Electric power data intelligent monitoring and predicting method and system based on financial data intelligence

Abstract

The invention relates to the field of predictive analysis, in particular to an intelligent power data monitoring and predicting method and system based on financial digital intelligence, and the method comprises the steps: obtaining a plurality of load data segments, and decomposing the load data segments to obtain cycle subitems and frequency subitems; calculating the rule availability according to the cycle analysis and the frequency sub-items, wherein the rule availability is positively correlated with the rule accessibility and the rule reasonability; the rule accessibility is in positive correlation with both the period accessibility and the trend accessibility, the rule reasoning is in positive correlation with both the period reasoning and the trend reasoning, and the weighting coefficient of the prediction algorithm is adjusted by using the rule availability so as to realize load data prediction. Therefore, accurate power load prediction is realized in a self-adaptive weighting coefficient mode.

Description

Technical Field

[0001] The present invention generally relates to the field of forecasting and analysis, and more specifically, to a method and system for intelligent monitoring and forecasting of power data based on financial digitization. Background Art

[0002] Power load forecasting plays an important role in power system operation and energy management. For example, by forecasting power load, it can help power companies adjust production plans in a timely manner to prevent overcapacity or undercapacity. The accuracy of power load forecasting will determine the accuracy of power production plans, so how to accurately forecast power load has become an urgent problem to be solved.

[0003] Exponential smoothing is a commonly used prediction algorithm. The weighting coefficient in the algorithm is used to determine the reference amount of the historical data change law during the prediction analysis. The setting quality of the weighting coefficient will directly affect the estimation accuracy. The existing patent application document CN112700288A shows a commodity price prediction method using the exponential smoothing method. The document uses the exponential smoothing method to predict commodity prices. The exponential smoothing method in the document adopts the traditional way of setting the weighting coefficient, that is, setting a fixed weighting coefficient, and using the weighting coefficient to complete the commodity price prediction at all times. Due to factors such as weather, the time series of power load data will vary greatly. If a fixed weighting coefficient is used, the reference amount of the historical data change law will be constant during the power prediction at each moment. This way of setting the weighting coefficient is not suitable for power load prediction scenarios with large change differences. For example, the future power load data is similar to the current power load data. Because the setting of inappropriate weighting coefficients leads to excessive reference to the historical change law, the reference amount to the current power load data is ignored, which leads to the inability to accurately predict the power load data at future moments.

[0004] Therefore, how to set a suitable weighting coefficient becomes the research focus of the present invention. Summary of the invention

[0005] In order to solve the problem of how to set appropriate weighting coefficients, the present invention provides an intelligent monitoring and prediction method and system for power data based on financial digitization.

[0006] In the first aspect, the present invention provides an intelligent monitoring and prediction method for power data based on financial digitalization, which adopts the following technical solutions: The intelligent monitoring and prediction method of power data based on financial digitalization includes the following steps: Obtaining load data time series; The load data time series is segmented to obtain a number of load data segments, and the load data segments are decomposed to obtain period sub-items and frequency sub-items; Calculate the law availability, which is positively correlated with both the law persistence and the law reasonability; The regularity persistence is positively correlated with both the period persistence and the trend persistence. The period persistence indicates the similarity of the frequency domain information of the period sub-items of two adjacent load data segments. The trend persistence indicates the similarity of the fitting relationship of the trend sub-items of two adjacent load data segments. The regular reasoning is positively correlated with both the periodic reasoning and the trend reasoning. The periodic reasoning indicates the autocorrelation of the frequency domain information variation of the periodic sub-items of all two adjacent load data segments. The trend reasoning indicates the autocorrelation of the fitting relationship variation of the trend sub-items of all two adjacent load data segments. The regular availability is used to adjust the weight coefficient of the prediction algorithm, and the adjusted weight coefficient is negatively correlated with the regular availability to achieve load data prediction.

