An intelligent acquisition method for charging pile load data

Abstract

This application relates to the field of data processing, and provides an intelligent acquisition method for charging pile load data, including: using a mean filtering algorithm to process a first data set according to a first filtering window to obtain a second data set; wherein, the first data set includes the load data of the current charging pile collected in the current detection period; calculating the filtering effect corresponding to the second data set; if the filtering effect is less than a preset value, adjusting the first filtering window to a second filtering window, and processing the first data set based on the adjusted second filtering window until the filtering effect is greater than or equal to the preset value. This method analyzes in combination with a specific scenario, and based on the filtering and denoising effect, analyzes whether the currently used filtering window is appropriate, so as to obtain a more ideal filtering and denoising effect.

Description

Technical Field

[0001] The present application relates to the field of data processing, and particularly to an intelligent acquisition method for charging pile load data. Background Art

[0002] A large number of new energy vehicles are connected to the power grid simultaneously during the same period to replenish electric energy. The load during charging of new energy vehicles is superimposed on the original load in the power grid system, resulting in a load peak, which poses a new challenge to the current bearing capacity of the substation. At the same time, it is necessary to consider the improvement and expansion of the substation scale, which requires a lot of manpower, financial resources and material resources. The load superimposition may also cause a series of power quality problems such as voltage offset, voltage fluctuation, line overload, three-phase imbalance, unbalance and network loss. Therefore, an intelligent acquisition method for charging pile load data is needed to facilitate the analysis of charging pile load data and solve the above problems.

[0003] In the prior art, when preprocessing the collected charging pile load data using the mean filtering denoising algorithm, since different denoising windows may result in different denoising effects, there may still be noise after denoising. Therefore, the existing method has an unsatisfactory denoising effect on the collected charging pile load data. Summary of the Invention

[0004] The present invention provides an intelligent acquisition method for charging pile load data. The method analyzes in combination with specific scenarios and determines whether the currently used filtering window is appropriate based on the filtering denoising effect, so as to obtain a more ideal filtering denoising effect.

[0005] In a first aspect, the present application provides an intelligent acquisition method for charging pile load data, including:

[0006] Processing a first data set according to a first filtering window by using a mean filtering algorithm to obtain a second data set; wherein, the first data set includes the load data of the current charging pile collected in the current detection period;

[0007] Calculating the filtering effect corresponding to the second data set;

[0008] If the filtering effect is less than a preset value, adjusting the first filtering window to a second filtering window, and processing the first data set based on the adjusted second filtering window until the filtering effect is greater than or equal to the preset value.

[0009] Optionally, calculating the filtering effect corresponding to the second data set includes:

[0010] Determining the possibility that each data in the second data set is noise data;

[0011] Determine the filtering effect corresponding to the second data set according to the possibility that each data in the second data set is noise data.

[0012] Optionally, determining the possibility that each data in the second data set is noise data includes:

[0013] Calculating a first possibility that each data in the second data set is noise data based on the relationship between the second data set of the current charging pile and the reference data set of the reference charging pile within the current detection period; the reference charging pile is other charging piles in the current charging network except the current charging pile;

[0014] Calculating a second possibility that each data in the second data set is noise data based on the relationship between the current data and other data except the current data in the second data set of the current charging pile within the current detection period;

[0015] Calculating a third possibility that each data in the second data set is noise data based on the relationship between the historical data set of the current charging pile in the historical detection period before the current detection period and the second data set of the current detection period;

[0016] Determine the product of the first possibility, the second possibility, and the third possibility as the possibility that each data in the second data set is noise data.

[0017] Optionally, calculating a first possibility that each data in the second data set is noise data based on the relationship between the second data set of the current charging pile and the reference data set of the reference charging pile within the current detection period includes:

[0018] Determine the correlation between the current charging pile and each reference charging pile based on the relationship between the historical load data of the current charging pile and the historical reference load data of each reference charging pile; wherein, the historical load data and the historical reference load data are collected in the same time period;

[0019] Use a sliding window with a length of n and a step size of 1 to divide the second data set into multiple detection data segments, and determine the possibility that each detection data segment of the current charging pile is an abnormal data segment based on the correlation between the current charging pile and each reference charging pile;

[0020] Determine the first possibility that each data in the second data set is noise data based on the possibility that each detection data segment is an abnormal data segment.

