first The comprehensive distance index corresponding to each historical day.

[0061] The cross-weighting factor of the bidding space is not less than 1. When the first... When the cross-feature values ​​of each time period in the bidding space cross-feature sequence corresponding to a historical day are relatively concentrated, When the value is close to 1, the correction to the distance of the basic multidimensional features is relatively small; when the dispersion of the cross-feature values ​​relative to the sequence mean increases, This corresponding increase increases the overall distance index corresponding to that historical day. Therefore, it reduces the likelihood that historical days with significant fluctuations in the time-period correlation between bidding space and electricity price will be selected as target historical days.

[0062] Multiple historical days are processed using the above method to obtain the comprehensive distance index for each historical day. The comprehensive distance index simultaneously represents the distance between the predicted day and the historical day in terms of multi-dimensional power operation characteristics, as well as the cross relationship between the bidding space and electricity price within the historical day. The smaller the comprehensive distance index, the closer the corresponding historical day is to the predicted day after being corrected by the bidding space cross-weighting factor.

[0063] It should be noted that the overall similarity in this embodiment is represented by a distance-type index. The smaller the overall distance index, the higher the similarity between the corresponding historical day and the day to be predicted. The similarity index threshold is a screening threshold set for this distance-type index.

[0064] Step S5: Based on the comprehensive distance index corresponding to each historical day, select the target historical day from multiple historical days, and construct the electricity price prediction sample set corresponding to the day to be predicted based on the multidimensional power operation characteristic data and electricity price data of the target historical day.

[0065] Specifically, multiple historical days are sorted in ascending order of comprehensive distance index, and the comprehensive distance index corresponding to each historical day is compared with a preset similarity index threshold in turn. Historical days with comprehensive distance index less than the similarity index threshold are determined as target historical days.

[0066] Let the total number of historical days be The historical days are arranged in ascending order of comprehensive distance index. Accordingly, the following conditions are met:

[0067] in, This represents the number of historical days where the overall distance index is less than the similarity index threshold after sorting. Therefore, All meet the similarity index threshold. When At that time, the following conditions are met:

[0068] Right now The last historical day that meets the similarity index threshold is considered the last historical day; subsequent historical days do not meet the similarity index threshold. At that time, all historical daily averages met the similarity index threshold.

[0069] The target historical day set obtained through filtering is represented as follows:

[0070] in, Indicates the target set of historical days. This represents the target historical days arranged in ascending order of comprehensive distance index and meeting the similarity index threshold. Indicates the sorted order of the first... A historical day The corresponding comprehensive distance index, To meet the similarity index threshold for the last historical day, This represents the threshold for the similarity index.

[0071] After selecting the target historical days, multidimensional power operation characteristic data and electricity price data for each target historical day are collected according to the correspondence between the target historical days and historical data. This ensures that the multidimensional power operation characteristic data for each target historical day corresponds to the electricity price data for the corresponding period of that historical day, resulting in an electricity price prediction sample set for the day to be predicted. The electricity price prediction sample set is used to provide historical samples for electricity price prediction on the day to be predicted.

[0072] In summary, this invention forms a complete process for constructing electricity price prediction samples through basic multidimensional feature distance calculation, construction of bidding space cross-feature sequences, correction of bidding space cross-weights, and selection of target historical days. This method not only considers the differences in multidimensional power operation characteristics between the day to be predicted and each historical day, but also uses the dispersion of the time-period deviation between the bidding space and electricity price within historical days to correct the basic multidimensional feature distance. This reduces the likelihood of historical days with unstable linkages between the bidding space and electricity price entering the electricity price prediction sample set, improves the representativeness of the constructed electricity price prediction sample set for the market state and price response characteristics of the day to be predicted, and provides more suitable historical samples for subsequent electricity price predictions.

[0073] Example 2 like Figure 2 As shown, based on the same inventive concept as the above embodiments, this embodiment provides an electricity price sample construction device based on bidding space cross weights, including a basic distance calculation module, a cross feature sequence construction module, a cross weight determination module, a distance correction module, and a sample set construction module.

[0074] The basic distance calculation module is used to obtain multi-dimensional power operation characteristic data of the day to be predicted and multiple historical days, and to calculate the basic multi-dimensional characteristic distance between the day to be predicted and each historical day.

