A method and system for power operation and maintenance management

Abstract

The present invention relates to the field of data processing technology, and specifically to a method and system for power operation and maintenance management, including: obtaining the power data trend intensity and power data seasonal intensity at each moment in the power data sequence according to the data distribution of all reference moments at each moment in the power trend data sequence, the power season data sequence, and the power residual data sequence; obtaining the weights of the trend item at each moment, the seasonal item at each moment, and the residual item at each moment according to the difference between the power data trend intensity at all moments in the local window at each moment in the power data sequence, and the difference between the power data seasonal intensity at all moments, and obtaining the final prediction value, so as to formulate a more scientific and reasonable power system planning and operation strategy. The present invention improves the accuracy of power data prediction and also improves the reliability of power system planning.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a power operation and maintenance management method and system. Background Art

[0002] Power operation and maintenance management refers to the effective management and maintenance of the power system to ensure its normal operation and reliability. It monitors and analyzes power data and predicts future data based on previously known data, so as to formulate more scientific and reasonable power system planning and operation strategies, which is a key activity to ensure the normal operation and reliability of the power system.

[0003] Since there are periodicities and trends in the use of electricity, the conventional STL algorithm can be used to decompose the power data to obtain three different data items: trend item, seasonal item and residual item. The corresponding prediction model is used to make predictions based on the data of different items. Based on the data after each prediction, the power data for each moment in the future is obtained to plan the power system. However, in the process of making predictions based on existing data, due to special circumstances such as special festivals, performances or competitions, the acquired power data will suddenly become higher, which makes the accuracy of the predicted data not high, and also causes the planning and operation of the power system to be seriously affected. Summary of the invention

[0004] The present invention provides a power operation and maintenance management method and system to solve the existing problems.

[0005] A power operation and maintenance management method and system of the present invention adopts the following technical solutions:

[0006] An embodiment of the present invention provides a power operation and maintenance management method, the method comprising the following steps:

[0007] Acquire power data at a number of consecutive moments, organize the power data at a number of consecutive moments into a power data sequence according to the time sequence, and decompose the power trend data sequence into a power trend data sequence, a power season data sequence, and a power residual data sequence;

[0008] The previous several moments of each moment are recorded as the reference moments of each moment, and the power data trend intensity and power data seasonal intensity of each moment in the power data sequence are obtained according to the data distribution of all reference moments of each moment in the power trend data sequence, the power seasonal data sequence and the power residual data sequence;

[0009] Obtain a local window at each moment in the power trend data sequence, and obtain the possibility of a special event occurring at each moment according to the difference between the trend strengths of power data at all moments in the local window of each moment in the power data sequence, the difference between the seasonal strengths of power data at all moments, and the mean of all residual data in the local window at each moment. Obtain the degree of preference at each moment according to the possibility of a special event occurring at each moment and the difference between the possibilities of a special event occurring at two moments. Obtain the weight of the trend item at each moment, the weight of the seasonal item at each moment, and the weight of the residual item at each moment according to the degree of preference at each moment.

[0010] According to the power trend data series, power season data series and power residual data series, each item of data is predicted to obtain each item of predicted data; according to the weights of the trend item, season item and residual item at each moment, each item of predicted data is corrected to obtain each item of corrected prediction data; according to each item of corrected prediction data, the final prediction value is obtained, and then a more scientific and reasonable power system planning and operation strategy is formulated based on the final prediction value.

[0011] Furthermore, the decomposition of the power trend data sequence into the power trend data sequence, the power season data sequence and the power residual data sequence includes the following specific steps:

[0012] The power data series is decomposed by the STL algorithm to obtain the power trend data series, power season data series and power residual data series.

[0013] Furthermore, the preceding several moments of each moment are recorded as reference moments of each moment, and according to the data distribution of all reference moments of each moment in the power trend data sequence, the power seasonal data sequence and the power residual data sequence, the power data trend intensity and the power data seasonal intensity of each moment in the power data sequence are obtained, and the specific steps included are as follows:

[0014] Record the A moments before each moment as the reference moment of each moment; wherein A is the preset first parameter;

[0015] According to the data distribution of all reference moments at each moment in the power trend data sequence and the power residual data sequence, the trend intensity of the power data at each moment in the power data sequence is obtained. According to the data distribution of all reference moments at each moment in the power season data sequence and the power residual data sequence, the seasonal intensity of the power data at each moment in the power data sequence is obtained.

[0016] Furthermore, the power data trend intensity at each moment in the power data sequence is obtained according to the data distribution of all reference moments at each moment in the power trend data sequence and the power residual data sequence, and the specific steps include the following:

[0017] The data at the time corresponding to the power trend data sequence and the power residual data sequence are added to obtain a set of sequences after addition, which is recorded as the first power data sequence; a coordinate system is established with the time as the horizontal axis and the power data as the vertical axis, and the constructed coordinate system is recorded as the reference coordinate system. The power trend data sequence and the first power data sequence are placed in the reference coordinate system, and the coordinate position of each time in each sequence can be obtained;

[0018] The slope between the data point of any reference moment at each moment in the power trend data sequence and the previous reference moment adjacent to any reference moment is recorded as the first slope of any reference moment at each moment, the slope between the data point of any reference moment at each moment in the power trend data sequence and the next reference moment adjacent to any reference moment is recorded as the second slope of any reference moment at each moment, the absolute value of the difference between the first slope and the second slope of any reference moment at each moment is recorded as the first difference of any reference moment at each moment in the power trend data sequence, and the reciprocal of the mean of the first differences of all reference moments at each moment in the power trend data sequence is recorded as the first mean at each moment in the power trend data sequence;

[0019] The variance of the data of all reference moments at each moment in the power residual data sequence is recorded as the first variance of each moment, the variance of the data of all reference moments at each moment in the first power data sequence is recorded as the second variance of each moment, the ratio of the first variance to the second variance at each moment is recorded as the first ratio at each moment, the absolute value of the difference between 1 and the first ratio at each moment is recorded as the second difference at each moment, and the product of the first mean at each moment in the power trend data sequence and the second difference at each moment is taken as the trend intensity of the power data at each moment in the power data sequence.