[0021] Optionally, determining the correlation between the current charging pile and each reference charging pile based on the relationship between the historical load data of the current charging pile and the historical reference load data of each reference charging pile, including:

[0022] Dividing the historical load data of the current charging pile into h historical load data segments with a length of b, and dividing the historical reference load data into h historical reference load data segments with a length of b;

[0023] Calculating the first similarity between the charge-time curve of the historical load data segment in the same time period and the charge-time curve of the historical reference load data segment;

[0024] Calculating the first difference between the historical load data and the historical reference load data at the same time point in the historical load data segment and the historical reference load data segment in the same time period;

[0025] Determining the correlation between the current charging pile and each reference charging pile based on the first similarity, the first difference, the length of the historical load data segment, and the number of historical load data segments;

[0026] Wherein, the correlation calculation method is:

[0027]

[0028] Bx represents the correlation between the current charging pile and the reference charging pile, norm represents normalization, h represents the number of historical load data segments, Xs v represents the first similarity between the charge-time curve of the v-th historical load data segment of the current charging pile and the charge-time curve of the v-th historical reference load data segment of the reference charging pile, ΔP c,b represents the first difference between the c-th historical load data in the v-th historical load data segment of the current charging pile and the c-th historical reference load data in the v-th historical reference load data segment of the reference charging pile, and b represents the length of the historical load data segment, that is, the number of historical load data in the historical load data segment.

[0029] Optionally, determining the possibility that each detection data segment of the current charging pile is an abnormal data segment based on the correlation between the current charging pile and each reference charging pile, including:

[0030] Calculating the possibility that each detection data segment of the current charging pile is an abnormal data segment based on the correlation between the current charging pile and the reference charging pile, the second similarity between the charge-time curve of the detection data segment of the current charging pile and the charge-time curve of the reference data segment of the reference load data of the reference charging pile, and the number of reference charging piles; wherein, the reference data segment and the detection data segment are load data collected in the same time period.

[0031] Among them, the calculation method for the probability that each detection data segment of the current charging pile is an abnormal data segment is as follows:

[0032]

[0033] Kx represents the probability that the detection data segment of the current charging pile is an abnormal data segment, and Bx g represents the correlation between the current charging pile and the g-th reference charging pile, and Xs g represents the second similarity of the charge-time curve between the detection data segment of the current charging pile and the reference data segment of the g-th reference charging pile. q represents the number of reference charging piles.

[0034] Optionally, determining the first probability that each data in the second data set is noise data based on the probability that each detection data segment is an abnormal data segment includes:

[0035] Calculating the first permutation entropy corresponding to the detection data segment containing the current data;

[0036] Replacing the current data in the first permutation entropy with the mean value of the detection data segment containing the current data to obtain the second permutation entropy;

[0037] Calculating the first probability that each data in the second data set is noise data based on the number of detection data segments containing the current data, the probability that the detection data segment containing the current data is an abnormal data segment, the first permutation entropy, and the second permutation entropy;

[0038] Among them, the calculation method for the first probability that each data in the second data set is noise data is as follows:

[0039]

[0040] Among them, Qr represents the first probability that the current data in the second data set is noise data, b represents the number of detection data segments containing the current data, and Kx j represents the probability that the detection data segment containing the current data is an abnormal data segment, Pr j represents the first permutation entropy, Pr′ j represents the second permutation entropy, and exp represents the exponential function with the natural constant e as the base.

[0041] Optionally, calculating the second probability that each data in the second data set is noise data based on the relationship between the current data and other data except the current data in the second data set of the current charging pile during the current detection period includes:

[0042] Calculate the second difference between the u-th data and the current data in the current data segment centered on the current data;

[0043] Calculate the first average difference between the u-th data and its adjacent data, and calculate the second average difference between the current data and its adjacent data. Calculate the third difference between the first average difference and the second average difference;

[0044] Calculate the second probability that each data in the second data set is noise data based on the number of data in the current data segment, the second difference, and the third difference;

[0045] Wherein, the calculation method of the second probability is:

[0046]

[0047] Wherein, Tf represents the second probability that the current data in the second data set is noise data, b represents the number of data in the current data segment, ΔP u represents the second difference, and ΔP' u represents the third difference.

