Analysis Method of Typical Enterprise Power Consumption Patterns Based on Multidimensional Isolate-Detect Change Point Detection

Abstract

The present invention proposes an analysis method for typical enterprise electricity consumption patterns based on multi-dimensional Isolate-Detect change point detection. First, data cleaning is performed on the collected industrial electricity consumption time series data; then, seasonal decomposition is carried out on the data after data cleaning to eliminate the periodic patterns of the data and extract the residual terms of the data; multi-dimensional change point detection is performed on the residual terms obtained after decomposition based on multi-dimensional Isolate-Detect, and only a single change point is detected in each divided time interval, making the change point detection in the multi-dimensional electricity consumption curve more accurate and the calculation efficiency higher; then, a possible set of change points is obtained by combining the mean test statistic and the maximum test statistic; the multi-sequence fusion is used to screen the possible set of change points for the detected set of change points to determine the final change point positions of the electricity consumption data; finally, based on the obtained change point positions, the standardized time series data is divided to extract the typical electricity consumption behaviors of the electricity consumption.

Description

Technical Field

[0001] The present invention relates to the technical field of power data analysis, and in particular to a method for analyzing typical enterprise power consumption patterns based on multi-dimensional Isolate-Detect change point detection. Background Art

[0002] With the continuous improvement of the informatization level of the power grid and the rapid development of digitalization and intelligence, the power industry has entered the era of big power data. As an energy system relied on by economic development and human life, a large amount of data with rapid growth and rich types will be generated during the operation of the power system. Analyzing power consumption behavior can discover the laws, relationships, trends, etc. in big power data to obtain power consumption characteristics. Therefore, detecting common change points and extracting typical power consumption behaviors for each industry can better analyze and understand power consumption, understand power consumption characteristics and behaviors, and provide strong support for power supply and management.

[0003] However, industrial power consumption data shows disadvantages such as large power consumption and complex power consumption rules for each industry. At the same time, the data also has characteristics such as periodicity and seasonality, which bring challenges to power consumption behavior analysis. Today's power data has the characteristics of high dimension, multiple types and large volume, posing higher requirements for data analysis technology. Summary of the Invention

[0004] Therefore, aiming at the gaps and deficiencies in the prior art, the present invention proposes a method for analyzing typical enterprise power consumption patterns based on multi-dimensional Isolate-Detect change point detection, which is used to improve the shortcomings of traditional power consumption behavior analysis models that do not consider temporal and phased changes. For industrial power consumption time series data, a multi-dimensional Isolate-Detect model is established to detect the common change points of the industry and extract typical power consumption behaviors. Based on historical power consumption data, one day of a weekday and one day of a weekend are extracted to detect the differences in power consumption behaviors between weekdays and weekends; the multi-dimensional Isolated-Detect algorithm is used for change point detection, improving the accuracy of change point detection; finally, based on the detected change point positions, typical power consumption behaviors are extracted by segmenting the power consumption time series data.

[0005] The main steps include data collection, data cleaning, residual item extraction, change point detection, change point screening, and extraction of typical electricity consumption behavior. First, data cleaning is performed according to the collected industrial electricity consumption time series data; then the cleaned data is seasonally decomposed to eliminate the periodic pattern of the data and extract the residual items of the data; based on multidimensional Isolate-Detect, multidimensional change point detection is performed on the residual items obtained after decomposition, and only a single change point is detected in each divided time interval, so that the change point detection in the multidimensional electricity consumption curve is more accurate and the calculation efficiency is higher. Combine the mean test statistic and the maximum test statistic to obtain a possible change point set; the detected change point set is screened by multi-sequence fusion to determine the final change point position of the electricity consumption data; finally, the standardized time series data is divided based on the obtained change point position to extract the typical electricity consumption behavior of electricity consumption. The multidimensional Isolate-Detect method proposed in the present invention takes into account the serial correlation between multidimensional data, can effectively detect the common change points of multidimensional data, and improves the accuracy of multidimensional change point detection. This will provide a more realistic and objective time-series electricity usage scenario, allowing for the formulation of a more reasonable electricity usage strategy in subsequent planning.