[0020] Furthermore, the seasonal intensity of power data at each moment in the power data sequence is obtained according to the data distribution of all reference moments at each moment in the power seasonal data sequence and the power residual data sequence, and the specific steps include the following:

[0021] The data at the time corresponding to the power season data sequence and the power residual data sequence are added to obtain a set of sequences after addition, which is recorded as the second power data sequence; the power season data sequence and the second power data sequence are placed in correspondence with each other in the reference coordinate system, and the coordinate position of each time in each sequence can be obtained;

[0022] The slope between the data point of any reference moment and the previous reference moment adjacent to any reference moment at each moment in the power seasonal data sequence is recorded as the third slope of any reference moment at each moment, the slope between the data point of any reference moment and the next reference moment adjacent to any reference moment at each moment in the power seasonal data sequence is recorded as the fourth slope of any reference moment at each moment, the absolute value of the difference between the third slope and the fourth slope at any reference moment at each moment is recorded as the third difference of any reference moment at each moment in the power seasonal data sequence, and the reciprocal of the mean of the third differences of all reference moments at each moment in the power seasonal data sequence is recorded as the second mean at each moment in the power seasonal data sequence;

[0023] The variance of the data of all reference moments at each moment in the second power data series is recorded as the third variance at each moment, the ratio of the first variance to the third variance at each moment is recorded as the second ratio at each moment, the absolute value of the difference between 1 and the second ratio at each moment is recorded as the fourth difference at each moment, and the product of the second mean at each moment in the power seasonal data series and the fourth difference at each moment is taken as the seasonal intensity of the power data at each moment in the power data series.

[0024] Furthermore, the local window at each moment in the power trend data sequence is obtained, and the possibility of a special event occurring at each moment is obtained according to the difference between the trend strength of power data at all moments in the local window at each moment in the power data sequence, the difference between the seasonal strength of power data at all moments, and the mean of all residual data in the local window at each moment, and the specific steps included are as follows:

[0025] Taking the power data at each moment in the power data sequence as the center and 2B+1 as the local window size, a local window at each moment in the power data sequence is obtained; wherein B is a preset second parameter;

[0026] The absolute value of the difference in the intensity of the trend of the power data between each moment in the power data sequence and any moment in the corresponding local window is recorded as the first difference between each moment and any moment in the corresponding local window, the product of the first difference between each moment and any moment in the corresponding local window and the time interval between each moment and any moment in the corresponding local window is recorded as the first product value of each moment and any moment in the corresponding local window, and the cumulative sum of the first product values ​​of each moment and all moments in the corresponding local window is recorded as the first value of each moment;

[0027] The absolute value of the difference in seasonal intensity of power data between each moment in the power data sequence and any moment in the corresponding local window is recorded as the second difference between each moment and any moment in the corresponding local window, the product of the second difference between each moment and any moment in the corresponding local window and the time interval between each moment and any moment in the corresponding local window is recorded as the second product value of each moment and any moment in the corresponding local window, and the cumulative sum of the second product values ​​of each moment and all moments in the corresponding local window is recorded as the second value of each moment; the sum of the first value and the second value of each moment is recorded as the third value a1 of each moment, and exp(-a1) is recorded as the fourth value of each moment;

[0028] The mean of the data in the power residual data sequence corresponding to all the moments in the local window at each moment is recorded as the third mean at each moment, and the product of the fourth value at each moment and the third mean is taken as the possibility of the occurrence of a special event at each moment;

[0029] Among them, exp() represents an exponential function with a natural constant as the base.

[0030] Furthermore, the step of obtaining the preference level at each moment according to the possibility of a special event occurring at each moment and the difference between the possibilities of a special event occurring at two moments includes the following specific steps:

[0031] Record any moment as the current moment, and in reverse order of time, group the probability of special events occurring at the same moment as the current moment into a sequence, recorded as the daily data sequence; then, in reverse order of time, group the probability of special events occurring at the same moment as the current moment into a sequence, recorded as the weekly data sequence;

[0032] The absolute value of the difference between the probability of a special event occurring at the current moment and each data in the daily data sequence at the current moment is recorded as the first eigenvalue of each data in the daily data sequence at the current moment, and the mean of the first eigenvalues ​​of all data in the daily data sequence at the current moment is recorded as the fourth mean at the current moment;

[0033] The absolute value of the difference between the probability of a special event occurring at the current moment and each data in the weekly data sequence at the current moment is recorded as the second eigenvalue of each data in the weekly data sequence at the current moment, and the mean of the second eigenvalues ​​of all data in the weekly data sequence at the current moment is recorded as the fifth mean at the current moment;

[0034] The minimum value between the fourth mean value at the current moment and the fifth mean value at the current moment is recorded as the third eigenvalue a2 at the current moment, exp(-a2) is recorded as the fourth eigenvalue at the current moment, and the product of the fourth eigenvalue at the current moment and the inverse of the probability of a special event occurring at the current moment is taken as the preference level at each moment;

[0035] Among them, exp() represents an exponential function with a natural constant as the base.

