An intelligent power supply reliability index grading prediction method

Abstract

The application discloses a kind of intelligent power supply reliability index classification prediction method, it is related to power grid operation and maintenance management field.The related historical data of power supply reliability of power grid system is obtained, the power supply reliability index value in fixed period is calculated, the power supply reliability index value and historical data are analyzed by grey correlation degree, the data category within the correlation degree threshold is filtered as the important influence factor of power supply reliability index, the real-time data of the important influence factor of power supply reliability of power grid system is collected, the homogeneity between the mean value of real-time data and the mean value of historical data is calculated by matching model, the historical data of the important influence factor within the homogeneity threshold range is filtered as homogenization parameter as input, with the corresponding power supply reliability index value calculation result as output to train and obtain neural network prediction model, the prediction value of the power supply reliability index in real-time data is output by prediction model.

Description

Technical Field

[0001] This invention relates to the field of power grid operation and maintenance management, and in particular to an intelligent method for predicting power supply reliability indicators in a graded manner. Background Technology

[0002] The quality of a power grid needs to be measured through the calculation of multiple reliability data. Key indicators include average user outage time (AIHC-1, AIHC-2, AIHC-3), average duration of fault outage (MID-F), and power supply reliability (RS-1, RS-2, RS-3).

[0003] Existing methods for predicting power supply reliability indicators in power grids mainly include probabilistic estimation and neural network prediction based on machine learning. Both methods rely on analyzing historical data of the power grid to achieve prediction results. However, because power supply reliability indicators are easily affected by external factors, such as environmental factors and actual usage conditions, the values ​​obtained using existing prediction methods often deviate significantly from the actual values. Therefore, we propose an intelligent hierarchical prediction method for power supply reliability indicators. Summary of the Invention

[0004] The main objective of this invention is to provide an intelligent method for predicting power supply reliability indicators in a graded manner. By selecting historical data of important influencing factors within the homogeneity threshold range as homogeneity parameters as input, and using the corresponding power supply reliability indicator values ​​as output, a neural network prediction model is trained to obtain a prediction model. The prediction model outputs the predicted value of the power supply reliability indicator in real-time data, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for hierarchical prediction of intelligent power supply reliability indicators, comprising:

[0007] Step 1: Obtain historical data related to the power supply reliability of the power grid system, and calculate the power supply reliability index value within any fixed period.

[0008] The power supply reliability indicators include average user outage time, specifically AIHC-1, AIHC-2, AIHC-3, average duration of fault outages, power supply reliability rate, average number of users experiencing power outages, average number of user outages, and average power supply reliability rate. Historical data includes ledger data, operational data, load data, and environmental data.

[0009] The ledger data includes reliability parameters such as model, installation information, nameplate parameters, rated current, rated voltage, and line length of each component in the power grid system.

[0010] Operational data includes measurement data such as the power factor, active power, reactive power, voltage, and current of components; status data includes dispatch data, power grid safety information, load node anomaly data, and equipment start-up and disconnection status.

[0011] Load data includes real-time load size of each distribution area and predicted load of each load node;

[0012] Environmental data includes atmospheric pressure, temperature, solar radiation intensity, wind speed, and wind direction.

[0013] Step two involves performing grey relational analysis on the obtained power supply reliability index values ​​and historical data to calculate the correlation degree between the power supply reliability index values ​​and different categories of data. Based on the obtained correlation degree values, the data category association order of the degree of influence of the power supply reliability index is determined.

[0014] The specific steps are as follows:

[0015] S21, construct a data sequence of power supply reliability index values ​​and historical data of different categories, denoted as:

[0016] E = (E1, E2, ..., E i )

[0017] F = (F1, F2, ..., F j )

[0018] In the formula, E is the data sequence of power supply reliability index values, E i Let F be the i-th power supply reliability index, and F be the data sequence of historical data for each category. j For the j-th category data value;

[0019] S22, standardizes each data point in the data sequence, using the following formula:

[0020]

[0021]

[0022] In the formula, E iq Let e ​​be the standardized value of the q-th value of the i-th power supply reliability index. iq Let q be the value of the i-th power supply reliability index. Let F be the mean of the i-th power supply reliability index. jq f is the standardized value of the q-th value of the j-th category data.jq For the q-th value of the j-th category, Let m1, m2, and n be the mean of the data in the j-th category, where m1, m2, and n are the data sample sizes.

