A Power Grid Security Situation Analysis Method for Big Data Analysis

Abstract

The present invention relates to the technical field of power grid security situation, and discloses a power grid security situation analysis method based on big data analysis. The method includes: preprocessing and standardizing a power grid historical data sequence; performing periodic prediction on the standardized power grid historical data sequence; using a power grid equipment operation life evaluation model to receive the standardized power grid historical data sequence and perform power grid equipment operation life evaluation; and performing power grid security situation prediction analysis based on the remaining operation life of the power grid equipment and the standardized power grid prediction data sequence. The present invention adopts a time-frequency conversion method to convert the standardized power grid historical data sequence into the frequency domain, extracts periodic information and performs data prediction, improves the interpretability of the prediction result, constructs a likelihood function representing the life distribution of power grid equipment, and realizes power grid security situation analysis based on multi-dimensional historical data of power grid equipment based on the standardized power grid prediction data and the remaining operation life of the power grid equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid security situation, and in particular to a power grid security situation analysis method based on big data analysis. Background Art

[0002] With the continuous expansion of the scale of power systems, the profound adjustment of energy structures and the rapid development of information technology, the operation of power grids faces unprecedented challenges. Especially in the context of global energy transformation, the proportion of clean energy continues to increase, and the large-scale access of renewable energy (such as wind energy and solar energy) makes the load fluctuation of power grids more complicated, increasing the difficulty of power grid load prediction and security situation analysis. The safety of power grid load is directly related to the stable operation of the power grid and the reliability of power supply. How to effectively analyze the load security situation has become one of the key issues that need to be solved in the current power industry. Traditional power grid load security analysis methods mostly rely on static mathematical models and statistical analysis based on historical data. Although these methods can analyze power grid loads to a certain extent, with the expansion of power grid scale and the dynamic changes of renewable energy, traditional methods require high-frequency power grid adjustments to cope with the impact of potential power grid load fluctuations, but high-frequency power grid adjustments will greatly reduce the life of power grid loads. Summary of the invention

[0003] In view of this, the present invention provides a power grid security situation analysis method based on big data analysis. Based on the historical operation status of large-scale power grid equipment, the periodic characteristics of the operation of power grid equipment are extracted, and periodic prediction and remaining operating life assessment of power grid equipment are performed. The power grid security situation assessment is achieved while ensuring the load life of the power grid, which helps to improve the safety and reliability of the power grid.

[0004] To achieve the above purpose, the present invention provides a power grid security situation analysis method based on big data analysis, comprising the following steps:

[0005] S1: collecting a historical data sequence of a power grid, and performing preprocessing and standardization on the historical data sequence of the power grid to obtain a standardized historical data sequence of the power grid, wherein the preprocessing includes interpolation and data denoising, and the historical data sequence of the power grid is a historical data sequence of power grid indicators, and the power grid indicators include voltage, current, power, frequency, temperature and equipment fault records of power grid equipment;

[0006] S2: Perform periodic forecasting on the standardized power grid historical data series to obtain the standardized power grid forecast data series.

[0007] The standardized power grid prediction data sequence is standardized power grid prediction data at multiple prediction moments;

[0008] S3: Use the power grid equipment operation life evaluation model to receive the standardized power grid historical data sequence, conduct the operation life evaluation of the power grid equipment, and obtain the remaining operation life of the power grid equipment;

[0009] S4: Generate the power grid security situation prediction and analysis result based on the remaining operation life of the power grid equipment and the standardized power grid prediction data sequence. The power grid security situation prediction and analysis result is the power grid security situation at multiple prediction times.

[0010] As a further improved method of the present invention:

[0011] Optionally, the power grid equipment is various equipment used for constructing, operating, and maintaining the power system, including power generation equipment, power transmission equipment, power transformation equipment, power distribution equipment, and power consumption equipment. The power grid historical data sequence is x:

[0012] x = (x(t - N), x(t - N + 1),..., x(t - n),..., x(t - 1)), n ∈ [1, N]

[0013] x(t - n) = {x i (t - n)|i ∈ [1, 6]}

[0014] Where:

[0015] x(t - N), x(t - N + 1),..., x(t - n),..., x(t - 1) represent the power grid data at N historical times, x(t - n) represents the power grid data at the nth historical time, t represents the timing information parameter, t - n represents the timing information of the power grid data, and the time interval between adjacent historical times is The time interval between the t - n and the current time is

[0016] x i (t - n) represents the data value of the ith power grid index in the power grid data x(t - n), i ∈ [1, 6]. The 1st - 6th power grid indexes are the voltage, current, power, frequency, temperature of the power grid equipment, and the equipment failure record in sequence;

[0017] Pre - process and standardize the power grid historical data sequence.