[0048] Optionally, calculate the third probability that each data in the second data set is noise data based on the relationship between the historical data set of the current charging pile in the historical detection period before the current detection period and the second data set of the current detection period, including:

[0049] Determine the initial probability that the current data collected at the current time is noise data based on the difference between the current data in the second data set and the historical data in the historical data set;

[0050] Determine the correlation of the load data on different days based on the differences between the historical data at the same time period on different days in the historical data set;

[0051] Determine the third probability that each data in the second data set is noise data based on the initial probability and the correlation of the load data on different days.

[0052] Optionally, determine the initial probability that the current data collected at the current time is noise data based on the difference between the current data in the second data set and the historical data in the historical data set, including:

[0053] Divide each historical data set into historical data segments of length b, and divide the second data set into current data segments of length b;

[0054] Calculate the average difference between the current data segment and the historical data segment corresponding to the current data segment in each historical data set;

[0055] Determine the initial possibility that the current data collected at the current time is noise data based on the average difference;

[0056] Among them, the calculation method of the initial possibility is:

[0057]

[0058] Among them, Rz u,w represents the initial possibility that the current data u is noise data on the w-th day. m represents the number of historical data sets, and each historical data set includes one-week load data. ΔPw i represents the average difference between the current data segment and the historical data segment corresponding to the current data segment time in the i-th historical data set on the w-th day;

[0059] Determine the correlation between the current data and the load data on different days based on the differences between the historical data in the same time period on different days in the historical data set, including:

[0060]

[0061] Among them, Rz u,r，w represents the correlation between the u-th load data segments in the same time period on the w-th day and the r-th day in the i-th historical data set. ΔPw i，r,w represents the average value of the differences in the load data between the w-th day and the r-th day in the i-th historical data set;

[0062] Determine the third possibility that each data in the second data set is noise data based on the initial possibility and the correlation of the load data on different days, including:

[0063]

[0064] Among them, Zx represents the third possibility that the data in the second data set is noise data. ΔPw r,w,u represents the similarity between the u-th load data segment on the r-th day and the u-th load data segment on the w-th day in the i-th historical data set. b represents the length of the load data segment, and e represents the number of days in a week except the current data.

[0065] The beneficial effects of this application are different from the prior art. An intelligent acquisition method for charging pile load data in this application includes: using a mean filtering algorithm to process a first data set according to a first filtering window to obtain a second data set; wherein, the first data set includes the load data of the current charging pile collected in the current detection period; calculating the filtering effect corresponding to the second data set; if the filtering effect is less than a preset value, adjusting the first filtering window to a second filtering window, and processing the first data set based on the adjusted second filtering window until the filtering effect is greater than or equal to the preset value. This method analyzes in combination with a specific scenario, and based on the filtering and denoising effect, analyzes whether the currently used filtering window is appropriate, so as to obtain a more ideal filtering and denoising effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic flowchart of the first embodiment of an intelligent acquisition method for charging pile load data of the present invention;

[0067] Figure 2 is Figure 1 a schematic flowchart of an embodiment of step S12;

[0068] Figure 3 is Figure 2 a schematic flowchart of an embodiment of step S21. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0070] The method of this application is used for collecting the load data of charging piles in a science and technology park. Specifically, when analyzing the load data of charging piles in a science and technology park to determine the current load status of the science and technology park, it is necessary to perform filtering processing on the collected charging pile load data. The prior art uses a mean filtering and denoising algorithm to filter the collected charging pile load data. However, since different denoising windows may result in different denoising effects, there may still be noise after denoising. Therefore, the existing method has an unsatisfactory denoising effect on the collected charging pile load data. The method of this application analyzes in combination with a specific scenario, and based on the filtering and denoising effect, analyzes whether the currently used filtering window is appropriate, so as to obtain a filtering window with a better denoising effect for filtering, thereby obtaining a more ideal filtering and denoising effect. The present application will be described in detail below with reference to the drawings and embodiments.

[0071] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of the first embodiment of the intelligent acquisition method for the load data of the charging pile of the present invention, specifically including:

[0072] Step S11: Process the first data set according to the first filtering window by using the mean filtering algorithm to obtain a second data set; wherein, the first data set includes the load data of the current charging pile collected in the current detection period.

[0073] An independent charging pile is equipped with sensors for monitoring the power load situation. The load data of the charging pile is mainly obtained by monitoring the load situation of the power line. The sensors then transmit the collected data to the data processing program of the collector for further analysis. Among them, the collected data all have time stamps, that is, the collection date and the day of the corresponding collection week corresponding to the current detection period. For example, the collected data is recorded as P w , where w represents the day of the current detection period corresponding to the collection, and the first data set finally collected is P = (P1, P2, P3, P4, P5, P6, P7).