[0075] Specifically, the basic distance calculation module obtains multidimensional power operation feature data corresponding to multiple historical days from the historical dataset, and also obtains multidimensional power operation feature data corresponding to the day to be predicted. The basic distance calculation module normalizes various types of power operation features for the day to be predicted and each historical day, and maps the normalized power operation feature values ​​of the same category and time period for the day to be predicted to each historical day. It calculates the difference between the corresponding normalized power operation feature values ​​for each category and time period, and determines the basic multidimensional feature distance corresponding to each historical day based on the sum of the squares of each difference.

[0076] The various power operation features used to calculate the distance of the basic multidimensional features do not include the bidding space. The bidding space is determined separately by the cross-feature sequence construction module and is used to construct the cross-feature sequence of the bidding space.

[0077] The cross-feature sequence construction module is used to determine the electricity price feature sequence and the bidding space feature sequence corresponding to each historical day, calculate the deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence for each time period, obtain the cross-feature value corresponding to each time period, and form the bidding space cross-feature sequence corresponding to the historical day by the cross-feature values ​​corresponding to each time period.

[0078] Specifically, regarding the first For each historical day, the cross-feature sequence construction module extracts the electricity price feature values ​​corresponding to each time period of the historical day from the historical dataset, and assembles the electricity price feature sequence corresponding to the historical day according to the time period order.

[0079] Regarding the first The first historical day For each time period, the cross-feature sequence construction module determines the corresponding bidding space feature value according to the following formula:

[0080] in, Indicates the first The first historical day in China The characteristic value of the bidding space corresponding to the time period, Indicates the first The first historical day in China System load forecast for the time period Indicates the first The first historical day in China The power load of the tie line during the time period, Indicates the first The first historical day in China Wind power output forecast for the period Indicates the first The first historical day in China Forecasted photovoltaic output for the period This indicates that new energy vehicles did not submit bids for a certain percentage of their participation, and , Indicates the first The first historical day in China Total output of non-market-based generating units during a given period.

[0081] The cross-feature sequence construction modules are arranged in time period order. The bidding space feature values ​​corresponding to each time period in a historical day are used to obtain the bidding space feature sequence corresponding to that historical day.

[0082] For each historical day, the cross-feature sequence construction module processes the electricity price feature sequence and bidding space feature sequence corresponding to that historical day using the maximum-minimum normalization method to obtain the normalized electricity price feature value and normalized bidding space feature value for each time period.

[0083] No. The first historical day in China The normalized electricity price characteristic value and the normalized bidding space characteristic value corresponding to the time period are as follows:

[0084]

[0085] in, Indicates the first The first historical day in China Electricity price characteristic value corresponding to the time period and These represent the minimum and maximum electricity price characteristic values ​​in the electricity price characteristic sequence corresponding to that historical day, respectively. and These represent the minimum and maximum bid space feature values ​​in the bid space feature sequence corresponding to that historical day, respectively.

[0086] After normalization, the cross-feature sequence construction module determines the first cross-feature sequence according to the following formula. The first historical day in China Cross-feature values ​​corresponding to the time period:

[0087] Calculate the cross-feature values ​​corresponding to each time period in the historical day in turn, and arrange them in the order of time periods to obtain the bidding space cross-feature sequence corresponding to the historical day.

[0088] The cross-weight determination module is used to determine the cross-weight factor of the bidding space corresponding to the historical day based on the degree of dispersion of the cross-feature values ​​of each time period in the bidding space cross-feature sequence relative to the sequence mean.

[0089] Specifically, regarding the first For each historical day, the cross-weight determination module first determines the sequence mean corresponding to that historical day according to the following formula:

[0090] Then, the cross-weight determination module determines the cross-weight factor of the bidding space corresponding to that historical day according to the following formula:

[0091] in, Indicates the first The first historical day in China Cross-feature values ​​corresponding to time periods Indicates the first The series mean corresponding to each historical day This indicates the total number of time periods in a single day. Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day.

[0092] When the cross-feature values ​​in the bidding space cross-feature sequence are relatively concentrated, the bidding space cross-weight factor is close to 1; when the dispersion of each cross-feature value relative to the sequence mean increases, the bidding space cross-weight factor increases accordingly. The cross-weight determination module processes multiple historical days in the same way to obtain the bidding space cross-weight factor corresponding to each historical day.