[0007] The present invention uses the law availability to accurately describe the content of historical laws retained in the data law of the current moment, thereby providing a basis for accurately setting the weighting coefficient; further, when calculating the law availability, not only the directly retained historical law information is considered, but also the historical law information that can be analyzed by reasoning is considered, so the present invention more comprehensively describes the content of historical laws retained in the data law of the current moment; further, when analyzing the directly retained historical law information, the periodic information and the trend information are studied separately, thereby more accurately describing the directly retained historical law information; further, when analyzing the historical law information analyzed by reasoning, the periodic information and the trend information are studied separately, thereby more accurately describing the historical law information that can be analyzed by reasoning; further, when calculating the cycle continuity, the frequency domain information is introduced to accurately describe the similarity of the periodic information of each load data segment, and then by analyzing the two adjacent load data The similarity of frequency domain information in the segments can accurately reflect the situation of retaining the periodic information in the previous load data segment in each load data segment, thereby providing a basis for accurately analyzing the content of historical laws retained in the data law at the current moment; further, when calculating the trend continuity, the present invention accurately describes the situation of retaining the trend information in the previous load data segment in each load data segment by analyzing the similarity of the fitting relationship in the trend sub-items in the two previous and next load data segments, thereby providing a basis for accurately analyzing the content of historical laws retained in the data law at the current moment; further, when calculating the periodic reasoning, the present invention accurately analyzes the regularity of the periodic information variation by analyzing the autocorrelation of the periodic information variation of all two adjacent load data segments, thereby accurately reflecting how much regular information in the data law at the current moment can be obtained by reasoning and analyzing the historical regular information, thereby providing a basis for accurately analyzing the content of historical laws retained in the data law at the current moment. Furthermore, when calculating trend reasoning, the present invention accurately analyzes the regularity of trend information changes by analyzing the autocorrelation of trend information changes between all two adjacent load data segments, thereby accurately reflecting how much regularity information in the data regularity at the current moment can be obtained by reasoning and analyzing historical regularity information, thereby providing a basis for accurately analyzing the content of historical regularities retained in the data regularity at the current moment. Preferably, the method for obtaining the periodic continuity includes: Obtaining frequency spectrum data of periodic sub-items of the load data segment, recording the vectors composed of each frequency and the corresponding amplitude in the frequency spectrum data as frequency domain description vectors, and arranging all frequency domain description vectors according to the frequency to obtain a frequency domain description vector sequence; Calculating the local periodicity of each load data segment, wherein the local periodicity is positively correlated with the similarity of the frequency domain description vector sequences of two adjacent load data segments; The average of the local periodic persistence of all load data segments is taken as the periodic persistence.

[0008] Preferably, the local periodic continuity of each load data segment satisfies the relationship: Preferably, the local periodic continuity of each load data segment satisfies the relationship: ; in, The first frequency domain description vector sequence representing a load data segment The frequency domain describes the frequencies in the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The frequency domain describes the frequencies in the vector, The first frequency domain description vector sequence of the load data segment The amplitude in the frequency domain describes the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The amplitude in the frequency domain describes the vector, Indicates the number of frequency domain description vectors contained in the frequency domain description vector sequence, Indicates the local periodic continuity of the load data segment.

[0009] When calculating the periodic continuity, the present invention introduces frequency domain information to accurately describe the similarity of the periodic information of each load data segment, and then by analyzing the similarity of the frequency domain information in two adjacent load data segments, it can accurately reflect the situation of each load data segment retaining the periodic information in the previous load data segment, thereby providing a basis for accurately analyzing the content of historical laws retained in the data laws at the current moment.

[0010] Preferably, the method for obtaining the trend continuity includes: Using the least square method to process the trend sub-item fitting polynomial of each load data segment to obtain a first fitting formula, obtaining constant coefficients in the first fitting formula, and forming a coefficient sequence with all constant coefficients of the first fitting formula; The cosine similarity of the coefficient sequence of a load data segment and the previous load data segment is calculated and recorded as the local trend persistence of the load data segment, and the average of the local trend persistence of all load data segments is recorded as the trend persistence.

[0011] When calculating the continuity of the trend, the present invention accurately describes the retention of the trend information in the previous load data segment in each load data segment by analyzing the similarity of the fitting relationship in the trend sub-items in the two previous and subsequent load data segments, thereby providing a basis for accurately analyzing the content of historical laws retained in the data laws at the current moment.

[0012] Preferably, the periodic reasoning acquisition method comprises: Calculating the periodic variation of each load data segment, wherein the periodic variation is positively correlated with the difference between the frequency description vector sequences of two adjacent load data segments; The autocorrelation value obtained by performing autocorrelation detection on the periodic variation of all load data segments is used as the periodic inference.

[0013] When calculating the periodic reasoning, the present invention accurately analyzes the regularity of the periodic information variation by analyzing the autocorrelation of the periodic information variation of all two adjacent load data segments, thereby accurately reflecting how much regular information in the data regularity at the current moment can be obtained by reasoning and analyzing the historical regular information, thereby providing a basis for accurately analyzing the content of historical regularities retained in the data regularity at the current moment.