[0036] Furthermore, the method of obtaining the weight of the trend item at each moment, the weight of the seasonal item at each moment, and the weight of the residual item at each moment according to the preferred degree at each moment includes the following specific steps:

[0037] The periodicity corresponding to the third eigenvalue at the current moment is recorded as the optimal periodicity;

[0038] The preferred degrees of the F moments before each moment are selected to form a set, which is recorded as the trend item data set. According to the optimal periodicity, the preferred degrees of the moments within the F periods before each moment and corresponding to each moment are selected to form a set, which is recorded as the season item data set; wherein F is the preset third parameter;

[0039] The ratio between the degree of preference at each moment and the cumulative sum of all data in the trend item data set is used as the factor weight of the trend item at each moment;

[0040] The ratio between the degree of preference at each moment and the cumulative sum of all data in the seasonal item data set is used as the factor weight of the seasonal item at each moment;

[0041] According to the factor weight of the trend term at each moment and the factor weight of the seasonal term at each moment, the factor weight of the residual term at each moment is obtained, which is expressed as follows:

[0042]

[0043] In the formula, γ1 represents the factor weight of the trend term at each moment, γ2 represents the factor weight of the seasonal term at each moment, and γ3 represents the factor weight of the residual term at each moment;

[0044] The cumulative sum of the factor weights of the trend item at each moment, the factor weights of the seasonal item at each moment, and the factor weights of the residual item at each moment is recorded as the factor weight at each moment; the ratio of the factor weight of the trend item at each moment to the factor weight at each moment is recorded as the weight of the trend item at each moment, the ratio of the factor weight of the seasonal item at each moment to the factor weight at each moment is recorded as the weight of the seasonal item at each moment, and the ratio of the factor weight of the residual item at each moment to the factor weight at each moment is recorded as the weight of the residual item at each moment.

[0045] Further, the method predicts each item of data according to the power trend data sequence, the power season data sequence and the power residual data sequence to obtain each item of predicted data; corrects each item of predicted data according to the weights of the trend item, the season item and the residual item at each moment to obtain each item of corrected predicted data; and obtains the final predicted value according to each item of corrected predicted data, including the following specific steps:

[0046] According to the power trend data sequence, the trend item data is predicted by the exponential smoothing algorithm to obtain the data after the trend item prediction; according to the power season data sequence, the seasonal item data is predicted by the seasonal autoregressive moving average model to obtain the data after the seasonal item prediction; according to the power residual data sequence, the residual item data is predicted by the autoregressive integrated moving average model to obtain the data after the residual item prediction;

[0047] The product of the predicted data of the trend item at each moment and the weight of the trend item at the closest moment is recorded as the predicted value of the trend item at each moment, the product of the predicted data of the seasonal item at each moment and the weight of the seasonal item at the closest moment is recorded as the predicted value of the seasonal item at each moment, and the product of the predicted data of the residual item at each moment and the weight of the residual item at the closest moment is recorded as the predicted value of each residual item;

[0048] The predicted value of the trend item at each moment, the predicted value of the seasonal item at each moment, and the predicted value of each residual item are added together to obtain the adjusted predicted value at each moment, and all the final predicted values ​​are obtained based on the adjusted predicted values ​​at all moments.

[0049] The present invention also provides an electric power operation and maintenance management system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any one of the above-mentioned electric power operation and maintenance management methods when executing the computer program.

[0050] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the power data trend intensity and the power data seasonal intensity at each moment in the power data sequence according to the data distribution of all reference moments in the power trend data sequence, the power season data sequence and the power residual data sequence, thereby improving the accuracy of trend and seasonal analysis in the power data; according to the difference between the power data trend intensity at all moments in the local window of each moment in the power data sequence, the difference between the power data seasonal intensity at all moments, and the mean of all residual data in the local window of each moment, the possibility of special events occurring at each moment is obtained, and according to the possibility of special events occurring at each moment, the possibility of special events occurring at two moments The difference between the possibilities of special events is used to obtain the degree of preference at each moment, and the weight of the trend item at each moment, the weight of the seasonal item at each moment and the weight of the residual item at each moment are obtained according to the degree of preference at each moment, thereby improving the degree of adjustment of the predicted value at each moment; the weight of the trend item at each moment, the weight of the seasonal item at each moment and the weight of the residual item at each moment are obtained according to the degree of preference at each moment, and the data after each prediction are corrected to obtain the corrected data of each prediction; the final prediction value is obtained according to the corrected data of each prediction, and then a more scientific and reasonable power system planning and operation strategy is formulated according to the final prediction value, thereby improving the accuracy of power data prediction and the reliability of power system planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 A flowchart of the steps of a power operation and maintenance management method of the present invention;

[0053] Figure 2 This is a flowchart of power operation and maintenance management. DETAILED DESCRIPTION

[0054] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a power operation and maintenance management method and system proposed according to the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0055] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0056] The following describes in detail a specific scheme of a power operation and maintenance management method and system provided by the present invention in conjunction with the accompanying drawings.

[0057] See also Figure 1 , which shows a flow chart of a method for power operation and maintenance management provided by an embodiment of the present invention, the method comprising the following steps:

[0058] Step S001: continuously collect power data at several moments.

[0059] It should be noted that in order to analyze the changes in power data, it is necessary to collect power data, predict the power data based on the collected power data, and plan and operate the power system based on the predicted power data.

[0060] Specifically, all power data within three months are collected with a sampling interval of ten minutes.

[0061] At this point, the power data at several moments are obtained.

[0062] Step S002: Record the first several moments of each moment as the reference moments of each moment, and obtain the power data trend intensity and power data seasonal intensity of each moment in the power data sequence according to the data distribution of all reference moments of each moment in the power trend data sequence, the power season data sequence and the power residual data sequence.

[0063] It should be noted that the size of the electricity data is not only related to the trend of electricity consumption, but also to different time periods, that is, it is related to seasonality. Therefore, by decomposing the electricity data, we can obtain trend terms, seasonal terms and residual terms. Therefore, by decomposing the electricity data and decomposing the different terms, we can further analyze and predict the data.