[0093] The distance correction module is used to correct the basic multidimensional feature distances by utilizing the cross-weighting factors of the bidding space corresponding to each historical day, so as to obtain the comprehensive distance index corresponding to each historical day.

[0094] Specifically, regarding the first For each historical day, the distance correction module corrects the distance of the underlying multidimensional feature corresponding to that historical day according to the following formula:

[0095] in, Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day Indicates the first The basic multidimensional feature distance corresponding to each historical day Indicates the first The comprehensive distance index corresponding to each historical day.

[0096] When the time-period deviation between the bidding space and the electricity price fluctuates significantly on a given historical day, the cross-weighting factor of the bidding space corresponding to that historical day increases, thereby increasing the comprehensive distance index corresponding to that historical day. Therefore, the distance correction module can use the stability of the cross-relationship between the bidding space and the electricity price to correct the distance of the basic multi-dimensional features.

[0097] The sample set construction module is used to select target historical days from multiple historical days based on the comprehensive distance index corresponding to each historical day, and construct the electricity price prediction sample set corresponding to the day to be predicted based on the multidimensional power operation characteristic data and electricity price data of the target historical days.

[0098] Specifically, the sample set construction module sorts multiple historical days in ascending order of comprehensive distance index, compares the comprehensive distance index corresponding to each historical day with a preset similarity index threshold, and determines the historical days whose comprehensive distance index is less than the similarity index threshold as target historical days.

[0099] After the target historical days are selected, the sample set construction module collects multidimensional power operation characteristic data and electricity price data for each target historical day according to the correspondence between each target historical day and historical data, and maintains the correspondence between multidimensional power operation characteristic data and electricity price data for the same period in the same target historical day, thus obtaining the electricity price prediction sample set corresponding to the day to be predicted.

[0100] In this embodiment, the basic multidimensional feature distances corresponding to each historical day output by the basic distance calculation module, and the bidding space cross-weight factors corresponding to each historical day output by the cross-weight determination module, are both used as inputs to the distance correction module. The distance correction module outputs the comprehensive distance index corresponding to each historical day and provides the comprehensive distance index to the sample set construction module, which then completes the selection of target historical days and the construction of the electricity price prediction sample set.

[0101] It should be noted that the above module division is a logical division based on function. In practical applications, two or more modules can be integrated as needed, or the function of a module can be divided into multiple functional units, as long as the same or corresponding processing procedure as in Embodiment 1 can be achieved. Other specific implementation methods of each module can be referred to the description of the corresponding method steps in Embodiment 1, and will not be repeated here.

[0102] Example 3 like Figure 3 As shown, the present invention also provides an electronic device 100 for implementing a method for constructing electricity price samples based on bidding space cross weights.

[0103] The electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on at least one processor 102, and at least one communication bus 104.

[0104] The memory 101 can be used to store the computer program 103. The processor 102 implements the method for constructing electricity price samples based on the cross weight of bidding space in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101.

[0105] The memory 101 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0106] At least one processor 102 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 102 may be a microprocessor or any conventional processor. Processor 102 is the control center of electronic device 100, connecting various parts of electronic device 100 via various interfaces and lines.

[0107] The memory 101 in the electronic device 100 stores multiple instructions to implement a method for constructing electricity price samples based on the cross-weight of the bidding space, and the processor 102 can execute multiple instructions to achieve the following: Acquire multidimensional power operation characteristic data for the day to be predicted and multiple historical days, and calculate the basic multidimensional feature distance between the day to be predicted and each of the historical days respectively; For each historical day, the electricity price feature sequence and the bidding space feature sequence corresponding to that historical day are determined. The deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence is calculated for each time period to obtain the cross feature value corresponding to each time period. The cross feature values ​​corresponding to each time period are used to form the bidding space cross feature sequence corresponding to that historical day. Based on the degree of dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean, the bidding space cross weight factor corresponding to the historical day is determined. By using the cross-weighting factor of the bidding space corresponding to each historical day to correct the corresponding basic multidimensional feature distance, the comprehensive distance index corresponding to each historical day is obtained. Based on the comprehensive distance index corresponding to each of the historical days, a target historical day is selected from multiple historical days, and based on the multidimensional power operation characteristic data and electricity price data of the target historical day, an electricity price prediction sample set corresponding to the day to be predicted is constructed.