[0014] Preferably, the periodic variation satisfies the relationship: ; in, The first frequency domain description vector sequence representing a load data segment The frequency domain describes the frequencies in the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The frequency domain describes the frequencies in the vector, The first frequency domain description vector sequence of the load data segment The amplitude in the frequency domain describes the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The amplitude in the frequency domain describes the vector, Indicates the number of frequency domain description vectors contained in the frequency domain description vector sequence, Indicates the periodic variation of the load data segment.

[0015] Preferably, the trend inference acquisition method includes: The difference between 1 and the first cosine similarity is used as the trend change amount of the load data segment, wherein the first cosine similarity is the cosine similarity between the coefficient sequence of a load data segment and the coefficient sequence of the previous load data segment; The autocorrelation value obtained by performing autocorrelation detection on the trend variation of all load data segments is used as trend inference.

[0016] When calculating trend reasoning, the present invention accurately analyzes the regularity of trend information variation by analyzing the autocorrelation of trend information variation of all two adjacent load data segments, thereby accurately reflecting how much regularity information in the data regularity at the current moment can be obtained by reasoning and analyzing historical regularity information, thereby providing a basis for accurately analyzing the content of historical regularities retained in the data regularity at the current moment.

[0017] Preferably, the method for acquiring the load data segment includes: The load data time series sequence is segmented to obtain unit segments; the first unit segment is taken as the first starting segment, and several temporary data segments of the first starting segment are obtained, and the temporary data segments are obtained by splicing the unit segments; the prediction error change of the temporary data segments of the first starting segment is calculated; the temporary data segment in which the prediction error change is greater than the change reference value for the first time is taken as the first load data segment, and the change reference value is positively correlated with the prediction error change of all previous temporary data segments; in response to the existence of a unit segment after the first load data segment, the next unit segment of the first load data segment is taken as the second starting segment, and the second load data segment is obtained according to the second starting segment; in response to the absence of a unit segment after the second load data segment, the iteration ends; and all load data segments are obtained.

[0018] When acquiring load data segments, the present invention controls the division of data segments according to the conformity of the fitting law of the latter data to the former data, thereby realizing the segmentation of data with similar laws into one data segment, and providing a basis for the subsequent analysis of the content of historical laws retained in the data laws at the current moment.

[0019] Preferably, calculating the prediction error variation of the temporary data segment includes: The fitting relationship obtained by fitting the temporary data segment is used as a prediction model, and the prediction model is used to perform prediction analysis to obtain the load data prediction values ​​at the next L moments, and the load data prediction values ​​at the next L moments constitute a prediction value data segment, and the Euclidean distance between the prediction value data segment and the next unit segment of the temporary data segment is divided by L to obtain the prediction error of the temporary data segment; The absolute value of the difference between the prediction error of the temporary data segment and the prediction error of the previous temporary data segment is taken as the prediction error change.

[0020] In the second aspect, the present invention provides an intelligent monitoring and prediction system for power data based on financial digitalization, which adopts the following technical solutions: The intelligent monitoring and prediction system for electric power data based on financial digitization includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned intelligent monitoring and prediction method for electric power data based on financial digitization is implemented.

[0021] By adopting the above technical solution, the above-mentioned intelligent monitoring and prediction method of power data based on financial digitization is generated into a computer program and stored in a memory to be loaded and executed by a processor, so that a terminal device is made according to the memory and the processor for easy use.