[0064] Specifically, the power data of several moments are sorted in chronological order to obtain a power data sequence. The power data sequence is decomposed by the STL algorithm to obtain a power trend data sequence, a power season data sequence and a power residual data sequence. Among them, the STL algorithm is not a known technology and will not be described in detail here.

[0065] At this point, the power trend data series, power season data series and power residual data series are obtained.

[0066] It should be noted that due to the time periods corresponding to some special holidays, performances or competitions in daily life, the corresponding power data may suddenly become high. The suddenly high power data appears abnormal compared with the surrounding adjacent power data. Therefore, the power data is further analyzed by analyzing the changes in the power data.

[0067] It should be further explained that in order to analyze the degree to which power data are affected by trend items and seasonal items, the data changes of the power trend data series and the power season data series are analyzed respectively to obtain the strength of the trend items and seasonal items; the degree to which the trend item is affected is analyzed based on the data changes and distribution in the power trend data series. The greater the degree of impact on the trend item, the smaller the trend strength of the corresponding power data; the smaller the degree of impact on the trend item, the greater the trend strength of the corresponding power data.

[0068] Specifically, the data at the time corresponding to the power trend data sequence and the power residual data sequence are added to obtain a set of sequences after addition, which is recorded as the first power data sequence; the data at the time corresponding to the power season data sequence and the power residual data sequence are added to obtain a set of sequences after addition, which is recorded as the second power data sequence. A coordinate system is established with time as the horizontal axis and power data as the vertical axis. The constructed coordinate system is recorded as the reference coordinate system. The power trend data sequence, the first power data sequence and the second power data sequence are corresponding to each other in the reference coordinate system, and the coordinate position of each moment in each sequence can be obtained. The slope between adjacent data is calculated by the position coordinates.

[0069] A first parameter A is preset, wherein this embodiment is described by taking A=100 as an example, and this embodiment is not specifically limited, wherein A can be determined according to specific implementation conditions. The first A moments of each moment are recorded as the reference moments of each moment, and thus the reference moment of each moment is obtained.

[0070] According to the difference between the data of all adjacent reference moments at each moment in the power trend data sequence, the distribution of the data of all reference moments at each moment in the power residual data sequence, and the distribution of the data of all reference moments at each moment in the first power data sequence, the power data trend intensity at each moment in the power data sequence is obtained. As an embodiment, the specific calculation method is:

[0071]

[0072] In the formula, K j,i-1,i K represents the slope between the data points at the i-1th reference time and the i-th reference time at the jth time in the power trend data series. j,i,i+1represents the slope between the data points at the i-th reference time and the i+1-th reference time at the j-th time in the power trend data sequence, A represents the number of all reference times at each time, σ j represents the variance of all reference data at the jth moment in the power residual data sequence, σ1 j represents the variance of the data of all reference times at the jth time in the first power data sequence, || is the absolute value symbol, Q j Represents the trend strength of the power data at the jth moment in the power data sequence.

[0073] Among them, |K j,i-1,i -K j,i,i+1 | represents the difference between adjacent slopes. When the difference is smaller, it means that the data in the power trend data sequence has higher consistency, that is, the data changes gradually, and it also indicates that the possibility of abnormal noise data in the data is smaller, that is, the corresponding power data trend is less affected, and the intensity of the power data trend is greater; when the difference is larger, it means that the consistency of the data in the power trend data sequence is lower, that is, the data does not change gradually, and the distribution of the data may be one high and one low, which also indicates that the possibility of abnormal noise data in the data is greater, that is, the corresponding power data trend is more affected, and the intensity of the power data trend is smaller. It represents the ratio of the variance of the data of all reference moments at each moment in the power residual data sequence to the variance of the data of all reference moments at each moment in the first power data sequence. The greater the difference between the ratio and 1, the greater the trend strength of the power data; the smaller the difference between the ratio and 1, the smaller the trend strength of the power data.

[0074] It should be noted that when analyzing the degree of influence of seasonal items based on the data changes and distribution in the power seasonal data series, the greater the influence of seasonal items, the smaller the seasonal intensity of the corresponding power data; the smaller the influence of seasonal items, the greater the seasonal intensity of the corresponding power data.

[0075] According to the difference between the data of all adjacent reference moments at each moment in the power seasonal data sequence, the distribution of the data of all reference moments at each moment in the power residual data sequence, and the distribution of the data of all reference moments at each moment in the first power data sequence, the seasonal intensity of the power data at each moment in the power data sequence is obtained. As an embodiment, the specific calculation method is:

[0076]

[0077] In the formula, K j ′ ,i-1,iK represents the slope between the data points at the i-1th reference time and the i-th reference time at the jth time in the power season data series. j ′ ,i,i+1 represents the slope between the data points at the i-th reference time and the i+1-th reference time at the j-th time in the power season data series, A represents the number of all reference times at each time, σ j represents the variance of all reference data at the jth moment in the power residual data sequence, σ2 j represents the variance of the data of all reference times at the jth time in the second power data sequence, || is the absolute value symbol, W j Represents the seasonal intensity of power data at the jth moment in the power data series.

[0078] Among them, |K j ′ ,i-1,i -K j ′ ,i,i+1 | represents the difference between adjacent slopes. When the difference is larger, it means that the data in the power seasonal data sequence has a higher consistency, that is, the data changes gradually, and it also indicates that the possibility of abnormal noise data in the data is smaller, that is, the corresponding power data seasonality is affected less, and the seasonal intensity of the power data is greater; when the difference is smaller, it means that the consistency of the data in the power seasonal data sequence is lower, that is, the data does not change gradually, and the distribution of the data may be one high and one low, which also indicates that the possibility of abnormal noise data in the data is greater, that is, the corresponding power data seasonality is affected more, and the seasonal intensity of the power data is smaller. The ratio of the variance of the data of all reference moments at each moment in the power residual data sequence to the variance of the data of all reference moments at each moment in the second power data sequence. The greater the difference between the ratio and 1, the greater the seasonal intensity of the power data; the smaller the difference between the ratio and 1, the smaller the seasonal intensity of the power data.