[0006] The specific technical solutions are as follows:

[0007] A method for analyzing typical patterns of enterprise electricity consumption based on multi-dimensional Isolate-Detect and multi-change point detection, characterized in that: first, data cleaning is performed according to the collected industrial electricity consumption time series data; then, seasonal decomposition is performed on the cleaned data to eliminate the periodic pattern of the data and extract the residual term of the data; multi-dimensional change point detection is performed on the residual term obtained after decomposition based on multi-dimensional Isolate-Detect, and only a single change point is detected in each divided time interval, so that the change point detection in the multi-dimensional electricity consumption curve is more accurate and the calculation efficiency is higher; then, a possible change point set is obtained by combining the mean test statistic and the maximum test statistic; the detected change point set is screened by multi-sequence fusion to determine the final change point position of the electricity consumption data; finally, the standardized time series data is divided based on the obtained change point position to extract the typical electricity consumption behavior.

[0008] Further, the following steps are included:

[0009] Step 1: Collect the time series data of electricity consumption of each industry;

[0010] Step 2: Fill missing values ​​in the electricity consumption time series data of each industry, extract electricity consumption data on a certain day on a weekday or a certain day on a weekend, use the 3σ principle to eliminate outliers, and then perform standardization;

[0011] Step 3: Smooth the cleaned data first, and then perform seasonal decomposition on the smoothed data to eliminate the periodicity and trend of the data to obtain the residual term of the smoothed data;

[0012] Step 4: Calculate the CUSUM statistic of the residual item of each industry electricity consumption data at each time point, and then calculate the mean and maximum value of the CUSUM statistic at each time point, which are recorded as M test statistics and T test statistics respectively;

[0013] Step 5: Use the Isolate-Detect algorithm to detect the common change points of the electricity consumption data of each industry. Preset the threshold according to the threshold formula of the Isolate-Detect algorithm, determine the time point when the M test statistic exceeds the threshold as the change point, put the obtained change point into the change point set, recorded as the change point set CP_M; determine the time point when the T test statistic exceeds the threshold as the change point, put the obtained change point into the change point set, recorded as the change point set CP_T;

[0014] Step 6: Record the change point set obtained by data analysis as the change point set CP, take out the two elements whose absolute difference value is less than 3 from the set CP_M and the set CP_T, select the smaller value and put it into the set CP to obtain the final change point set;

[0015] Step 7: Segment the standardized electricity consumption time series data based on the obtained change point positions, calculate the average electricity consumption of each industry in each segment of the time series data, and obtain the typical electricity consumption behavior of each segment of the time series data.

[0016] Furthermore, in step 2, after filling the missing values ​​in the electricity consumption data of each industry, the electricity consumption data of a certain day on a weekday and the electricity consumption data of a certain day on a weekend are extracted for subsequent data analysis, and the differences in industrial electricity consumption behaviors on weekdays and weekends are analyzed; the 3σ principle is used to eliminate outliers to avoid the impact of outliers on subsequent electricity consumption behavior analysis.

[0017] Furthermore, in step 3, the cleaned data is smoothed using a Savitzky-Golay filter. When the loss function reaches a minimum value, the fitting effect of the original data is optimal. The fitting value of the smoothed original data is obtained through a sliding window to effectively reduce the noise of the data.

[0018] Furthermore, in step 3, for the smoothed data, an additive model is used to perform trend decomposition, and the smoothed power consumption data is decomposed into a trend part, a periodic part, and a residual part to eliminate the periodicity and trend of the data, which is expressed as follows:

[0019] X i,t =T i,t +S i,t +Ci,t +I i,t , (7)

[0020] Among them, T i represents the long-term time trend of the i-th industry, S i represents the seasonal time trend of the i-th industry, C i represents the cyclical time trend of the i-th industry, I i represents the remaining residual term of the i-th industry.