[0074] During the process of collecting data, the load data of the charging pile may have noise due to external interference and other factors, so it is necessary to eliminate these noise data to avoid the result obtained from analyzing the collected data not meeting the expectation due to abnormal data. Specifically, the first data set is processed according to the first filtering window by using the mean filtering algorithm to obtain a second data set. Among them, the first filtering window is the preset filtering window in the mean filtering algorithm, generally with a size of 3; a too large filtering window can better remove noise, but it will reduce the information contained in the data, and a too small filtering window may lead to poor data denoising effect, and there will still be a large amount of interference noise after denoising. In order to avoid an unsatisfactory filtering effect or reducing the information contained in the data, the filtering effect of the second data set is analyzed in this application.

[0075] Step S12: Calculate the filtering effect corresponding to the second data set.

[0076] Please combine with Figure 2 , step S12 includes:

[0077] Step S21: Determine the possibility that each data in the second data set is noise data.

[0078] Specifically, please combine with Figure 3 , step S21 specifically includes:

[0079] Step S31: Calculate the first possibility that each data in the second data set is noise data based on the relationship between the second data set of the current charging pile and the reference data set of the reference charging pile within the current detection period.

[0080] Obtain the load data of multiple charging piles, and calculate the relationship between the second data set of the current charging pile and the reference data set of the reference charging pile. It should be noted that the reference charging pile is other charging piles in the current charging network except the current charging pile. It can be understood that since the load situation in the charging network such as in a science and technology park is calculated, the reference charging pile is the charging pile in the same usage state as the current charging pile at the same time.

[0081] In a specific embodiment, obtain the historical load data of the current charging pile and the historical reference load data of each reference charging pile collected in the same time period; determine the correlation between the current charging pile and each reference charging pile based on the relationship between the historical load data of the current charging pile and the historical reference load data of each reference charging pile collected in the same time period.

[0082] Specifically, divide the historical load data of the current charging pile into h historical load data segments with a length of b, and divide the historical reference load data into h historical reference load data segments with a length of b. It should be noted that the time of the data points in each historical load data segment and historical reference load data segment corresponds one by one.

[0083] Calculate the first similarity between the charge-time curve of the historical load data segment in the same time period and the charge-time curve of the historical reference load data segment. In a specific embodiment, the shape context algorithm can be used to calculate the first similarity between the charge-time curve of the historical load data segment in the same time period and the charge-time curve of the historical reference load data segment.

[0084] Calculate the first difference between the historical load data and the historical reference load data at the same time point in the historical load data segment and the historical reference load data segment in the same time period.

[0085] Determine the correlation between the current charging pile and each reference charging pile based on the first similarity, the first difference, the length of the historical load data segment, and the number of historical load data segments.

[0086] In a specific embodiment, the correlation calculation method is:

[0087]

[0088] Bx represents the correlation between the current charging pile and the reference charging pile, norm represents normalization, h represents the number of historical load data segments, Xsv It represents the first similarity, ΔP, between the charge-time curve of the v-th historical load data segment of the current charging pile and the charge-time curve of the v-th historical reference load data segment of the reference charging pile. c,v It represents the first difference between the c-th historical load data in the v-th historical load data segment of the current charging pile and the c-th historical reference load data in the v-th historical reference load data segment of the reference charging pile. b represents the length of the historical load data segment, that is, the number of historical load data in the historical load data segment.

[0089] In the above formula, when the first similarity Xs between the charge-time curve of the v-th historical load data segment of the current charging pile to be obtained and the charge-time curve of the v-th historical reference load data segment of the reference charging pile v is higher, the first difference ΔP between the c-th historical load data in the v-th historical load data segment of the current charging pile and the c-th historical reference load data in the v-th historical reference load data segment of the reference charging pile c,v is smaller, indicating that the correlation between the current charging pile and the reference charging pile is higher, and this correlation represents the degree of correlation.

[0090] The second data set is divided into multiple detection data segments by using a sliding window with a length of n and a step size of 1. For example, the second data set includes data 1, 2, 3, 4, 5, 6. If the length is set to 3 and the step size is 1, the obtained detection data segments are 1, 2, 3; 2, 3, 4; 3, 4, 5; 4, 5, 6. Among them, the number of detection data segments containing the current data 1 is 1 (1, 2, 3), the number of detection data segments containing the current data 2 is 2 (1, 2, 3; 2, 3, 4), the number of detection data segments containing the current data 3 is 3 (1, 2, 3; 2, 3, 4; 3, 4, 5), and others are not elaborated here.