[0022] The present invention has the following beneficial effects: the present invention accurately describes the content of historical rules retained in the data rules at the current moment through the rule availability, thereby providing a basis for accurately setting the weighting coefficient; further, when calculating the rule availability, not only the directly retained historical rule information is considered, but also the historical rule information that can be analyzed by reasoning is considered, so the present invention more comprehensively describes the content of historical rules retained in the data rules at the current moment; Further, when analyzing the directly retained historical regularity information, the periodic information and the trend information are studied separately, so as to more accurately describe the directly retained historical regularity information; further, when analyzing the historical regularity information derived by inference, the periodic information and the trend information are studied separately, so as to more accurately describe the historical regularity information that can be derived by inference; Furthermore, when calculating the cycle continuity, the frequency domain information is introduced to accurately describe the similarity of the cycle information of each load data segment, and then by analyzing the similarity of the frequency domain information in two adjacent load data segments, it can accurately reflect the situation of each load data segment retaining the cycle information in the previous load data segment, thereby providing a basis for accurately analyzing the content of historical laws retained in the data law at the current moment; Furthermore, when calculating the continuity of the trend, the present invention accurately describes the retention of the trend information in the previous load data segment in each load data segment by analyzing the similarity of the fitting relationship in the trend sub-items in the two previous load data segments, thereby providing a basis for accurately analyzing the content of the historical law retained in the data law at the current moment; Furthermore, when calculating the periodic reasoning, the present invention accurately analyzes the regularity of the periodic information variation by analyzing the autocorrelation of the periodic information variation of all two adjacent load data segments, thereby accurately reflecting how much regular information in the data regularity at the current moment can be obtained by reasoning and analyzing the historical regular information, thereby providing a basis for accurately analyzing the content of historical regularities retained in the data regularity at the current moment.

[0023] Furthermore, when calculating trend reasoning, the present invention accurately analyzes the regularity of trend information changes by analyzing the autocorrelation of trend information changes between all two adjacent load data segments, thereby accurately reflecting how much regularity information in the data regularity at the current moment can be obtained by reasoning and analyzing historical regularity information, thereby providing a basis for accurately analyzing the content of historical regularities retained in the data regularity at the current moment.

[0024] Furthermore, when acquiring load data segments, the present invention controls the division of data segments according to the conformity of the later data with the fitting law of the previous data, thereby realizing the segmentation of data with similar laws into one data segment, providing a basis for the subsequent analysis of the content of historical laws retained in the data laws at the current moment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a flowchart of the steps of the method for intelligent monitoring and prediction of power data based on financial digitization in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0027] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0028] The embodiment of the present invention discloses a method for counting goods in and out of a warehouse based on target tracking, referring to Figure 1 , comprising steps S1 to S4: S1. Obtain the load data time series.

[0029] Specifically, the load data of the power company is collected once every n hours, collected N times, and the load data collected at all times are arranged in time sequence to obtain a load data time series sequence. n represents a preset collection interval, and N represents a preset number of collections. This embodiment is described by taking n as 1 and N as 2000 as an example. Other embodiments may take other values, and this embodiment does not specifically limit it.

[0030] S2. Segment the load data time series to obtain several load data segments, and decompose the load data segments to obtain period sub-items and frequency sub-items.

[0031] S20: Segment the load data time series to obtain a number of load data segments.

[0032] It should be noted that in order to set a suitable weighting coefficient, it is necessary to analyze the changes in historical rules. Due to the influence of factors such as temperature, the power load changes unstably in time series, so the power load has different change rules in different stages. In order to better analyze the changes in the historical rules of power load, it is necessary to segment data with similar change rules together.

[0033] Optionally, as an embodiment, the load data time series sequence is segmented to obtain a plurality of load data segments, including: The load data time series is evenly divided into data segments of length M, which are recorded as load data segments. M represents a preset segment length. This embodiment takes M as 100 as an example for description. Other embodiments may take other values, which are not specifically limited in this embodiment.

[0034] It should be noted that, since the predetermined segment length is artificially given, it is not set according to the data variation law. Therefore, the data segments obtained by this segmentation method may divide data with similar variation laws into different segments, and may also divide data with different variation laws into one data segment. Therefore, this segmentation method will affect the accuracy of the subsequent setting of the weighting coefficient.

[0035] Preferably, as an example, the load data time series sequence is segmented to obtain a plurality of load data segments, including: The load data time series sequence is evenly divided into several data segments of length L, which are recorded as unit segments; the first unit segment is taken as the first starting segment, several temporary data segments of the first starting segment are obtained, the prediction error change of the temporary data segments of the first starting segment is calculated, and the temporary data segment whose prediction error change is greater than the change reference value appears for the first time in all temporary data segments of the first starting segment is taken as the first load data segment; in response to the existence of a unit segment after the first load data segment, the next unit segment of the first load data segment is taken as the second starting segment, several temporary data segments of the second starting segment are obtained, and the prediction error change of the temporary data segment of the second starting segment is calculated, and the temporary data segment whose prediction error change is greater than the change reference value appears for the first time in all temporary data segments of the second starting data segment is taken as the second load data segment; in response to the absence of a unit segment after the second load data segment, the iteration ends to obtain all load data segments. L represents the preset unit segment length. This embodiment is described by taking L as 10 as an example. Other embodiments may take other values, and this embodiment does not make specific restrictions.