[0023] S23, Calculate the correlation values ​​of all data points in the data sequence. The calculation formula is:

[0024]

[0025] In the formula, ξ jq Let F be the correlation coefficient between each data point in data sequence F and data sequence E, where min i [min q (|E iq -F jq |)] is used to obtain |E iq -F jq |Minimum value of the calculated result q Then take the minimum value among all the calculated minimum values ​​(·). i [·], max i [max q (|E iq -F jq |)] is used to obtain |E iq -F jq The maximum value of the calculated result is max. q Then, take the maximum value among all calculated maximum values ​​(·). i [·], where ρ is the resolution coefficient, with a value range of (0,1], and is usually taken as ρ=0.5;

[0026] S24. Calculate the correlation degree value based on the correlation coefficient calculation results. The calculation formula is as follows:

[0027]

[0028] In the formula, R ij The correlation between the i-th power supply reliability index and the j-th category of data;

[0029] S25, the correlation between the i-th power supply reliability index and the j-th category data is arranged in order of magnitude to obtain the data category correlation order of the influence degree of the power supply reliability index;

[0030] Step 3: Set the correlation threshold [Δr] for the degree of influence of each of the power supply reliability indicators. imin ,1), where Δr iminLet Δr be the minimum correlation threshold for the i-th power supply reliability index value. Based on the obtained data category correlation order, data categories within this correlation threshold are selected as important influencing factors for the power supply reliability index. imin It is determined based on an empirical formula, which is:

[0031]

[0032] In the formula, r imax r is the maximum correlation value in the correlation order of the i-th power supply reliability index. imin Let γ be the minimum correlation value in the correlation sequence of the i-th power supply reliability index, γ be the number of correlation values ​​in the correlation sequence of the i-th power supply reliability index, and ν be the correlation value in the correlation sequence of the i-th power supply reliability index that is greater than the mean correlation value. The number of;

[0033] Step four: Collect real-time data on key influencing factors of power supply reliability in the power grid system. The sampling time for the collected real-time data is less than a fixed period. Calculate the mean of the key influencing factors in historical data within any fixed period and the mean of the real-time data. Calculate the homogeneity between the mean of the real-time data and the mean of the historical data using a matching model. Set a homogeneity threshold range and select historical data of the key influencing factors within the homogeneity threshold range as homogeneity parameters. The expression for the matching model is:

[0034]

[0035] In the formula, D λk Let be the homogeneity between the real-time data mean of the k-th important influencing factor and the historical data mean of the λ-th period. Let be the mean of the real-time data for the k-th important influencing factor. This represents the average of the historical data averages of the k-th important influencing factor over the λ-th period.

[0036] The steps for determining the homogeneity threshold are as follows:

[0037] S41, arbitrarily select θ fixed periods from the historical data, and calculate the mean of the historical data for the μ-th period within the θ fixed periods that is the k-th important influencing factor, denoted as .

[0038] S42, calculate the homogeneity D between the mean of real-time data and the mean of historical data in the μth period using a matching model. μk ;

[0039] S43, using the obtained homogeneity D μkNumerical dataset creation, denoted as, Calculate homogeneity D μk mean in the dataset Among them, the mean The calculation formula is:

[0040]

[0041] In the formula, D μkmin D is the minimum value in the dataset. μkmax The maximum value in the dataset;

[0042] S44, through homogeneity D μk mean in the dataset Determine the range of values ​​for the homogeneity threshold, where the range is:

[0043] Step 5: Using the homogenized parameters of the key influencing factors as input and the corresponding power supply reliability index value calculation result as output, train a neural network prediction model to obtain the predicted value of the power supply reliability index in real-time data. The number of hidden layer nodes in the neural network prediction model is determined according to the following formula:

[0044]

[0045] In the formula, s is the number of hidden layer nodes in the neural network prediction model; m is the number of input layer nodes in the neural network prediction model; n is the number of output layer nodes; and α is an integer from 1 to 9. In this prediction model, n = 1.