[0021] Furthermore, in step 4, using the data residual term obtained after decomposition, calculate the mean value of the CUSUM statistic at each time point. The formula is as follows:

[0022]

[0023]

[0024] Among them, i refers to the i-th industry, s refers to the starting point of the detection interval, e refers to the ending point of the detection interval, b refers to the detection time point, n refers to the total length of the detection interval, I i,t refers to the smoothed data residual term of the i-th industry at time t, refers to the value of the CUSUM statistic of the i-th industry at the b time point within the detection interval [s, e], p refers to the total number of data dimensions, refers to the mean value of the CUSUM statistic at the b time point within the detection interval [s, e], denoted as

[0025] Furthermore, in step 4, using the data residual term obtained after decomposition, calculate the maximum value of the CUSUM statistic at each time point. The formula is as follows:

[0026]

[0027] Among them, i refers to the i-th industry, s refers to the starting point of the detection interval, e refers to the ending point of the detection interval, b refers to the detection time point, n refers to the total length of the detection interval, I i,t refers to the smoothed data residual term of the i-th industry at time t, refers to the value of the CUSUM statistic of the i-th industry at the b time point within the detection interval [s, e], p refers to the total number of data dimensions, refers to the maximum value of the CUSUM statistic at the b time point within the detection interval [s, e].

[0028] Further, in step 5, based on the calculated M-test statistic and T-test statistic, the Isolate-Detect algorithm is used to detect the common change points of the electricity consumption of each industry:

[0029] First, create the interval to be detected. For a data sequence of length T, first set a positive constant λ T , and then create two sets of ordered left and right extended intervals with K = [T / λ T ; the jth right extended interval is R j = [1, min{jλ T , T}], and the jth left extended interval is L j = [max{1, T - jλ T + 1}, T]; collect these intervals in the ordered set S RL = {R1, L1, R2, L2,..., R K , L K}; then identify the point with the largest test statistic value in R1;

[0030] Based on the obtained test statistic, set a threshold, and the calculation formula of the threshold is as follows:

[0031]

[0032] In the formula, σ is the standard deviation of the input data, and C is the given parameter value; compare the threshold with the test statistic to determine whether this time point is a change point: if the value of the test statistic exceeds the threshold, the corresponding point is regarded as a change point; if it does not exceed the threshold, continue to detect the next interval of S RL .

[0033] Further, in step 7, based on the detected change point positions, segment the standardized time series data, calculate the mean value of the electricity consumption of each industry in each segment, and extract the typical electricity consumption scenarios of each segment of time series data. The calculation formula for the mean value of the electricity consumption of the ith industry is as follows:

[0034]

[0035] In the formula, τ k refers to the position of the kth change point, refers to the standardized electricity consumption value of the ith industry at the position of the (k + 1)th change point.

[0036] Compared with the prior art, the beneficial effects of the present invention and its preferred solutions at least include:

[0037] 1. Considering that there are differences in electricity consumption behaviors on weekdays and weekends, extract the data of a certain day on weekdays and a certain day on weekends for subsequent data analysis, and distinguish the different electricity consumption characteristics between weekdays and weekends;

[0038] 2. Trend decomposition is performed on the electricity consumption time series data. Considering that the cycle may make the abnormal detection effect of the model unstable, the residual sequence without trend and cycle terms is used for modeling, which improves the accuracy of subsequent modeling.

[0039] 3. By calculating the CUSUM mean statistic, the influence of all dimensions on the change point detection is considered.

[0040] 4. Calculate the CUSUM maximum statistic. Only the sequence with the largest CUSUM statistic value is used, which is not easily affected by outliers.

[0041] 5. Compared with the traditional change point detection method, the ID method only detects a single change point in each isolated interval, with higher calculation efficiency.

[0042] 6. Considering that the M-test statistic is more easily affected by outliers and the T-test statistic is not robust, the common change points detected by the two test statistics are selected to improve the accuracy of change point detection.