[0079] At this point, the trend intensity and seasonal intensity of the power data at each moment in the power data series are obtained.

[0080] Step S003: Obtain a local window at each moment in the power trend data sequence, and obtain the possibility of special events occurring at each moment according to the difference between the trend strengths of power data at all moments in the local window of each moment in the power data sequence, the difference between the seasonal strengths of power data at all moments, and the mean of all residual data in the local window at each moment. Obtain the degree of preference for each moment according to the possibility of special events occurring at each moment and the difference between the possibilities of special events occurring at two moments, and obtain the weight of the trend item at each moment, the weight of the seasonal item at each moment, and the weight of the residual item at each moment according to the degree of preference for each moment.

[0081] It should be noted that in addition to smaller random fluctuations, power data is also prone to larger fluctuations due to special events, such as large-scale activities in some areas, sudden weather changes, etc. These events will cause the stability and regularity of power data to be destroyed, resulting in the weakening of the trend and seasonal strength of the data. Therefore, the possibility of special events occurring at each moment is analyzed by analyzing the difference between the trend and seasonal strength of the power data at each moment in the power data sequence and the trend and seasonal strength of the power data at the surrounding adjacent moments.

[0082] Specifically, a second parameter B is preset, wherein this embodiment is described by taking B=20 as an example, and this embodiment is not specifically limited, wherein B may be determined according to specific implementation conditions. The power data at each moment in the power data sequence is taken as the center, and 2B+1 is taken as the local window size to obtain a local window at each moment in the power data sequence; wherein the number of left data points in the local window at each moment is equal to the number of right data points, which is B.

[0083] According to the difference between the trend strength of power data at all moments in the local window at each moment in the power data sequence, the difference between the seasonal strength of power data at all moments in the local window at each moment, and the mean of all residual data in the local window at each moment, the possibility of a special event occurring at each moment is obtained. As an embodiment, the specific calculation method is:

[0084]

[0085] In the formula, Q c represents the trend strength of the power data at the cth moment in the power data sequence, Q c,v represents the trend strength of the power data at the vth moment in the local window at the cth moment in the power data sequence, W c represents the seasonal intensity of the power data at the cth moment in the power data series, W c,v represents the seasonal intensity of the power data at the vth moment in the local window at the cth moment in the power data series, Tc,v represents the time interval between the cth moment in the power data sequence and the vth moment in the local window of the cth moment, R c,v represents the data in the power residual data sequence corresponding to the vth moment in the local window of the cth moment in the power data sequence, M represents the number of all moments in the local window of each moment in the power data sequence, || is the absolute value symbol, P c It represents the probability of a special event occurring at the cth moment, and exp() represents an exponential function with a natural constant as the base.

[0086] Among them, |Q c -Q c,v | represents the difference between the trend strength of power data at two moments in the local window of each moment in the power data series, |W c -W c,v | represents the difference between the seasonal intensity of power data at two moments in the local window of each moment in the power data sequence; when the two differences are larger, it means that there is a high probability that a special event has occurred at one of the two moments, causing a sudden change in power consumption; when the two differences are smaller, it means that there is a small probability that a special event has occurred at the two moments, and there is no sudden change in power consumption. AS |Q c -Q c,v | and |W c -W c,v |The weight of |, that is, the smaller the time interval between two moments, the higher the reliability of the obtained difference for analysis. It represents the mean of the data in the power residual data sequence corresponding to all moments in the local window at each moment. When the mean of the residual value is smaller, it means that the mutation occurs less frequently. When the mean of the residual value is larger, it means that the mutation occurs more frequently.

[0087] At this point, we get the probability of a special event occurring at each moment.

[0088] It should be noted that since the power data in different time periods are different, that is, the power consumption in the morning and evening will be higher, the power data in different time periods of the day are different, so the power data has daily periodicity. In a week, the power consumption is higher during the week and lower on weekends, so the power data also has weekly periodicity; therefore, the collected power data is analyzed by daily periodicity and weekly periodicity to determine which periodicity is more affected, and then the preferred degree of power data is obtained according to the strongest periodicity.

[0089] Specifically, any moment is recorded as the current moment, and the probability of special events occurring at the same moment as the current moment every day is grouped into a sequence in reverse order of time, which is recorded as the daily data sequence; then, the probability of special events occurring at the same moment as the current moment every week is grouped into a sequence in reverse order of time, which is recorded as the weekly data sequence.

[0090] According to the difference between the data in the daily data sequence and the difference between the data in the weekly data sequence corresponding to the current moment, the preference degree at the current moment is obtained. As an embodiment, the specific calculation method is as follows:

[0091]

[0092] In the formula, P represents the possibility of a special event occurring at the current moment, P r represents the rth data in the daily data sequence at the current time, p e It represents the e-th data in the weekly data sequence at the current moment, n represents the number of all data in the daily data sequence at the current moment, m represents the number of all data in the weekly data sequence at the current moment, || is the absolute value symbol, min() represents the minimum value function, exp() represents the exponential function with natural constant as the base, and Y represents the preference degree at the current moment.

[0093] in, The difference between the data in the daily data series representing the daily periodicity, The differences between the data in the weekly data series representing weekly periodicity are selected to analyze the preference of each moment. That is, when the difference between the data is smaller, the preference of each moment is greater, and vice versa. When the possibility of a special event occurring at each moment is greater, the preference of each moment is smaller, and vice versa, the preference of each moment is greater.