[0036] Preferably, as an example, the method for calculating the prediction error change of the temporary data segment is: The temporary data segment is fitted with a polynomial using the least squares method to obtain a fitting relationship, which is recorded as a prediction model. The prediction model is used to fit the load data for the next L moments, which is recorded as the load data prediction value. The load data prediction values ​​for the next L moments constitute a prediction value data segment. The Euclidean distance between the prediction value data segment and the next unit segment of the temporary data segment is divided by L to obtain the prediction error of the temporary data segment.

[0037] The absolute value of the difference between the prediction error of the temporary data segment and the prediction error of the previous temporary data segment is taken as the prediction error change.

[0038] Among them, the above embodiments involve a variation reference value, and the following needs to explain a method for determining the variation reference value, including: Twice the average of the prediction error changes of all temporary data segments before the temporary data segment is used as the reference value of the change.

[0039] Among them, the above embodiments involve temporary data segments, and the method for determining the temporary data segments needs to be described below, including: The starting data segment is concatenated with one unit segment, two unit segments, ..., S unit segments thereafter to obtain a number of temporary data segments, where S represents the number of unit segments following the starting data segment.

[0040] For example, the starting data segment is concatenated with the following unit segment to obtain a temporary data segment, and the starting data segment is concatenated with the following S data segments to obtain a temporary data segment. It should be noted that the amount of data in the unit segment is small, so there may be a change pattern in multiple data segments. Therefore, in order to segment the data belonging to the same change pattern into one data segment, the unit segment needs to be merged. Since the fitting relationship of the data is similar when there is the same change pattern, when the fitting relationship of the previous data can be used to better fit the subsequent data, it means that the subsequent data can be merged with the previous data. Therefore, the data segmentation based on this theory can better segment data with similar data patterns together, and segment data with large differences in data patterns into different data segments. Provide a basis for the subsequent law analysis.

[0041] S21: Decompose the load data segment into period sub-items and frequency sub-items.

[0042] Optionally, as an example, decomposing the load data segment into periodic sub-items and frequency sub-items includes: utilizing an ARIMA algorithm to decompose each load data segment into periodic sub-items and trend sub-items of each load data segment.

[0043] It should be noted that the process of decomposing each load data segment using the ARIMA algorithm to obtain the period sub-items and trend sub-items of each load data segment is a prior art and will not be described in detail here.

[0044] S3: Calculate the availability of the rule, the rule availability is positively correlated with both the rule persistence and the rule reasoning; the rule persistence is positively correlated with both the period persistence and the trend persistence, the period persistence indicates the similarity of the frequency domain information of the periodic sub-items of two adjacent load data segments, the trend persistence indicates the similarity of the fitting relationship of the trend sub-items of two adjacent load data segments; the rule reasoning is positively correlated with both the periodic reasoning and the trend reasoning, the periodic reasoning indicates the autocorrelation of the frequency domain information change of the periodic sub-items of all two adjacent load data segments, the trend reasoning indicates the autocorrelation of the fitting relationship change of the trend sub-items of all two adjacent load data segments.

[0045] It should be noted that the weighting coefficient in the exponential smoothing method determines the reference amount of historical rules and the reference amount of current data in the forecast analysis. In order to set a suitable weighting coefficient, it is necessary to determine how much historical change rules can be retained in the data change rule at the current moment. This embodiment uses the rule availability to reflect the retention of historical change rules in the data change rule at the current moment.

[0046] It should be further explained that the preservation of historical laws in the data change law at the current moment includes not only the historical law information directly preserved, but also the historical law information obtained through reasoning. Therefore, when analyzing the availability of laws, it is necessary to combine the two aspects of information to measure the preservation of historical laws.

[0047] Preferably, as an example, calculating the law availability includes: The normalized value of the cumulative sum of the law's persistence and the law's reasonability is taken as the law's availability.

[0048] It should be noted that this embodiment uses the sigmoid function to normalize the cumulative sum of regularity continuity and regularity reasoning, and other embodiments may use other functions for normalization.

[0049] Among them, the above embodiments involve the continuity of regularity, and the method for obtaining the continuity of regularity needs to be explained below.

[0050] It should be noted that the retention of the utilization rules can be reflected by the remaining historical rules, among which common data rules include periodic rules and trend rules, and thus the remaining historical rules can be reflected by analyzing the remaining periodic rules and the remaining trend rules. In this embodiment, the periodic continuity is used to reflect the remaining periodic rules, and the trend continuity is used to reflect the remaining trend rules.