[0108] Example 4 If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, and read-only memory (ROM).

[0109] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0110] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0111] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0113] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing electricity price samples based on cross-weighting of bidding space, characterized in that, include: Acquire multidimensional power operation characteristic data for the day to be predicted and multiple historical days, and calculate the basic multidimensional feature distance between the day to be predicted and each of the historical days respectively; For each historical day, the electricity price feature sequence and the bidding space feature sequence corresponding to that historical day are determined. The deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence is calculated for each time period to obtain the cross feature value corresponding to each time period. The cross feature value corresponding to each time period is used to form the bidding space cross feature sequence corresponding to that historical day. Based on the degree of dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean, the bidding space cross weight factor corresponding to the historical day is determined; By using the cross-weighting factor of the bidding space corresponding to each historical day to correct the corresponding basic multidimensional feature distance, the comprehensive distance index corresponding to each historical day is obtained. Based on the comprehensive distance index corresponding to each of the historical days, a target historical day is selected from multiple historical days, and based on the multidimensional power operation characteristic data and electricity price data of the target historical day, an electricity price prediction sample set corresponding to the day to be predicted is constructed.

2. The method for constructing electricity price samples based on cross-weighted bidding space according to claim 1, characterized in that, The step of calculating the deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence on a time-period basis includes: For each historical day, the electricity price feature sequence and the bidding space feature sequence are processed using the maximum-minimum normalization method to obtain the normalized electricity price feature value and the normalized bidding space feature value corresponding to each time period. The cross-feature values ​​corresponding to each time period in the historical day are determined according to the following formula: in, Indicates the first The first historical day in China The normalized electricity price characteristic value corresponding to the time period, Indicates the first The first historical day in China Normalized auction space eigenvalues ​​corresponding to the time period Indicates the first The first historical day in China Cross-feature values ​​corresponding to time periods; The steps for determining the bidding space cross weight factor corresponding to the historical day based on the dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean include: Based on the crossover characteristic values ​​corresponding to each time period in the historical day, the sequence mean and the bidding space crossover weight factor are determined according to the following formula: in, Indicates the first The series mean corresponding to each historical day This indicates the total number of time periods in a single day. Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day.

3. The method for constructing electricity price samples based on cross-weighted bidding space according to claim 1, characterized in that, The step of correcting the basic multidimensional feature distance corresponding to each historical day using the bidding space cross-weighting factor includes: Using the cross-weighting factor of the bidding space corresponding to the same historical day, the distance of the basic multidimensional features corresponding to that historical day is corrected according to the following formula: in, Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day Indicates the first The basic multidimensional feature distance corresponding to each historical day Indicates the first The comprehensive distance index corresponding to each historical day; The steps of selecting a target historical day from multiple historical days based on the comprehensive distance index corresponding to each historical day, and constructing a sample set for electricity price prediction corresponding to the day to be predicted based on the multidimensional power operation characteristic data and electricity price data of the target historical day include: The historical days are sorted in ascending order of the comprehensive distance index, and the historical days whose comprehensive distance index is less than the similarity index threshold are determined as the target historical days. Based on the multidimensional power operation characteristic data and electricity price data of each target historical day, the electricity price prediction sample set is obtained.

4. The method for constructing electricity price samples based on cross-weighted bidding space according to claim 1, characterized in that, The steps for determining the bidding space feature sequence corresponding to the historical day include: Regarding the first The first historical day For each time period, the corresponding bidding space characteristic value is determined according to the following formula: in, Indicates the first The first historical day in China The characteristic value of the bidding space corresponding to the time period, Indicates the first The first historical day in China System load forecast for the time period Indicates the first The first historical day in China The power load of the tie line during the time period, Indicates the first The first historical day in China Wind power output forecast for the period Indicates the first The first historical day in China Forecasted photovoltaic output for the period This indicates the proportion of new energy vehicles that did not submit bids. Indicates the first The first historical day in China Total output of non-market-based generating units during the specified time period; Arrange the bidding space feature values ​​corresponding to each time period of the historical day in chronological order to obtain the bidding space feature sequence corresponding to the historical day.