[0091] Based on the correlation between the current charging pile and each reference charging pile, determine the possibility that each detection data segment of the current charging pile is an abnormal data segment. Specifically, based on the correlation between the current charging pile and the reference charging pile, the second similarity between the charge-time curve of the detection data segment of the current charging pile and the charge-time curve of the reference data segment of the reference load data of the reference charging pile, and the number of reference charging piles, calculate the possibility that each detection data segment of the current charging pile is an abnormal data segment. It should be noted that the reference data segment and the detection data segment are load data collected in the same time period;

[0092] In a specific embodiment, the calculation method for the possibility that each detection data segment of the current charging pile is an abnormal data segment is:

[0093]

[0094] Kx represents the possibility that the detected data segment of the current charging pile is an abnormal data segment, Bx g represents the correlation between the current charging pile and the g-th reference charging pile, Xs g represents the second similarity of the charge-time curve between the detected data segment of the current charging pile and the reference data segment of the g-th reference charging pile. q represents the number of reference charging piles.

[0095] In the above formula, if the detected data segment corresponding to the current charging pile is more similar to the reference data segment of the reference charging pile with a high degree of correlation (i.e., high correlation), it indicates that the possibility that the detected data segment corresponding to the current charging pile is an abnormal data segment is lower. On the contrary, it indicates that the possibility that the detected data segment corresponding to the current charging pile is an abnormal data segment is higher.

[0096] Determine the first possibility that each data in the second data set is noise data based on the possibility that each detected data segment is an abnormal data segment. Specifically, calculate the first permutation entropy corresponding to the detected data segment containing the current data; replace the current data in the first permutation entropy with the mean value of the detected data segment containing the current data to obtain the second permutation entropy; calculate the first possibility that each data in the second data set is noise data based on the number of detected data segments containing the current data, the possibility that the detected data segment containing the current data is an abnormal data segment, the first permutation entropy, and the second permutation entropy.

[0097] In one embodiment, the calculation method of the first possibility that each data in the second data set is noise data is:

[0098]

[0099] where Qr represents the first possibility that the current data in the second data set is noise data, b represents the number of detected data segments containing the current data, Kx j represents the possibility that the detected data segment containing the current data is an abnormal data segment, Pr j represents the first permutation entropy, Pr′ j represents the second permutation entropy, and exp represents the exponential function with the natural constant e as the base. (Pr j -Pr′ j ) represents the difference between the first permutation entropy and the second permutation entropy. If the difference between the first permutation entropy and the second permutation entropy is larger, it indicates that the first possibility that the current data is noise data is higher.

[0100] In the above formula, if the first permutation entropy corresponding to each detection data segment where the current data is located is smaller, and the second permutation entropy becomes larger after replacing the current data with the data segment mean, and the possibility that the corresponding detection data segment is an abnormal data segment is smaller, that is, the possibility of containing noise data is smaller, then the corresponding detection data segment is more credible, which means that the first possibility that the current data is noise data is smaller.

[0101] It should be noted that the first permutation entropy is calculated by the permutation entropy calculation method. Among them, the embedding dimension of the permutation entropy calculation method is 3, and the time delay is 3.

[0102] Step S32: Calculate the second possibility that each data in the second data set of the current charging pile is noise data based on the relationship between the current data and other data except the current data in the second data set within the current detection period.

[0103] Analyze according to the load curves in adjacent time periods, construct a data curve with the acquisition time as the horizontal axis and the corresponding charging pile load as the vertical axis. Calculate the second possibility that each data in the second data set of the current charging pile is noise data based on the relationship between the current data and other data except the current data in the second data set.

[0104] Specifically, calculate the second difference between the u-th data and the current data in the current data segment centered on the current data; calculate the first average difference between the u-th data and the adjacent data of the u-th data, and calculate the second average difference between the current data and the adjacent data of the current data, and calculate the third difference between the first average difference and the second average difference; calculate the second possibility that each data in the second data set is noise data based on the number of data in the current data segment, the second difference, and the third difference;

[0105] In a specific embodiment, the calculation method of the second possibility is:

[0106]

[0107] Among them, Tf represents the second possibility that the current data in the second data set is noise data, b represents the number of data in the current data segment, ΔP u represents the second difference, and ΔP' u represents the third difference.