[0018] Optionally, the pre - processing process of the power grid data includes:

[0019] Obtain the timing information of the missing power grid data, calculate the distance between the timing information of the missing power grid data and the timing information of the non - missing power grid data, select M pieces of timing information of the non - missing power grid data with the closest distances, and extract the power grid data corresponding to the selected timing information;

[0020] Calculate the time-series information weight of the extracted power grid data, weight the extracted power grid data, and use the weighted result as the interpolation completion result of the power grid data lacking time-series information of the power grid data. Use the power grid historical data sequence without missing power grid data as the interpolated and completed power grid historical data sequence. The calculation formula for the interpolation completion result of the power grid data is as follows:

[0021]

[0022] Where:

[0023] y represents the interpolation completion result of the power grid data lacking time-series information of the power grid data, and x m represents the m-th extracted power grid data, and Inf m represents the time-series completion parameter of the power grid data x m . represents the time-series information weight of the power grid data x m , where m ∈ [1, M];

[0024] time(x m ) represents the time-series information of the power grid data x m , time represents the time-series information lacking power grid data, std represents the standard deviation of the distance between the time-series information of the M extracted power grid data and the time-series information lacking power grid data, and ε represents the standard deviation control coefficient;

[0025] Perform data denoising on the interpolated and completed power grid historical data sequence to obtain the preprocessed power grid historical data sequence. The data denoising method is median filtering.

[0026] Optionally, perform standardization processing on the power grid data in the preprocessed power grid historical data sequence to obtain the standardized power grid data. Sort the standardized power grid data in the order of time-series information to obtain the standardized power grid historical data sequence. The standardization process is as follows:

[0027] Obtain the power grid data in the preprocessed power grid historical data sequence, and perform standardization processing on the data values of different power grid indicators in the obtained power grid data. The standardization processing method for the data values under the voltage, current, power, frequency, and temperature indicators is normalization processing. The normalization processing method is the min-max normalization algorithm based on the preset maximum value and preset minimum value of the power grid indicator. The standardization processing method for the equipment failure record indicator adopts the {0, 1} coding method. If there is an equipment failure record in the power grid data, the standardization processing result of the equipment failure record indicator is 1, otherwise the standardization processing result of the equipment failure record indicator is 0;

[0028] The standardized power grid historical data sequence is:

[0029]

[0030] Wherein:

[0031] represents the normalized result of the grid data x(t - n), represents the data value x i (t - n) of the normalized result.

[0032] Optionally, perform time - frequency conversion on the normalized grid historical data sequence to obtain the frequency - amplitude distribution of the normalized grid historical data sequence under different grid metrics. The time - frequency conversion process is as follows:

[0033] Split the normalized grid historical data sequence into normalized index data value sequences under different grid metrics. The normalized index data value sequence under the i - th grid metric is

[0034]

[0035] Perform frequency - domain transformation on the normalized index data value sequence to obtain the frequency - domain representation result of the time - series sequence. The frequency - domain representation result of the normalized index data value sequence is:

[0036]

[0037] Wherein:

[0038] Y i s represents the frequency - domain representation result of the normalized index data value sequence at the s - th frequency component. j represents the imaginary unit;

[0039] Based on the frequency - domain representation result, calculate the amplitude of the normalized index data value sequence at the s - th frequency component as the frequency - amplitude distribution of the normalized index data value sequence. The amplitude of the normalized index data value sequence at the s - th frequency component is

[0040]

[0041] Wherein:

[0042] Re(Y i s ) represents the real part of the frequency - domain representation result Y i s Im(Y i s) represents the frequency-domain representation result Y i s The imaginary part of;

[0043] Extract the periodic information of the standardized index data value sequence from the frequency amplitude distribution, and the standardized index data value sequence The periodic information extraction process is as follows:

[0044] Obtain the standardized index data value sequence The frequency amplitude distribution of, extract the order of the frequency components with amplitudes higher than the preset amplitude threshold, where the order of the s-th frequency component is s;

[0045] Convert the extracted order into the expected period length in turn, and select the shortest expected period length as the periodic information L of the standardized index data value sequence where the expected period length corresponding to the order s is i Enhance the periodic characteristics of the standardized index data value sequence based on the periodic information of the grid index.