[0046] The present invention has the following beneficial effects:

[0047] Compared with existing technologies, the technical solution of this invention obtains historical data related to the power supply reliability of the power grid system, calculates the power supply reliability index value within any fixed period, performs grey relational analysis on the obtained power supply reliability index value and historical data, calculates the correlation degree between the power supply reliability index value and different categories of data, determines the data category correlation order of the influence degree of the power supply reliability index based on the obtained correlation degree value, and sets a correlation degree threshold [Δr] for the influence degree of each power supply reliability index. imin ,1), where Δr iminThe minimum correlation threshold for the i-th power supply reliability index value is used. Data categories within the correlation threshold are selected as important influencing factors of the power supply reliability index based on the obtained data category correlation order. Real-time data of important influencing factors of power supply reliability in the power grid system are collected. The mean values ​​of the important influencing factors in historical data within any fixed period and in real-time data are calculated. The homogeneity between the mean of real-time data and the mean of historical data is calculated using a matching model. A homogeneity threshold range is set, and historical data of the important influencing factors within the homogeneity threshold range are selected as homogenization parameters. The homogenization parameters of the obtained important influencing factors are used as input, and the corresponding power supply reliability index value calculation result is used as output to train and obtain a neural network prediction model. The prediction model outputs the predicted value of the power supply reliability index in real-time data, which can reduce the deviation between the prediction result and the actual value caused by changes in external factors and improve the prediction accuracy of the power supply reliability index value. Attached Figure Description

[0048] Figure 1 This is a flowchart of a method for predicting the hierarchical reliability index of intelligent power supply according to the present invention. Detailed Implementation

[0049] The present invention will be further described below with reference to specific embodiments. The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the present invention. In order to better illustrate the specific embodiments of the present invention, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size.

[0050] Example 1

[0051] like Figure 1 The flowchart shown is a method for predicting the hierarchical reliability index of intelligent power supply according to the present invention.

[0052] The specific implementation process of the technical solution of this invention includes the following steps:

[0053] Step 1) Obtain historical data related to the power supply reliability of the power grid system, and calculate the power supply reliability index value within any fixed period.

[0054] In this embodiment, AIHC-1, the average user outage time index, which is the ratio of the total outage time of all users to the total number of users within a certain period, is used as an example to illustrate the power supply reliability index. Firstly, the value of the power supply reliability index can be calculated using historical data.

[0055]

[0056] In the formula, T1 represents the duration of a power outage for a single user within a certain time period; n1 represents the number of users affected by each power outage.

[0057] Step 2) Perform grey relational analysis on the obtained power supply reliability index values ​​and historical data, calculate the correlation degree between the power supply reliability index values ​​and different categories of data, and determine the data category association order of the influence of the power supply reliability index based on the obtained correlation degree values. The specific steps are as follows:

[0058] S21), construct a data sequence of the average power outage time for users and historical data of different categories, denoted as:

[0059] E = (E1)

[0060] F = (F1, F2, ..., F j )

[0061] In the formula, E represents the data sequence of the average power outage time index value for users, and F represents the data sequence of historical data for each category, including data values ​​from ledger data, operational data, load data, and environmental data. Specifically, F... j For the j-th category data value;

[0062] S22), standardize each data point in the data sequence, using the following formula:

[0063]

[0064]

[0065] In the formula, E 1q e is the standardized value of the q-th value of the average power outage time for users. 1q This is the q-th value representing the average power outage time for users. F represents the average power outage time for users. jq f is the standardized value of the q-th value of the j-th category data. jq For the q-th value of the j-th category, Let m2 be the mean of the data in the j-th category, and m2 and n be the data sample sizes.

[0066] S23), calculate the correlation values ​​of all data points in the data sequence. The calculation formula is:

[0067]

[0068] In the formula, ξ jq Let be the correlation coefficient between each data point in data sequence F and data sequence E, where min1[min q (|E 1q -F jq|)] is used to obtain |E 1q -F jq |Minimum value of the calculated result q Then take the minimum value among all the calculated minimum values ​​(·). i [·],max1[max q (|E 1q -F jq |)] is used to obtain |E iq -F jq The maximum value of the calculated result is max. q Then, take the maximum value among all the calculated maximum values, max1[·], where ρ is the resolution coefficient, and its value range is (0,1], usually ρ=0.5;

[0069] S24), calculate the correlation degree value based on the correlation coefficient calculation result. The calculation formula is:

[0070]

[0071] In the formula, R 1j The correlation between the average power outage time for users and the data of the j-th category;

[0072] S25) The correlation between the obtained average power outage time index and the data of the j-th category is arranged in order of magnitude, that is, the data category correlation order of the influence of the average power outage time index is obtained, and the order of the influence of historical data of which categories on the average power outage time is determined.