[0108] In the above formula, if the difference between each data and the current data in the current data segment is larger, that is, the second difference ΔP u is larger, and the difference in the corresponding data change is also larger, that is, the third difference ΔP' u is larger, then it means that the second possibility that the current data is noise data is larger.

[0109] Step S33: Calculate a third possibility that each data in the second data set is noise data based on the relationship between the historical data set of the current charging pile in the historical detection period before the current detection period and the second data set in the current detection period.

[0110] Based on the difference between the current data in the second data set and the historical data in the historical data set, determine an initial possibility that the current data collected at the current time is noise data.

[0111] Since in the science and technology park, the charging pile load data has a strong correlation with the working hours of people in the park, such as going to work and getting off work, the charging pile load data can be analyzed according to a specific time period. Since the changes in the park are caused by the weekly changes, in this embodiment, the load change data corresponding to m historical detection periods adjacent to the current detection period is collected for analysis to determine an initial possibility that the current data collected at the current time is noise data. Each historical detection period corresponds to a historical data set. Collect and obtain the corresponding time information at the current time.

[0112] Specifically, divide each historical data set into historical data segments of length b, and divide the second data set into current data segments of length b; calculate the average difference between the current data segment and the historical data segment corresponding to the current data segment time in each historical data set; based on the average difference, determine an initial possibility that the current data collected at the current time is noise data.

[0113] Among them, the calculation method of the initial possibility is:

[0114]

[0115] Among them, Rz u,w represents the initial possibility that the current data u is noise data on the wth day, m represents the number of historical data sets, for example, m = 20, and each historical data set includes one-week load data, ΔPw i represents the average difference between the current data segment and the historical data segment corresponding to the current data segment time in the ith historical data set on the wth day. If the current data and the historical data are on the same day in different detection periods, for example, both on Wednesday, the higher the similarity between the current data segment and the historical data segment, the lower the possibility that the current data segment contains a noise data segment, that is, the lower the initial possibility that the current data collected at the current time is noise data.

[0116] Based on the differences between the historical data at the same time period on different days in the historical data set, determine the correlation of the load data on different days.

[0117] In one embodiment, the calculation method of the correlation of the load data on different days is:

[0118]

[0119] wherein, Rz u,r，w represents the correlation of the u-th load data segment in the same time period on the w-th day and the r-th day in the i-th historical data set, and ΔPw i，r,w represents the average value of the differences in load data between the w-th day and the r-th day in the i-th historical data set. It can be understood that the calculated ΔPw i，r,w is smaller, indicating that the load data on the r-th day is more correlated with the load data on the w-th day.

[0120] Determine the third possibility that each data in the second data set is noise data based on the initial possibility and the correlation of load data on different days. Specifically, for different working hours, such as Monday and Tuesday, there may also be a correlation. Therefore, the correlation degree between each day of the week and the day of the week where the currently analyzed load data is located can be analyzed according to the time of a detection cycle, so as to determine the third possibility that each data in the second data set is noise data.

[0121] In one embodiment, the calculation method for the third possibility that each data in the second data set is noise data is:

[0122]

[0123] wherein, Zx represents the third possibility that the data in the second data set is noise data, and ΔPw r,w,u represents the similarity between the u-th load data segment on the r-th day and the u-th load data segment on the w-th day in the i-th historical data set, b represents the length of the load data segment, e represents the number of days in a week except the current data, that is, e = 6, and Rz u,w represents the initial possibility that the current data u is noise data on the w-th day.

[0124] In the above formula, when the data of the currently calculated data segment is more similar to the data obtained in the corresponding time period on the same day of each week in the historical data segment, that is, the required Rz u,w is smaller, the correlation between the other days of the week and the w-th day is greater, and the correlation degree of the same data segment on different days under the current collection week is closer to the historical correlation degree, that is, the required Rz u,r，w -ΔPw r,w,u is smaller, indicating that the possibility that the current data is noise data is smaller.

[0125] Step S34: Determine the product of the first possibility, the second possibility, and the third possibility as the possibility that each data in the second data set is noise data.

[0126] Specifically, the product of the first possibility Qr, the second possibility Tf, and the third possibility Zx is determined as the possibility that each data in the second data set is noise data.

[0127] Step S22: Determine the filtering effect corresponding to the second data set according to the possibility that each data in the second data set is noise data.