[0043] 9. The standardized time series data is segmented based on the final change point position, and the mean value of the electricity consumption data of each industry in each segment of the time series data is calculated respectively, which can accurately and efficiently extract the typical electricity consumption behavior of each segment. Description of the Drawings

[0044] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments:

[0045] Figure 1 It is the flowchart of the method for analyzing the typical electricity consumption pattern of enterprises based on multi-dimensional Isolate-Detect multi-change point detection according to the embodiment of the present invention;

[0046] Figure 2 It is the original data diagram after filling in the missing values according to the embodiment of the present invention;

[0047] Figure 3 It is the time series diagram of the electricity consumption data on Monday according to the embodiment of the present invention;

[0048] Figure 4 It is the time series diagram of the electricity consumption data on Saturday according to the embodiment of the present invention;

[0049] Figure 5 It is the time series diagram of the electricity consumption data of each industry on Monday after data cleaning according to the embodiment of the present invention;

[0050] Figure 6 It is the time series diagram of the electricity consumption data of each industry on Saturday after data cleaning according to the embodiment of the present invention;

[0051] Figure 7Smoothing graph of electricity consumption data of each industry on Monday using the Savitzky-Golay filter in the embodiments of the present invention;

[0052] Figure 8 Smoothing graph of electricity consumption data of each industry on Monday using the Savitzky-Golay filter in the embodiments of the present invention;

[0053] Figure 9 Residual sequence graph of electricity consumption data of each industry on Monday after using the additive model in the embodiments of the present invention;

[0054] Figure 10 Residual sequence graph of electricity consumption data of each industry on Saturday after using the additive model in the embodiments of the present invention;

[0055] Figure 11 Time series graph of M-test statistic and T-test statistic of electricity consumption data on Monday in the embodiments of the present invention;

[0056] Figure 12 Time series graph of M-test statistic and T-test statistic of electricity consumption data on Saturday in the embodiments of the present invention;

[0057] Figure 13 Final change point distribution graph of electricity consumption on Monday detected by using the multi-dimensional Isolate-Detect algorithm in the embodiments of the present invention;

[0058] Figure 14 Final change point distribution graph of electricity consumption on Saturday detected by using the multi-dimensional Isolate-Detect algorithm in the embodiments of the present invention;

[0059] Figure 15 Piecewise mean curve graph of electricity consumption data on Monday in the embodiments of the present invention;

[0060] Figure 16 Piecewise mean curve graph of electricity consumption data on Saturday in the embodiments of the present invention. Detailed implementation manners

[0061] To make the features and advantages of this patent more obvious and understandable, specific embodiments are given below and described in detail in conjunction with the accompanying drawings as follows:

[0062] In this embodiment, the technical solutions in the embodiments of the present application are clearly and completely described using the electricity consumption data set of each industry in Fuzhou City, Fujian Province. As Figure 1 shown in the detailed process of the present invention, specific application examples for implementing this solution are provided below:

[0063] Step 1: Collect the electricity consumption time series data of each industry in the industry, where the data includes Gregorian date, industry type, and electricity consumption.

[0064] Step 2: Fill missing values ​​in the electricity consumption time series data. Mark the data with empty or zero electricity consumption as missing values. The data of electricity consumption of electronic machinery and wood processing on July 6, 2020 are missing. The electricity consumption data of electronic machinery and wood processing on every Monday from June to September 2020 are averaged and used to interpolate the missing values. The electricity consumption time series diagram is obtained, as shown in Figure 2 As shown. The data of Monday and Saturday are extracted from the power consumption time series data, as shown Figure 3 , Figure 4 As shown. Then the 3σ principle is used to remove outliers, that is, in each dimension of data, data points with a deviation from the mean value exceeding 3 times the standard deviation are identified as abnormal data, and the detected abnormal data are removed. Due to the large differences in electricity consumption between industries, in order to avoid the impact of large differences in the order of magnitude of the time series data of electricity consumption of each industry on the change point detection, the electricity consumption of each industry is standardized separately, and the formula is as follows:

[0065]

[0066] Among them, x i,t represents the electricity consumption of the ith industry at the tth time point, μ i represents the average electricity consumption of the ith industry, σ i represents the standard deviation of electricity consumption of the ith industry, X i,t Represents the standardized electricity consumption. After data cleaning, the electricity consumption time series of each industry on Monday and Saturday are as follows Figure 5 , Figure 6 As shown;