[0073] Step 3), set the correlation threshold [Δr] for the influence of the user's average power outage time indicator. 1min ,1), where Δr 1min The minimum correlation threshold for the average power outage time value is used as the basis for selecting data categories within the correlation threshold based on the obtained data category correlation order. These categories are considered as important influencing factors for the average power outage time indicator. The minimum correlation threshold Δr for the average power outage time value is defined as follows: 1min It is determined based on an empirical formula, which is:

[0074]

[0075] In the formula, r 1max r is the maximum correlation value in the correlation order of the user average power outage time index. 1min γ represents the minimum correlation coefficient in the correlation order of the average power outage time index for users, γ represents the number of correlation coefficients in the correlation order of the average power outage time index for users, and ν represents the correlation coefficient value in the correlation order of the average power outage time index for users that is greater than the mean correlation coefficient. The number of.

[0076] Step 4) Collect real-time data on key factors affecting the reliability of the power grid system. It should be noted that the sampling time for the collected real-time data is less than a fixed period. Calculate the mean of key factors in historical data within any fixed period and the mean of real-time data. Calculate the homogeneity between the mean of real-time data and the mean of historical data using a matching model. Set a homogeneity threshold range and select historical data of key factors within the homogeneity threshold range as homogeneity parameters. The expression for the matching model is:

[0077]

[0078] In the formula, D λk Let be the homogeneity between the real-time data mean of the k-th important influencing factor and the historical data mean of the λ-th period. Let be the mean of the real-time data for the k-th important influencing factor. This represents the average of the historical data averages of the k-th important influencing factor over the λ-th period.

[0079] The steps for determining the homogeneity threshold are as follows:

[0080] S41), arbitrarily select θ fixed periods from the historical data, and calculate the mean of the historical data for the μ-th period within the θ fixed periods that is the k-th important influencing factor, denoted as .

[0081] S42), calculate the homogeneity D between the mean of real-time data and the mean of historical data in the μth period using a matching model. μk ;

[0082] S43), using the obtained homogeneity D μk Numerical dataset creation, denoted as, Calculate homogeneity D μk mean in the dataset Among them, the mean The calculation formula is:

[0083]

[0084] In the formula, D μkmin D is the minimum value in the dataset. μkmax The maximum value in the dataset;

[0085] S44), through homogeneity D μk mean in the dataset Determine the range of values ​​for the homogeneity threshold, where the range is:

[0086] Step 5) uses the homogenized parameters of the key influencing factors as input and the corresponding average user outage time index value as output to train a neural network prediction model. The prediction model outputs the predicted value of the average user outage time index in real-time data. The number of hidden layer nodes of the neural network prediction model is determined according to the following formula, where the expression of the formula is:

[0087]

[0088] In the formula, s is the number of hidden layer nodes in the neural network prediction model; m is the number of input layer nodes in the neural network prediction model; n is the number of output layer nodes; and α is an integer from 1 to 9. In this prediction model, n = 1, and the number of input layer nodes is equal to the number of important influencing factors of power supply reliability index.