[0094] The periodicity corresponding to the smallest difference between the differences between the data in the daily data sequence with daily periodicity and the differences between the data in the weekly data sequence with weekly periodicity is recorded as the optimal periodicity.

[0095] A third parameter F is preset, wherein this embodiment is described by taking F=5 as an example, and this embodiment is not specifically limited, wherein F can be determined according to specific implementation conditions.

[0096] The preferred degrees of the F moments before each moment are selected to form a set, which is recorded as the trend item data set. According to the optimal periodicity, the preferred degrees of the moments within the F periods before each moment and corresponding to each moment are selected to form a set, which is recorded as the season item data set.

[0097] According to the preference degree of each moment and the preference degrees of all moments in the trend item data set, the factor weight of the trend item at each moment is obtained. As an embodiment, the specific calculation method is:

[0098]

[0099] In the formula, Y represents the preference level at each moment, Y s It represents the preference degree at the sth moment, H represents the number of data in the trend item data set at each moment, and γ1 represents the factor weight of the trend item at each moment.

[0100] According to the preference degree of each moment and the preference degrees of all moments in the seasonal item data set, the factor weight of the seasonal item at each moment is obtained. As an embodiment, the specific calculation method is:

[0101]

[0102] In the formula, Y represents the preference level at each moment, Y z It represents the preference degree of the zth moment, G represents the number of data in the seasonal item data set at each moment, and γ2 represents the factor weight of the seasonal item at each moment.

[0103] According to the factor weight of the trend item at each moment and the factor weight of the seasonal item at each moment, the factor weight of the residual item at each moment is obtained. As an embodiment, the specific calculation method is:

[0104]

[0105] In the formula, γ1 represents the factor weight of the trend term at each moment, γ2 represents the factor weight of the seasonal term at each moment, γ3 represents the factor weight of the residual term at each moment, and F is the preset third parameter.

[0106] The cumulative sum of the factor weights of the trend item at each moment, the factor weights of the seasonal item at each moment, and the factor weights of the residual item at each moment is recorded as the factor weight at each moment; the ratio of the factor weight of the trend item at each moment to the factor weight at each moment is recorded as the weight of the trend item at each moment, the ratio of the factor weight of the seasonal item at each moment to the factor weight at each moment is recorded as the weight of the seasonal item at each moment, and the ratio of the factor weight of the residual item at each moment to the factor weight at each moment is recorded as the weight of the residual item at each moment.

[0107] At this point, the weight of the trend term at each moment, the weight of the seasonal term at each moment, and the weight of the residual term at each moment are obtained.

[0108] Step S004: predict each item of data according to the power trend data sequence, the power season data sequence and the power residual data sequence to obtain each item of predicted data; correct each item of predicted data according to the weights of the trend item, the season item and the residual item at each moment to obtain each item of corrected prediction data; obtain the final prediction value according to the corrected prediction data, and then formulate a more scientific and reasonable power system planning and operation strategy based on the final prediction value.

[0109] According to the power trend data sequence, the trend item data is predicted by the exponential smoothing algorithm to obtain the data after the trend item prediction; according to the power season data sequence, the seasonal item data is predicted by the seasonal autoregressive moving average model to obtain the data after the seasonal item prediction; according to the power residual data sequence, the residual item data is predicted by the autoregressive integrated moving average model to obtain the data after the residual item prediction.

[0110] The product of the predicted data of the trend item at each moment and the weight of the trend item at the closest moment to each moment is recorded as the predicted value of the trend item at each moment. The product of the predicted data of the seasonal item at each moment and the weight of the seasonal item at the closest moment to each moment is recorded as the predicted value of the seasonal item at each moment. The product of the predicted data of the residual item at each moment and the weight of the residual item at the closest moment to each moment is recorded as the predicted value of each residual item.

[0111] The predicted value of the trend item at each moment, the predicted value of the seasonal item at each moment, and the predicted value of each residual item are added together to obtain the final predicted value at each moment.

[0112] According to the final prediction values ​​of all future moments, a more scientific and reasonable power system planning and operation strategy is formulated to ensure the key activities of normal operation and reliability of the power system. Among them, the flowchart of power operation and maintenance management is as follows Figure 2 shown.

[0113] It should be noted that the exp(-x) model used in this embodiment is only used to indicate that the negative correlation and the result of the constraint model output are in the interval (0,1). In specific implementation, it can be replaced with other models with the same purpose. This embodiment only uses the exp(-x) model as an example for description without making specific limitations on it, where x refers to the input of the model.