[0067] Step 3: First use the Savitzky-Golay filter to smooth the cleaned data. The Savitzky-Golay filter is a filtering method based on the local polynomial least squares fitting in the time domain. Assume that the original time series data is X i ={X i,1 ,X i,2 ,...X i,t ...,X i,T}, with X i,t As the origin, take X i,t There are m sample points on each side, and a window array containing 2m+1 sample points is constructed. Then a p-order polynomial is constructed to fit the data in the window. The p-order polynomial is as follows:

[0068]

[0069] Where -m≤n≤m, p≤2m+1; the loss function is defined as follows:

[0070]

[0071] When the loss function reaches the minimum value, the fitting effect of the original data reaches the optimal. By using a sliding window to obtain the fitted values of the smoothed original data, the noise of the data can be effectively reduced. The trend of the smoothed electricity consumption data on Monday is as Figure 7 shown, and the trend of the smoothed electricity consumption data on Saturday is as Figure 8 shown;

[0072] Next, the additive model is used to decompose the trend of the smoothed data, and the model is expressed as follows:

[0073]

[0074] where, T i represents the long-term time trend of the i-th industry, S i represents the seasonal time trend of the i-th industry, C i represents the cyclical time trend of the i-th industry, and I i represents the remaining residual term of the i-th industry. In the present invention, the smoothed electricity consumption data is decomposed into a trend part, a cycle part, and a residual term part, and the residual sequences of the electricity consumption data on Monday and the residual sequence of the electricity consumption data on Saturday are respectively obtained, as Figure 9 , Figure 10 shown.

[0075] Step 4: Calculate the CUSUM statistic of the electricity consumption residual term of each industry at each time point. The main formula is as follows:

[0076]

[0077] In the formula, i refers to the i-th industry, s refers to the starting point of the detection interval, e refers to the end point of the detection interval, b refers to the detection time point, n refers to the total length of the detection interval, and I i,t refers to the residual term of the smoothed data of the i-th industry at time t, refers to the CUSUM statistic value of the i-th industry at the b time point within the detection interval [s, e].

[0078] Then, calculate the mean and maximum value of the CUSUM statistic of the residual term of each industry at each time point, that is, the M test statistic and the T test statistic. The formulas are as follows:

[0079]

[0080]

[0081] In the formula, p refers to the total number of data dimensions. In this embodiment, p = 9, refers to the mean of the CUSUM statistic at the b time point within the detection interval [s, e]. Refers to the maximum value of the CUSUM statistic at time point b within the detection interval [s, e]. The time series diagrams of the M-test statistic and the T-test statistic for the electricity consumption data on Monday are as Figure 11 shown, and the time series diagrams of the M-test statistic and the T-test statistic for the electricity consumption data on Saturday are as Figure 12 shown.

[0082] Step 5: Based on the M-test statistic and the T-test statistic, use the ID method to detect change points and output the common change point positions of the multi-dimensional electricity consumption data. The ID method mainly screens change points through a threshold method, and its principle is as follows:

[0083] First, create the intervals to be detected. For a data sequence of length T, first set a positive constant λ T , and then create two sets of ordered left and right extended intervals with K = [T / λ T . The jth right extended interval is R j = [1, min{jλ T , T}], and the jth left extended interval is L j = [max{1, T - jλ T + 1}, T]. Collect these intervals in the ordered set S RL = {R1, L1, R2, L2,..., R K , L K}. Then ID identifies the point with the largest test statistic value in R1. Based on the obtained test statistic, set the threshold, and the calculation formula of the threshold is as follows:

[0084]

[0085] where ζ T is the obtained threshold, σ is the standard deviation of the input data, C is the given parameter value, T is the length of the input data sequence. In this embodiment, select T Saturday = 111.