[0128] Specifically, the reciprocal of the product of the first possibility Qr, the second possibility Tf, and the third possibility Zx is used as the filtering effect corresponding to the second data set.

[0129] Specifically, the calculation method of the filtering effect corresponding to the second data set is:

[0130]

[0131] where norm() represents the normalization function, ζ represents the number of data in the second data set, and Zx γ , Tf γ , Qr γ respectively represent the third possibility, the second possibility, and the first possibility that the γ-th data in the second data set is noise data.

[0132] Step S13: If the filtering effect is less than the preset value, adjust the first filtering window to the second filtering window, and process the first data set based on the adjusted second filtering window until the filtering effect is greater than or equal to the preset value.

[0133] Specifically, set the preset value η = 0.6, that is, when the required denoising effect Gx is greater than or equal to the preset value, it can be considered that a good denoising effect can be obtained by using the current window, that is, the size of the first filtering window, and the denoising result of the data obtained using this window size is used for subsequent analysis.

[0134] If the wave effect is less than the preset value, continue to adjust the size of the first filtering window until the above conditions are met.

[0135] It should be noted that the transformation size range of the first filtering window is 2k + 1, and the maximum value of k is 20.

[0136] The present invention combines the correlation between charging piles in the same area and the time relationship of using charging piles in the science and technology park. According to historical data analysis, the possibility that the currently collected data is noise is determined, and a preferred filtering window is obtained accordingly, which greatly increases the denoising effect of load data and enhances the accuracy of the collected data.

[0137] The above are only the embodiments of the present application, and do not limit the patent scope of the present application accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present application.

Claims

1. An intelligent acquisition method for charging pile load data, characterized in that, Including: Processing a first data set according to a first filtering window by using a mean filtering algorithm to obtain a second data set; wherein, the first data set includes load data of a current charging pile collected in a current detection period; Determining the correlation between the current charging pile and each reference charging pile based on the relationship between the historical load data of the current charging pile and the historical reference load data of each reference charging pile; wherein, the historical load data and the historical reference load data are collected in the same time period; Dividing the second data set into multiple detection data segments by using a sliding window with a length of n and a step size of 1, and determining the possibility that each detection data segment of the current charging pile is an abnormal data segment based on the correlation between the current charging pile and each reference charging pile; Determining a first possibility that each data in the second data set is noise data based on the possibility that each detection data segment is an abnormal data segment; Calculating a second difference between the u-th data and the current data in a current data segment centered on the current data; Calculating a first average difference between the u-th data and the adjacent data of the u-th data, and calculating a second average difference between the current data and the adjacent data of the current data, and calculating a third difference between the first average difference and the second average difference; Calculating a second possibility that each data in the second data set is noise data based on the number of data in the current data segment, the second difference, and the third difference; Determining an initial possibility that the current data collected at the current time is noise data based on the difference between the current data in the second data set and the historical data in the historical data set; Determining the correlation of the load data on different days based on the difference between the historical data in the historical data set at the same time period on different days; Determining a third possibility that each data in the second data set is noise data based on the initial possibility and the correlation of the load data on different days; Determining the product of the first possibility, the second possibility, and the third possibility as the possibility that each data in the second data set is noise data; determining the filtering effect corresponding to the second data set according to the possibility that each data in the second data set is noise data; If the filtering effect is less than a preset value, adjusting the first filtering window to a second filtering window, and processing the first data set based on the adjusted second filtering window until the filtering effect is greater than or equal to the preset value.

2. The intelligent acquisition method of charging pile load data according to claim 1, characterized in that Determining the correlation between the current charging pile and each reference charging pile based on the relationship between the historical load data of the current charging pile and the historical reference load data of each reference charging pile includes: Dividing the historical load data of the current charging pile into h historical load data segments with a length of b, and dividing the historical reference load data into h historical reference load data segments with a length of b; Calculating a first similarity between the charge-time curve of the historical load data segment in the same time period and the charge-time curve of the historical reference load data segment; Calculating a first difference between the historical load data and the historical reference load at the same time point in the historical load data segment and the historical reference load data segment in the same time period; Determine the correlation between the current charging pile and each reference charging pile based on the first similarity, the first difference, the length of the historical load data segment, and the number of historical load data segments; Wherein, the calculation method of the correlation is: Bx represents the correlation between the current charging pile and the reference charging pile, norm represents normalization, h represents the number of historical load data segments, and Xs v represents the first similarity between the charge-time curve of the v-th historical load data segment of the current charging pile and the charge-time curve of the v-th historical reference load data segment of the reference charging pile, and ΔP c,v represents the first difference between the c-th historical load data in the v-th historical load data segment of the current charging pile and the c-th historical reference load data in the v-th historical reference load data segment of the reference charging pile. b represents the length of the historical load data segment, that is, the number of historical load data in the historical load data segment.