5. A device for constructing electricity price samples based on cross-weighted bidding space, characterized in that, include: The basic distance calculation module is used to acquire multi-dimensional power operation characteristic data of the day to be predicted and multiple historical days, and to calculate the basic multi-dimensional characteristic distance between the day to be predicted and each of the historical days respectively. The cross-feature sequence construction module is used to determine the electricity price feature sequence and the bidding space feature sequence corresponding to each historical day, calculate the deviation between the electricity price feature value in the electricity price feature sequence and the bidding space feature value in the bidding space feature sequence for each time period, obtain the cross-feature value corresponding to each time period, and form the bidding space cross-feature sequence corresponding to the historical day by the cross-feature values ​​corresponding to each time period. The cross weight determination module is used to determine the cross weight factor of the bidding space corresponding to the historical day based on the degree of dispersion of the cross feature values ​​of each time period in the bidding space cross feature sequence relative to the sequence mean. The distance correction module is used to correct the corresponding basic multidimensional feature distance using the bidding space cross-weighting factor corresponding to each historical day, so as to obtain the comprehensive distance index corresponding to each historical day. The sample set construction module is used to select target historical days from multiple historical days based on the comprehensive distance index corresponding to each historical day, and construct the electricity price prediction sample set corresponding to the day to be predicted based on the multidimensional power operation characteristic data and electricity price data of the target historical days.

6. The electricity price sample construction device based on bidding space cross-weights according to claim 5, characterized in that, The cross-feature sequence construction module is specifically used to process the electricity price feature sequence and the bidding space feature sequence for each historical day using a maximum-minimum normalization method, to obtain the normalized electricity price feature value and the normalized bidding space feature value corresponding to each time period, and to determine the cross-feature value corresponding to each time period in the historical day according to the following formula: in, Indicates the first The first historical day in China The normalized electricity price characteristic value corresponding to the time period, Indicates the first The first historical day in China Normalized auction space eigenvalues ​​corresponding to the time period Indicates the first The first historical day in China Cross-feature values ​​corresponding to time periods; The cross-weight determination module is specifically used to determine the sequence mean and the bidding space cross-weight factor according to the cross-feature values ​​corresponding to each time period in the historical day, using the following formula: in, Indicates the first The series mean corresponding to each historical day This indicates the total number of time periods in a single day. Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day.

7. The electricity price sample construction device based on bidding space cross-weights according to claim 5, characterized in that, The distance correction module is specifically used to correct the basic multidimensional feature distance corresponding to the same historical day using the bidding space cross-weight factor, according to the following formula: in, Indicates the first The cross-weighting factor of the bidding space corresponding to each historical day Indicates the first The basic multidimensional feature distance corresponding to each historical day Indicates the first The comprehensive distance index corresponding to each historical day; The sample set construction module is specifically used to sort multiple historical days in ascending order of the comprehensive distance index, determine the historical days whose comprehensive distance index is less than the similarity index threshold as the target historical days, and obtain the electricity price prediction sample set based on the multidimensional power operation characteristic data and electricity price data of each target historical day.

8. The electricity price sample construction device based on bidding space cross-weights according to claim 5, characterized in that, The cross-feature sequence construction module is specifically used for: Regarding the first The first historical day For each time period, the corresponding bidding space characteristic value is determined according to the following formula: in, Indicates the first The first historical day in China The characteristic value of the bidding space corresponding to the time period, Indicates the first The first historical day in China System load forecast for the time period Indicates the first The first historical day in China The power load of the tie line during the time period, Indicates the first The first historical day in China Wind power output forecast for the period Indicates the first The first historical day in China Forecasted photovoltaic output for the period This indicates the proportion of new energy vehicles that did not submit bids. Indicates the first The first historical day in China Total output of non-market-based generating units during the specified time period; Arrange the bidding space feature values ​​corresponding to each time period of the historical day in chronological order to obtain the bidding space feature sequence corresponding to the historical day.

9. An electronic device, characterized in that, It includes a processor and a memory, the processor being used to execute a computer program stored in the memory to implement the method for constructing electricity price samples based on the cross-weight of the bidding space as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, which, when executed by a processor, implements the method for constructing electricity price samples based on cross-weighted bidding space as described in any one of claims 1 to 4.

Images