[0086] If the value of the test statistic exceeds the threshold, then this point is regarded as a change point. If it does not exceed the threshold, continue to detect the next interval of S RL . After detection, the ID algorithm starts a new round of detection from the end point (starting point) of the right (or left) extended interval where the detection occurs.

[0087] Step 6: Using the two test statistics, obtain two change point sets CP_MM and CP_MT for the electricity consumption data on Monday. Take out the two elements with the absolute value of the difference less than 3 in set CP_MM and set CP_MT, and select the smaller value and put it into set CP_M to obtain the final change point set. The final change point detection result diagram for the electricity consumption data on Monday is as Figure 13As shown, a total of 11 change points are detected. The same detection method is applied to the electricity consumption data on Saturday, and the final set of change points CP_S for the Saturday electricity consumption data is obtained. A total of 10 change points are detected. The graph of the final change point detection result is as shown in Figure 14 shown.

[0088] Step 7: Based on the detected change point positions, segment the standardized time series data, calculate the mean value of the electricity consumption of each industry in each segment, and extract the typical electricity consumption scenarios of each segment of time series data. The calculation formula for the mean value of the electricity consumption of the i-th industry is as follows:

[0089]

[0090] In the formula, τ k refers to the position of the k-th change point, refers to the standardized electricity consumption value of the i-th industry at the position of the (k + 1)-th change point. The segmented mean curve of the Monday electricity consumption data is as shown in Figure 15 shown, and the segmented mean curve of the Saturday electricity consumption data is as shown in Figure 16 shown. At the same time, the mean values of each industry in the time series sub-scenarios are marked in the figure.

[0091] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more flows or multiple flows and / or blocks Figure 1 one or more blocks or multiple blocks.

[0093] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions in the processFigure 1 one process or multiple processes and / or boxes Figure 1 the functions specified in one box or multiple boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one process Figure 1 one process or multiple processes and / or boxes Figure 1 in one box or multiple boxes.

[0095] As described above, it is only the preferred embodiment of the present invention, and it is not a limitation to the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still belong to the protection scope of the technical solution of the present invention.

[0096] This patent is not limited to the above best mode. Anyone inspired by this patent can obtain various other forms of the analysis method for typical enterprise electricity consumption patterns based on multi-dimensional Isolate-Detect change-point detection. All equal changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope covered by this patent.