[0046] Optionally, the periodic characteristic enhancement method is:

[0047] Obtain the standardized index data value sequence and the periodic information of the standardized index data value sequence, slice the standardized index data value sequence using the periodic information, and stack the sliced results to obtain the sliced matrix of the standardized index data value sequence as the periodic characteristic enhancement result;

[0048] Construct a grid data prediction model to perform periodic prediction on the standardized grid historical data sequence. The grid data prediction model is a Transformer structure, including an encoder and a decoder. The encoder is used to receive the sliced matrix of the standardized index data value sequence, perform convolution processing and activation function processing on the sliced matrix, and perform residual connection on the processing result and the sliced matrix to obtain the encoded representation result of the sliced matrix. The decoder is used to perform periodic cross-attention calculation on the encoded representation result of the sliced matrix to obtain the periodic attention values of the standardized index data value sequence at different prediction times;

[0049] Generate the standardized index prediction data values of the standardized index data value sequence at different prediction times based on the periodic attention values, and form the generated results into the standardized grid prediction data sequence. The standardized index data value sequence

[0050] The standardized index prediction data value at the h-th prediction time t+h is x (t+h): * (t+h):

[0051]

[0052] Wherein:

[0053] is the standardized index data value sequence is the periodic attention value at the h-th prediction time t+h;

[0054] softmax(·) is the softmax activation function, G represents the value matrix, and H+1 is the total number of prediction times;

[0055] The standardized power grid prediction data sequence is:

[0056] x * = (x * (t), x * (t+1),..., x * (t+h),..., x * (t+H))

[0057] x * (t+h) = {x * (t+h)|i∈[1,6]}

[0058] Wherein:

[0059] x * (t+h) represents the standardized power grid prediction data at the h-th prediction time t+h.

[0060] Optionally, the power grid equipment operation life evaluation model includes a life distribution parameter evaluation module and a power grid equipment operation life calculation module. The life distribution parameter evaluation module is used to receive the standardized power grid historical data sequence, construct a likelihood function of the life distribution, solve the likelihood function to obtain the life distribution parameters, and the power grid equipment operation life calculation module is used to receive the life distribution parameters and calculate the remaining operation life of the power grid equipment by using the life distribution parameters and the standardized index data value sequence of the equipment failure record index;

[0061] The likelihood function of the life distribution constructed by the life distribution parameter evaluation module is F(λ1,λ2):

[0062]

[0063] Wherein:

[0064] λ1,λ2 represent the life distribution parameters, λ1 represents the life ratio parameter, and λ2 represents the life shape parameter;

[0065] Take the logarithm of the likelihood function \(F(\lambda_1,\lambda_2)\) to obtain the log-likelihood function, solve the first-order partial derivatives of the log-likelihood function, set the first-order partial derivatives to 0 to obtain the equation of the life ratio parameter and the life shape parameter, and solve the equation to obtain the life ratio parameter and the life shape parameter;

[0066] The calculation formula for the remaining operating life of the power grid equipment obtained by using the standardized index data value sequence of the life distribution parameter and the equipment failure record index is:

[0067]

[0068] Where:

[0069] \(V_1\) represents the original operating life of the power grid equipment, \(V\) represents the remaining operating life of the power grid equipment, and \(\alpha\) represents the life control parameter.

[0070] Optionally, the analysis process of the power grid security situation prediction and analysis result is as follows:

[0071] Extract the standardized power grid prediction data at different prediction times and the remaining operating life of the power grid equipment, and calculate the security situation of the power grid equipment at the prediction time. The security situation of the power grid equipment at the \(h\)th prediction time is:

[0072]

[0073] Where:

[0074] \(P(t + h)\) represents the security situation of the power grid equipment at the \(h\)th prediction time, \(\rho\) represents the preset standardized power grid data of the normally operating power grid equipment, \(\beta\) represents the scale control parameter, \(\|\cdot\|_1\) represents the \(L1\) norm, and \(V\) t+h represents the retention ratio of the operating life of the power grid equipment at the \(h\)th prediction time;

[0075] The higher the security situation, the safer the power grid equipment. The mean value of the security situations of all power grid equipment at the prediction time is used as the power grid security situation at the prediction time.