[0051] S30: Calculate the continuity of the law.

[0052] Preferably, as an example, the method for obtaining the regularity continuity includes steps S301-S303: S301: Calculate cycle continuity.

[0053] Preferably, as an example, the method for obtaining periodic continuity includes: The spectrum data of the periodic sub-items of the load data segment are obtained, and the vectors composed of each frequency and the corresponding amplitude in the spectrum data are recorded as frequency domain description vectors. All frequency domain description vectors are arranged according to the frequency to obtain a frequency domain description vector sequence.

[0054] Calculate the local periodicity of each load data segment.

[0055] The average of the local periodic persistence of all load data segments is taken as the periodic persistence.

[0056] Preferably, as an example, the local periodic continuity of each load data segment satisfies the relationship: ; in, The first frequency domain description vector sequence representing a load data segment The frequency domain describes the frequencies in the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The frequency domain describes the frequencies in the vector, The first frequency domain description vector sequence of the load data segment The amplitude in the frequency domain describes the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The amplitude in the frequency domain describes the vector, Indicates the number of frequency domain description vectors contained in the frequency domain description vector sequence, Indicates the local periodic continuity of the load data segment.

[0057] It is understandable that the local periodicity of a load data segment can be reflected by the fact that the load data segment inherits the periodicity of the previous load data segment. When the periodicity similarity between the load data segment and the previous load data segment is high, it means that the load data segment inherits more periodicity of the previous load data segment. Cosine similarity can measure the similarity of two sequence information. The traditional cosine similarity calculation formula is This method of calculating cosine similarity only considers the frequency similarity between the load data segment and the previous load data segment. However, this analysis method does not consider the content similarity between the corresponding frequencies of the load data segment and the previous load data segment. The period information contains not only frequency information but also frequency content information. The amplitude can better reflect the content of each frequency. Therefore, in order to reflect the content similarity between the corresponding frequencies of the load data segment and the previous load data segment, this embodiment uses As The weight of Weighted. The larger the frequency, the and The higher the content similarity, the higher the content similarity. Through this adjustment method, it is possible to measure not only the frequency similarity between the load data segment and the previous data segment, but also the frequency content similarity between the load data segment and the previous data segment. This achieves a more accurate description of the similarity between the periodic regularity of the load data segment and the previous data segment.

[0058] S302: Calculate the trend continuity.

[0059] Preferably, as an example, the method for obtaining trend persistence includes: The trend sub-item fitting polynomial of each load data segment is processed by the least square method to obtain a first fitting formula, and the constant coefficient of each sub-item in the first fitting formula is obtained. For example, the first fitting formula is: . Among them, 4, 6, 3, 5, 9, and 7 are respectively The constant coefficients of the sub-items are formed into a coefficient sequence by combining all the constant coefficients of the first fitting formula.

[0060] The cosine similarity of the coefficient sequence of a load data segment and the previous load data segment is calculated and recorded as the local trend persistence of the load data segment, and the average of the local trend persistence of all load data segments is recorded as the trend persistence.

[0061] S303: Calculate the continuity of the rule based on the continuity of the cycle and the continuity of the trend.

[0062] Preferably, as an example, calculating the regularity persistence according to the cycle persistence and the trend persistence includes: Take the mean of all data in the cycle sub-item as the first mean, and the mean of all data in the trend sub-item as the second mean. Take the first mean as the weight of the cycle persistence, and the second mean as the weight of the trend persistence. Take the weighted sum of the cycle persistence and the trend persistence to obtain the regular persistence.

[0063] Among them, the above embodiments also involve regular reasoning, and the method for obtaining regular reasoning needs to be explained below.

[0064] It should be noted that in addition to the direct laws that have been preserved in the data change law, there are also laws that can be analyzed through reasoning. For example, the change of the trend sub-items of power load conforms to the law of doubling step by step, such as 1, 2, 4, and 8. This change law can be obtained through reasoning, so this law is also an applicable law in the historical change law.

[0065] S31: Calculation of regularity reasoning.

[0066] Preferably, as an example, the regular reasoning acquisition method includes steps S310-S312: S310: Calculate cycle reasoning.

[0067] Preferably, as an example, the periodic reasoning acquisition method includes: Calculate the periodic variation of each load data segment.