[0114] This embodiment provides an electric power operation and maintenance management system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, an electric power operation and maintenance management method in steps S001 to S004 is implemented.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A power operation and maintenance management method, characterized in that: The method comprises the following steps: Acquire power data at a number of consecutive moments, organize the power data at a number of consecutive moments into a power data sequence according to the time sequence, and decompose the power trend data sequence into a power trend data sequence, a power season data sequence, and a power residual data sequence; The previous several moments of each moment are recorded as the reference moments of each moment, and the power data trend intensity and power data seasonal intensity of each moment in the power data sequence are obtained according to the data distribution of all reference moments of each moment in the power trend data sequence, the power seasonal data sequence and the power residual data sequence; Obtain a local window at each moment in the power trend data sequence, and obtain the possibility of a special event occurring at each moment according to the difference between the trend strengths of power data at all moments in the local window of each moment in the power data sequence, the difference between the seasonal strengths of power data at all moments, and the mean of all residual data in the local window at each moment. Obtain the degree of preference at each moment according to the possibility of a special event occurring at each moment and the difference between the possibilities of a special event occurring at two moments. Obtain the weight of the trend item at each moment, the weight of the seasonal item at each moment, and the weight of the residual item at each moment according to the degree of preference at each moment. Predict various data according to the power trend data series, power season data series and power residual data series to obtain the data after each prediction; correct the data after each prediction according to the weight of the trend item, season item and residual item at each moment to obtain the corrected data after each prediction; obtain the final prediction value according to the corrected data of each prediction, and then formulate a more scientific and reasonable power system planning and operation strategy according to the final prediction value; The method records the previous several moments of each moment as the reference moment of each moment, and obtains the power data trend intensity and power data seasonal intensity of each moment in the power data sequence according to the data distribution of all reference moments of each moment in the power trend data sequence, the power seasonal data sequence and the power residual data sequence, including the following specific steps: The previous moment of each The moments are recorded as the reference moments of each moment; among them, is to preset the first parameter; According to the data distribution of all reference moments at each moment in the power trend data sequence and the power residual data sequence, the trend intensity of the power data at each moment in the power data sequence is obtained. According to the data distribution of all reference moments at each moment in the power season data sequence and the power residual data sequence, the seasonal intensity of the power data at each moment in the power data sequence is obtained.

2. A power operation and maintenance management method according to claim 1, characterized in that: The specific steps of decomposing the power trend data sequence into the power trend data sequence, the power season data sequence and the power residual data sequence are as follows: The power data series is decomposed by the STL algorithm to obtain the power trend data series, power season data series and power residual data series.

3. According to claim 1, a power operation and maintenance management method is characterized in that: The method of obtaining the power data trend intensity at each moment in the power data sequence according to the data distribution of all reference moments at each moment in the power trend data sequence and the power residual data sequence includes the following specific steps: Adding the data of the power trend data sequence and the power residual data sequence at the corresponding time to obtain a set of added sequences, which is recorded as a first power data sequence; A coordinate system is established with time as the horizontal axis and power data as the vertical axis. The constructed coordinate system is recorded as a reference coordinate system. The power trend data sequence and the first power data sequence are mapped to the reference coordinate system, and the coordinate position of each moment in each sequence can be obtained. The slope between the data point of any reference moment at each moment in the power trend data sequence and the previous reference moment adjacent to any reference moment is recorded as the first slope of any reference moment at each moment, the slope between the data point of any reference moment at each moment in the power trend data sequence and the next reference moment adjacent to any reference moment is recorded as the second slope of any reference moment at each moment, the absolute value of the difference between the first slope and the second slope of any reference moment at each moment is recorded as the first difference of any reference moment at each moment in the power trend data sequence, and the reciprocal of the mean of the first differences of all reference moments at each moment in the power trend data sequence is recorded as the first mean at each moment in the power trend data sequence; The variance of the data of all reference moments at each moment in the power residual data sequence is recorded as the first variance of each moment, the variance of the data of all reference moments at each moment in the first power data sequence is recorded as the second variance of each moment, the ratio of the first variance to the second variance at each moment is recorded as the first ratio at each moment, the absolute value of the difference between 1 and the first ratio at each moment is recorded as the second difference at each moment, and the product of the first mean at each moment in the power trend data sequence and the second difference at each moment is taken as the trend intensity of the power data at each moment in the power data sequence.

4. A power operation and maintenance management method according to claim 3, characterized in that: The method of obtaining the seasonal intensity of power data at each moment in the power data sequence according to the data distribution of all reference moments at each moment in the power seasonal data sequence and the power residual data sequence includes the following specific steps: The data at the time corresponding to the power season data sequence and the power residual data sequence are added to obtain a set of sequences after addition, which is recorded as the second power data sequence; the power season data sequence and the second power data sequence are placed in correspondence with each other in the reference coordinate system, and the coordinate position of each time in each sequence can be obtained; The slope between the data point of any reference moment and the previous reference moment adjacent to any reference moment at each moment in the power seasonal data sequence is recorded as the third slope of any reference moment at each moment, the slope between the data point of any reference moment and the next reference moment adjacent to any reference moment at each moment in the power seasonal data sequence is recorded as the fourth slope of any reference moment at each moment, the absolute value of the difference between the third slope and the fourth slope at any reference moment at each moment is recorded as the third difference of any reference moment at each moment in the power seasonal data sequence, and the reciprocal of the mean of the third differences of all reference moments at each moment in the power seasonal data sequence is recorded as the second mean at each moment in the power seasonal data sequence; The variance of the data of all reference moments at each moment in the second power data series is recorded as the third variance at each moment, the ratio of the first variance to the third variance at each moment is recorded as the second ratio at each moment, the absolute value of the difference between 1 and the second ratio at each moment is recorded as the fourth difference at each moment, and the product of the second mean at each moment in the power seasonal data series and the fourth difference at each moment is taken as the seasonal intensity of the power data at each moment in the power data series.