3. The intelligent acquisition method of charging pile load data according to claim 1, characterized in that Determine the possibility that each detection data segment of the current charging pile is an abnormal data segment based on the correlation between the current charging pile and each reference charging pile, including: Calculate the possibility that each detection data segment of the current charging pile is an abnormal data segment based on the correlation between the current charging pile and the reference charging pile, the charge-time curve of the detection data segment of the current charging pile, the second similarity between the charge-time curve of the detection data segment of the current charging pile and the reference data segment of the reference load data of the reference charging pile, and the number of reference charging piles; wherein, the reference data segment and the detection data segment are load data collected in the same time period; Wherein, the calculation method of the possibility that each detection data segment of the current charging pile is an abnormal data segment is: Kx represents the possibility that the detected data segment of the current charging pile is an abnormal data segment, Bx g represents the correlation between the current charging pile and the g-th reference charging pile, Xs g represents the second similarity of the charge-time curve between the detected data segment of the current charging pile and the charge-time curve of the reference data segment of the g-th reference charging pile, where q represents the number of reference charging piles.

4. The intelligent acquisition method of charging pile load data according to claim 1, characterized in that, Determine the first possibility that each data in the second data set is noise data based on the possibility that each detection data segment is an abnormal data segment, including: Calculate the first permutation entropy corresponding to the detection data segment containing the current data; Replace the current data in the first permutation entropy with the mean value of the detection data segment containing the current data to obtain the second permutation entropy; Calculate the first possibility that each data in the second data set is noise data based on the number of detection data segments containing the current data, the possibility that the detection data segment containing the current data is an abnormal data segment, the first permutation entropy, and the second permutation entropy; Wherein, the calculation method of the first possibility that each data in the second data set is noise data is: Among them, Qr represents the first possibility that the current data in the second data set is noise data, b represents the number of detection data segments containing the current data, and Kx j represents the possibility that the detection data segment containing the current data is an abnormal data segment, Pr j represents the first permutation entropy, Pr′ j represents the second permutation entropy, and exp represents the exponential function with the natural constant e as the base.

5. The intelligent acquisition method of charging pile load data according to claim 1, characterized in that, The calculation method of the second possibility is: Among them, Tf represents the second possibility that the current data in the second data set is noise data, b represents the number of data in the current data segment, and ΔP u represents the second difference, and ΔP’ u represents the third difference.

6. The intelligent acquisition method of charging pile load data according to claim 1, characterized in that Determine the initial possibility that the current data collected at the current time is noise data based on the difference between the current data in the second data set and the historical data in the historical data set, including: Divide each historical data set into historical data segments with a length of b, and divide the second data set into current data segments with a length of b; Calculate the average difference between the current data segment and the historical data segment corresponding to the current data segment in each historical data set in terms of time; Determine the initial possibility that the current data collected at the current time is noise data based on the average difference; Wherein, the calculation method of the initial possibility is: Among them, Rz u,w represents the initial possibility that the current data u is noise data on the w-th day, m represents the number of historical data sets, and each historical data set includes one-week load data, ΔPw i represents the average difference between the current data segment and the historical data segment corresponding to the current data segment time in the i-th historical data set on the w-th day; Determine the correlation between the current data and the load data on different days based on the differences between the historical data at the same time period on different days in the historical data set, including: Among them, Rz u,r，w represents the correlation of the u-th load data segment in the same time period on the w-th day and the r-th day in the i-th historical data set, and ΔPw i，r,w represents the average value of the differences in load data between the w-th day and the r-th day in the i-th historical data set; Determine the third possibility that each data in the second data set is noise data based on the initial possibility and the correlation between the load data on different days, including: Among them, Zx represents the third possibility that the data in the second data set is noise data, and ΔPw r,w,u represents the similarity between the u-th load data segment on the r-th day and the u-th load data segment on the w-th day in the i-th historical data set. b represents the length of the load data segment, and e represents the number of days other than the current data within a week.

Images