[0068] An autocorrelation function is used to perform autocorrelation detection on the periodic variation of all load data segments to obtain a first autocorrelation value, and the obtained first autocorrelation value is used as the periodic inference.

[0069] It should be noted that the periodic variation has a strong autocorrelation, which means that the regularity of the periodic variation is more obvious, and further indicates that the following regularity information can be inferred by analyzing the variation relationship of the periodic information.

[0070] Preferably, as an example, the periodic variation satisfies the relationship: ; in, The first frequency domain description vector sequence representing a load data segment The frequency domain describes the frequencies in the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The frequency domain describes the frequencies in the vector, The first frequency domain description vector sequence of the load data segment The amplitude in the frequency domain describes the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The amplitude in the frequency domain describes the vector, Indicates the number of frequency domain description vectors contained in the frequency domain description vector sequence, Indicates the periodic variation of the load data segment.

[0071] S311: Calculate trend reasoning.

[0072] Preferably, as an example, the trend inference acquisition method includes: The difference between 1 and the first cosine similarity is used as the trend change amount of the load data segment, and the first cosine similarity is the cosine similarity between the coefficient sequence of a load data segment and the coefficient sequence of the previous load data segment.

[0073] The autocorrelation function is used to perform autocorrelation detection on the trend variation of all load data segments to obtain a second autocorrelation value, and the second autocorrelation value is used as trend inference.

[0074] S312: Calculate regular reasoning based on periodic reasoning and trend reasoning.

[0075] Preferably, as an example, the method for obtaining regular reasoning includes: The first mean is used as the weight of periodic reasoning, and the second mean is used as the weight of trend reasoning. The periodic reasoning and trend reasoning are weighted and summed to obtain the regular reasoning.

[0076] S4. Use the regular availability to adjust the weighting coefficient of the prediction algorithm to achieve load data prediction.

[0077] S40: Using the rule availability to adjust the weighting coefficient of the prediction algorithm.

[0078] Optionally, as an example, the weight coefficient of the prediction algorithm is adjusted by using the law availability, including: The adjusted weighting coefficient satisfies the relationship:

[0079] in, Represents a preset weighting coefficient. In this embodiment, Take 0.5 as an example for description. Other embodiments may take other values, and this embodiment does not impose specific restrictions. D represents the availability of the rule. The greater the availability of the rule, the more information that can be referenced by the historical rule. Therefore, more references to historical rules are needed when making load forecasts. The weighting coefficient is inversely proportional to the reference amount of the historical rule. Therefore, when the availability of the historical rule is higher, the optimization weighting coefficient is smaller. express Function, used for normalization, Represents the adjusted weighting coefficient.

[0080] S41: Perform load forecasting based on the adjusted weighting coefficient.

[0081] Optionally, as an example, performing load forecasting according to the adjusted weighting coefficient includes: The adjusted weighting coefficient is used as the weighting coefficient, and based on the load data in the last load data segment, the load data forecast value at the future moment is predicted using the exponential smoothing method.

[0082] An embodiment of the present invention also discloses an intelligent monitoring and prediction system for electric power data based on financial digitization, including a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, an intelligent monitoring and prediction method for electric power data based on financial digitization according to the present invention is implemented.

[0083] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0084] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid storage cube, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0085] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0086] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. The intelligent monitoring and prediction method of power data based on financial digitalization is characterized by: include: Obtaining load data time series; The load data time series is segmented to obtain a number of load data segments, and the load data segments are decomposed to obtain period sub-items and frequency sub-items; Calculate the law availability, which is positively correlated with both the law persistence and the law reasonability; The regularity persistence is positively correlated with both the period persistence and the trend persistence. The period persistence indicates the similarity of the frequency domain information of the period sub-items of two adjacent load data segments. The trend persistence indicates the similarity of the fitting relationship of the trend sub-items of two adjacent load data segments. The regular reasoning is positively correlated with both the periodic reasoning and the trend reasoning. The periodic reasoning indicates the autocorrelation of the frequency domain information variation of the periodic sub-items of all two adjacent load data segments. The trend reasoning indicates the autocorrelation of the fitting relationship variation of the trend sub-items of all two adjacent load data segments. The regular availability is used to adjust the weight coefficient of the prediction algorithm, and the adjusted weight coefficient is negatively correlated with the regular availability to achieve load data prediction.

2. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 1 is characterized in that: The method for obtaining the periodic continuity includes: Obtaining frequency spectrum data of periodic sub-items of the load data segment, recording the vectors composed of each frequency and the corresponding amplitude in the frequency spectrum data as frequency domain description vectors, and arranging all frequency domain description vectors according to the frequency to obtain a frequency domain description vector sequence; Calculating the local periodicity of each load data segment, wherein the local periodicity is positively correlated with the similarity of the frequency domain description vector sequences of two adjacent load data segments; The average of the local periodic persistence of all load data segments is taken as the periodic persistence.

3. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 2 is characterized in that: The local periodicity of each load data segment satisfies the relationship: ； in, The first frequency domain description vector sequence of a load data segment The frequency domain describes the frequencies in the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The frequency domain describes the frequencies in the vector, The first frequency domain description vector sequence of the load data segment The amplitude in the frequency domain describes the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The amplitude in the frequency domain describes the vector, Indicates the number of frequency domain description vectors contained in the frequency domain description vector sequence, Indicates the local periodic continuity of the load data segment.

4. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 1 is characterized in that: The method for obtaining trend continuity includes: Using the least square method to process the trend sub-item fitting polynomial of each load data segment to obtain a first fitting formula, obtaining constant coefficients in the first fitting formula, and forming a coefficient sequence with all constant coefficients of the first fitting formula; The cosine similarity of the coefficient sequence of a load data segment and the previous load data segment is calculated and recorded as the local trend persistence of the load data segment, and the average of the local trend persistence of all load data segments is recorded as the trend persistence.

5. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 2 is characterized in that: The periodic reasoning acquisition method comprises: Calculating the periodic variation of each load data segment, wherein the periodic variation is positively correlated with the difference between the frequency description vector sequences of two adjacent load data segments; The autocorrelation value obtained by performing autocorrelation detection on the periodic variation of all load data segments is used as the periodic inference.

6. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 5 is characterized in that: The periodic variation satisfies the relationship: ； in, The first frequency domain description vector sequence of a load data segment The frequency domain describes the frequencies in the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The frequency domain describes the frequencies in the vector, The first frequency domain description vector sequence of the load data segment The amplitude in the frequency domain describes the vector, The first one in the frequency domain description vector sequence of the previous load data segment of this load data segment The amplitude in the frequency domain describes the vector, Indicates the number of frequency domain description vectors contained in the frequency domain description vector sequence, Indicates the periodic variation of the load data segment.

7. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 4 is characterized in that: The trend inference acquisition method includes: The difference between 1 and the first cosine similarity is used as the trend change amount of the load data segment, wherein the first cosine similarity is the cosine similarity between the coefficient sequence of a load data segment and the coefficient sequence of the previous load data segment; The autocorrelation value obtained by performing autocorrelation detection on the trend variation of all load data segments is used as trend inference.

8. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 1 is characterized in that: The method for acquiring the load data segment comprises: The load data time series sequence is segmented to obtain unit segments; the first unit segment is taken as the first starting segment, and several temporary data segments of the first starting segment are obtained, and the temporary data segments are obtained by splicing the unit segments; the prediction error change of the temporary data segments of the first starting segment is calculated; the temporary data segment in which the prediction error change is greater than the change reference value for the first time is taken as the first load data segment, and the change reference value is positively correlated with the prediction error change of all previous temporary data segments; in response to the existence of a unit segment after the first load data segment, the next unit segment of the first load data segment is taken as the second starting segment, and the second load data segment is obtained according to the second starting segment; in response to the absence of a unit segment after the second load data segment, the iteration ends; and all load data segments are obtained.

9. The method for intelligent monitoring and prediction of power data based on financial digitization according to claim 8 is characterized in that: Calculate the change in prediction error for the temporary data segment, including: The fitting relationship obtained by fitting the temporary data segment is used as a prediction model, and the prediction model is used to perform prediction analysis to obtain the load data prediction values ​​at the next L moments, and the load data prediction values ​​at the next L moments constitute a prediction value data segment, and the Euclidean distance between the prediction value data segment and the next unit segment of the temporary data segment is divided by L to obtain the prediction error of the temporary data segment; The absolute value of the difference between the prediction error of the temporary data segment and the prediction error of the previous temporary data segment is taken as the prediction error change.

10. The power data intelligent monitoring and prediction system based on financial digitalization is characterized by: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for intelligent monitoring and prediction of power data based on financial digitization according to any one of claims 1 to 9 is implemented.

Images