[0089] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for hierarchical prediction of intelligent power supply reliability indicators, characterized in that, include: Step 1: Obtain historical data related to the power supply reliability of the power grid system and calculate the power supply reliability index value within any fixed period; Step 2: Perform grey relational analysis on the obtained power supply reliability index values ​​and historical data, calculate the correlation degree between the power supply reliability index values ​​and different categories of data, and determine the data category correlation order of the influence degree of the power supply reliability index based on the obtained correlation degree values. Step 3: Set the correlation threshold for the degree of influence of each of the power supply reliability indicators. ,in, For the first The minimum correlation threshold for each power supply reliability index value is used to filter data categories within the correlation threshold as important influencing factors for the power supply reliability index based on the obtained data category correlation order. It is determined based on an empirical formula, which is: In the formula, For the first The maximum correlation value in the correlation order of power supply reliability indicators For the first The minimum correlation degree value in the correlation order of power supply reliability indicators. For the first The number of correlation values ​​in the correlation order of each power supply reliability index. For the first The correlation value in the correlation order of the power supply reliability indicators is greater than the mean correlation value. The number of; Step four: Collect real-time data of key influencing factors on the reliability of the power grid system; calculate the mean of these key influencing factors in historical data over any fixed period and the mean of real-time data; calculate the homogeneity between the mean of real-time data and the mean of historical data using a matching model; set a homogeneity threshold range; and select historical data of the key influencing factors within the homogeneity threshold range as homogeneity parameters; the steps for determining the homogeneity threshold are as follows: S41, Select any historical data Each fixed period is calculated separately in the calculation. Within a fixed period, the first... The first important influencing factor The average of historical data over a period of time, denoted as ; S42, calculate the mean of the real-time data and the first value using a matching model. Homogeneity among the historical data means of each period ; S43, utilizing the obtained homogeneity Numerical dataset creation, denoted as, Calculate homogeneity mean in the dataset , where the mean The calculation formula is: In the formula, The minimum value in the dataset. The maximum value in the dataset; S44, through homogeneity mean in the dataset Determine the range of values ​​for the homogeneity threshold, where the range is: ; Step 5: Using the homogenized parameters of the important influencing factors as input and the corresponding power supply reliability index value calculation result as output, train the neural network prediction model to obtain the predicted value of the power supply reliability index in real time.

2. The intelligent power supply reliability index classification prediction method according to claim 1, characterized in that, In step four, the expression for the matching model is: In the formula, For the first The mean of real-time data for the first important influencing factor and the second Homogeneity among the historical data means of each period For the first The average of real-time data for several important influencing factors. For the first The first important influencing factor The average of historical data over a period of time.

3. The intelligent power supply reliability index classification prediction method according to claim 1, characterized in that, The specific steps for step two are as follows: S21, construct a data sequence of power supply reliability index values ​​and historical data of different categories, denoted as: In the formula, This is a data sequence of power supply reliability index values. For the first One power supply reliability indicator For the historical data sequences of each category, For the first Data values ​​for each category; S22, standardizes each data point in the data sequence, using the following formula: In the formula, For the first The first power supply reliability index Standardized values ​​of a number of values. For the first The first power supply reliability index A number, For the first The average of several power supply reliability indicators, For the first The first category of data Standardized values ​​of a number of values. For the first The first category of data A number, For the first The mean of the data for each category All of these are data sample sizes; S23, Calculate the correlation values ​​of all data points in the data sequence. The calculation formula is: In the formula, For data sequences Data points and data sequences The correlation coefficient, where, To obtain Minimum value of the calculation result Then take the minimum value among all the calculated minimum values. , To obtain Maximum value of the calculation result Then take the maximum value among all the calculated maximum values. , The resolution coefficient has a range of values. ,Pick ; S24. Calculate the correlation degree value based on the correlation coefficient calculation results. The calculation formula is as follows: In the formula, For the first The power supply reliability index and the first The correlation between data in each category; S25, the obtained first The power supply reliability index and the first The correlation degree of each category of data is arranged in order of magnitude, thus obtaining the data category correlation order of the degree of influence of the power supply reliability index.

4. The intelligent power supply reliability index classification prediction method according to claim 1, characterized in that, In step five, the number of hidden layer nodes in the neural network prediction model is determined according to the following formula, wherein the expression of the formula is: In the formula, This represents the number of hidden layer nodes in the neural network prediction model. This represents the number of nodes in the input layer of the neural network prediction model. This represents the number of nodes in the output layer. It is an integer from 1 to 9.

5. The intelligent power supply reliability index classification prediction method according to claim 1, characterized in that, The number of nodes in the input layer of the neural network prediction model is equal to the number of important influencing factors of power supply reliability index.

6. The intelligent power supply reliability index classification prediction method according to claim 1, characterized in that, In step four, the sampling time step of the real-time data is less than a fixed period.

Images