5. The power operation and maintenance management method according to claim 1, characterized in that: The method of obtaining a local window at each moment in the power trend data sequence, and obtaining the possibility of a special event occurring at each moment according to the difference between the power data trend strength at all moments in the local window at each moment in the power data sequence, the difference between the power data seasonal strength at all moments, and the mean of all residual data in the local window at each moment, includes the following specific steps: The power data at each moment in the power data sequence is taken as the center. is the local window size, and the local window at each moment in the power data sequence is obtained; where, To preset the second parameter; The absolute value of the difference in the intensity of the trend of the power data between each moment in the power data sequence and any moment in the corresponding local window is recorded as the first difference between each moment and any moment in the corresponding local window, the product of the first difference between each moment and any moment in the corresponding local window and the time interval between each moment and any moment in the corresponding local window is recorded as the first product value of each moment and any moment in the corresponding local window, and the cumulative sum of the first product values ​​of each moment and all moments in the corresponding local window is recorded as the first value of each moment; The absolute value of the difference in the seasonal intensity of the power data between each moment in the power data sequence and any moment in the corresponding local window is recorded as the second difference between each moment and any moment in the corresponding local window. The product of the second difference between each moment and any moment in the corresponding local window and the time interval between each moment and any moment in the corresponding local window is recorded as the second product value of each moment and any moment in the corresponding local window. The cumulative sum of the second product values ​​of each moment and all moments in the corresponding local window is recorded as the second value of each moment. The sum of the first value and the second value of each moment is recorded as the third value of each moment. ,Will Recorded as the fourth value at each moment; The mean of the data in the power residual data sequence corresponding to all the moments in the local window at each moment is recorded as the third mean at each moment, and the product of the fourth value at each moment and the third mean is taken as the possibility of the occurrence of a special event at each moment; in, Represents an exponential function with a natural constant as its base.

6. The power operation and maintenance management method according to claim 1, characterized in that: The method of obtaining the preferred degree of each moment according to the possibility of a special event occurring at each moment and the difference between the possibilities of a special event occurring at two moments comprises the following specific steps: Record any moment as the current moment, and in reverse order of time, group the probability of special events occurring at the same moment as the current moment into a sequence, recorded as the daily data sequence; then, in reverse order of time, group the probability of special events occurring at the same moment as the current moment into a sequence, recorded as the weekly data sequence; The absolute value of the difference between the probability of a special event occurring at the current moment and each data in the daily data sequence at the current moment is recorded as the first eigenvalue of each data in the daily data sequence at the current moment, and the mean of the first eigenvalues ​​of all data in the daily data sequence at the current moment is recorded as the fourth mean at the current moment; The absolute value of the difference between the probability of a special event occurring at the current moment and each data in the weekly data sequence at the current moment is recorded as the second eigenvalue of each data in the weekly data sequence at the current moment, and the mean of the second eigenvalues ​​of all data in the weekly data sequence at the current moment is recorded as the fifth mean at the current moment; The minimum value between the fourth mean at the current moment and the fifth mean at the current moment is recorded as the third eigenvalue at the current moment ,Will Recorded as the fourth eigenvalue at the current moment, the product of the fourth eigenvalue at the current moment and the inverse of the probability of a special event occurring at the current moment is taken as the preference level at each moment; in, Represents an exponential function with a natural constant as its base.

7. A power operation and maintenance management method according to claim 6, characterized in that: The method of obtaining the weight of the trend item at each moment, the weight of the seasonal item at each moment, and the weight of the residual item at each moment according to the degree of preference at each moment includes the following specific steps: The periodicity corresponding to the third eigenvalue at the current moment is recorded as the optimal periodicity; The preferred degrees of the F moments before each moment are selected to form a set, which is recorded as the trend item data set. According to the optimal periodicity, the preferred degrees of the moments within the F periods before each moment and corresponding to each moment are selected to form a set, which is recorded as the season item data set; wherein F is the preset third parameter; The ratio between the degree of preference at each moment and the cumulative sum of all data in the trend item data set is used as the factor weight of the trend item at each moment; The ratio between the degree of preference at each moment and the cumulative sum of all data in the seasonal item data set is used as the factor weight of the seasonal item at each moment; According to the factor weight of the trend term at each moment and the factor weight of the seasonal term at each moment, the factor weight of the residual term at each moment is obtained, which is expressed as follows: ； In the formula, Represents the factor weight of the trend item at each moment, represents the factor weight of the seasonal term at each moment, Represents the factor weight of the residual term at each moment; The cumulative sum of the factor weights of the trend item at each moment, the factor weights of the seasonal item at each moment, and the factor weights of the residual item at each moment is recorded as the factor weight at each moment; the ratio of the factor weight of the trend item at each moment to the factor weight at each moment is recorded as the weight of the trend item at each moment, the ratio of the factor weight of the seasonal item at each moment to the factor weight at each moment is recorded as the weight of the seasonal item at each moment, and the ratio of the factor weight of the residual item at each moment to the factor weight at each moment is recorded as the weight of the residual item at each moment.

8. The power operation and maintenance management method according to claim 1, characterized in that: The data are predicted according to the power trend data sequence, the power season data sequence and the power residual data sequence to obtain the predicted data; According to the weights of trend items, seasonal items and residual items at each moment, the data after each forecast is corrected to obtain the corrected data of each forecast; the final forecast value is obtained according to the corrected data of each forecast, including the following specific steps: According to the power trend data sequence, the trend item data is predicted by the exponential smoothing algorithm to obtain the data after the trend item prediction; according to the power season data sequence, the seasonal item data is predicted by the seasonal autoregressive moving average model to obtain the data after the seasonal item prediction; according to the power residual data sequence, the residual item data is predicted by the autoregressive integrated moving average model to obtain the data after the residual item prediction; The product of the predicted data of the trend item at each moment and the weight of the trend item at the closest moment is recorded as the predicted value of the trend item at each moment, the product of the predicted data of the seasonal item at each moment and the weight of the seasonal item at the closest moment is recorded as the predicted value of the seasonal item at each moment, and the product of the predicted data of the residual item at each moment and the weight of the residual item at the closest moment is recorded as the predicted value of each residual item; The predicted value of the trend item at each moment, the predicted value of the seasonal item at each moment, and the predicted value of each residual item are added together to obtain the adjusted predicted value at each moment, and all the final predicted values ​​are obtained based on the adjusted predicted values ​​at all moments.

9. An electric power operation and maintenance management system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the power operation and maintenance management method as described in any one of claims 1-8 are implemented.

Images