Claims

1. A method for analyzing typical enterprise electricity consumption patterns based on multi-dimensional Isolate-Detect change point detection, characterized in that The following steps are involved: Step 1: Collect the time series data of electricity consumption of each industry; Step 2: Fill missing values ​​in the electricity consumption time series data of each industry, extract electricity consumption data on a certain day on a weekday or a certain day on a weekend, use the 3σ principle to eliminate outliers, and then perform standardization; Step 3: Smooth the cleaned data first, and then perform seasonal decomposition on the smoothed data to eliminate the periodicity and trend of the data to obtain the residual term of the smoothed data; Step 4: Calculate the CUSUM statistic of the residual item of each industry electricity consumption data at each time point, and then calculate the mean and maximum value of the CUSUM statistic at each time point, which are recorded as M test statistics and T test statistics respectively; Step 5: Use the Isolate-Detect algorithm to detect the common change points of the electricity consumption data of each industry. Preset the threshold according to the threshold formula of the Isolate-Detect algorithm, determine the time point when the M test statistic exceeds the threshold as the change point, put the obtained change point into the change point set, recorded as the change point set CP_M; determine the time point when the T test statistic exceeds the threshold as the change point, put the obtained change point into the change point set, recorded as the change point set CP_T; Step 6: Record the change point set obtained by data analysis as the change point set CP, take out the two elements whose absolute difference value is less than 3 from the set CP_M and the set CP_T, select the smaller value and put it into the set CP to obtain the final change point set; Step 7: Segment the standardized electricity consumption time series data based on the obtained change point positions, calculate the average of the electricity consumption of each industry in each time series data segment, and obtain the typical electricity consumption behavior of each time series data segment; In step 4, the data residual term obtained after decomposition is used to calculate the mean of the CUSUM statistic at each time point. The formula is as follows: Among them, i refers to the i-th industry, s refers to the starting point of the detection interval, e refers to the ending point of the detection interval, b refers to the detection time point, n refers to the total length of the detection interval, and I i,t refers to the data residual term after smoothing at time t for the i-th industry, refers to the CUSUM statistic value at the b time point for the i-th industry within the detection interval [s, e], and p refers to the total number of data dimensions, refers to the mean of the CUSUM statistic at the b time point within the detection interval [s, e], denoted as In step 5, based on the calculated M test statistic and T test statistic, the Isolate-Detect algorithm is used to detect the common change points of electricity consumption in various industries: First, create the intervals to be detected. For a data sequence of length T, first set a positive constant λ T , and then create two sets of ordered left and right extended intervals with K = [T / λ T ; the j-th right extended interval is R j = [1, min{jλ T , T}], and the j-th left extended interval is L j = [max{1, T - jλ T + 1}, T]; collect these intervals in the ordered set S RL = {R1, L1, R2, L2,..., R K , L K}; then identify the point with the largest test statistic value in R1; Based on the obtained test statistic, a threshold is set, where the threshold is calculated as follows: where σ is the standard deviation of the input data, and C is a given parameter value; compare the threshold and the test statistic to determine whether this time point is a change point: if the value of the test statistic exceeds the threshold, then the corresponding point is regarded as a change point; if it does not exceed the threshold, then continue to detect the next interval of S RL of 2. The method for analyzing typical enterprise power consumption patterns based on multi-dimensional Isolate-Detect change point detection according to claim 1, wherein: In step 2, after filling the missing values ​​in the electricity consumption data of each industry, the electricity consumption data of a certain day on a weekday and the electricity consumption data of a certain day on a weekend are extracted for subsequent data analysis to analyze the differences in industrial electricity consumption behaviors on weekdays and weekends.

3. The method for analyzing typical enterprise power consumption patterns based on multi-dimensional Isolate-Detect change point detection according to claim 2, characterized in that: In step 3, the cleaned data is smoothed using a Savitzky-Golay filter. When the loss function reaches the minimum value, the fitting effect of the original data reaches the optimal value. The fitting value of the original data after smoothing is obtained through a sliding window.

4. The method for analyzing typical patterns of enterprise electricity consumption based on multi-dimensional Isolate-Detect and multi-change point detection according to claim 3 is characterized by: In step 3, for the smoothed data, the additive model is used to perform trend decomposition, and the smoothed power consumption data is decomposed into a trend part, a periodic part, and a residual part to eliminate the periodicity and trend of the data, which is expressed as follows: X i,t = T i,t + S i,t + C i,t + I i,t , (1) where, T i represents the long-term time trend of the i-th industry, S i represents the seasonal time trend of the i-th industry, C i represents the cyclical time trend of the i-th industry, I i represents the remaining residual term of the i-th industry.

5. The enterprise electricity consumption typical pattern analysis method based on multi-dimensional Isolate-Detect change point detection according to claim 1, characterized in that: In step 4, using the data residual term obtained after decomposition, calculate the maximum value of the CUSUM statistic at each time point, and the formula is as follows: Among them, s refers to the starting point of the detection interval, e refers to the ending point of the detection interval, and b refers to the detection time point. refers to the CUSUM statistic value of the i-th industry at the b time point within the detection interval [s, e], and p refers to the total number of data dimensions. refers to the maximum value of the CUSUM statistic at the b time point within the detection interval [s, e].

6. The enterprise electricity consumption typical pattern analysis method based on multi-dimensional Isolate-Detect change point detection according to claim 1, characterized in that: In step 7, based on the detected change point positions, segment the standardized time series data, calculate the mean value of the electricity consumption of each industry in each segment, and extract the typical electricity consumption scenarios of each segment of time series data. The formula for calculating the mean value of the electricity consumption of the i-th industry is as follows: where τ k denotes the position of the k-th change point, and denotes the normalized electricity consumption value of the i-th industry at the position of the (k + 1)-th change point.

Images