[0076] To solve the above problems, the present invention provides an electronic device, which includes:

[0077] A memory that stores at least one instruction;

[0078] A communication interface that enables communication of the electronic device; and

[0079] A processor that executes the instructions stored in the memory to implement the above-mentioned power grid security situation analysis method for big data analysis.

[0080] To solve the above problems, the present invention further provides a computer-readable storage medium, in which at least one instruction is stored, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned power grid security situation analysis method for big data analysis.

[0081] Compared with the prior art, the present invention proposes a power grid security situation analysis method for big data analysis, and this technology has the following advantages:

[0082] First of all, this solution proposes a data prediction method. By using an interpolation completion method that combines neighbor information, the missing power grid data in the power grid historical data sequence is completed. During the completion process, based on the temporal information distance between neighbor information, a temporal information weight is generated to improve the contribution rate of interpolation completion of temporal neighbor data and the effectiveness of the completion result. The standardized power grid historical data sequence is transformed into the frequency domain by using a time-frequency conversion method to obtain the frequency domain representation result and the frequency amplitude distribution, and the periodic information characterizing the cycle length of the standardized power grid historical data sequence is extracted. Through the periodic information, it is possible to avoid misidentifying periodic fluctuations as noise, thereby reducing the prediction error. The periodic information provides clear prior knowledge, reducing the complex patterns that the model needs to learn from the data, thus reducing the model complexity. The coding and decoding method is used to combine the periodic information for prediction, improving the interpretability of the prediction result.

[0083] At the same time, this solution proposes a power grid equipment life assessment method and a security situation assessment method. A likelihood function representing the life distribution of power grid equipment is constructed, and the life distribution parameters in the likelihood function are solved by combining the standardized power grid historical data sequence to obtain the life distribution parameters characterizing the life distribution length, peak position, and tail behavior. The remaining operating life of the power grid equipment is calculated by combining the standardized index data value sequence of the equipment failure record index. The standardized power grid prediction data at different prediction times and the remaining operating life of the power grid equipment are extracted, and the security situation of the power grid equipment at the prediction time is calculated. The lower the difference between the standardized power grid prediction data and the preset standardized power grid data, and the higher the retention ratio of the operating life of the power grid equipment at the prediction time, the higher the security situation of the power grid equipment, and thus the higher the power grid security situation, realizing the power grid security situation analysis based on multi-dimensional historical data of power grid equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 It is a schematic flowchart of a power grid security situation analysis method for big data analysis provided by an embodiment of the present invention.

[0085] The realization, functional characteristics, and advantages of the purpose of the present invention will be further described in conjunction with the embodiments with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0086] It should be understood that the specific embodiments described herein are only for explaining the present invention and are not used to limit the present invention.

[0087] An embodiment of the present application provides a power grid security situation analysis method for big data analysis. The execution subject of the power grid security situation analysis method for big data analysis includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the power grid security situation analysis method for big data analysis can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc.

[0088] Referring to Figure 1 , Embodiment 1 of the present invention is:

[0089] A power grid security situation analysis method for big data analysis includes the following steps:

[0090] S1: Collect the power grid historical data sequence, and perform preprocessing and standardization processing on the power grid historical data sequence to obtain a standardized power grid historical data sequence.

[0091] The power grid equipment is various equipment used to construct, operate, and maintain the power system, including power generation equipment, power transmission equipment, power transformation equipment, power distribution equipment, and power consumption equipment. The power grid historical data sequence is x:

[0092] x = (x(t - N), x(t - N + 1),..., x(t - n),..., x(t - 1)), n ∈ [1, N]

[0093] x(t - n) = {x i (t - n)|i ∈ [1, 6]}

[0094] Wherein:

[0095] x(t - N), x(t - N + 1),..., x(t - n),..., x(t - 1) represent the power grid data at N historical moments, x(t - n) represents the power grid data at the nth historical moment, t represents the time series information parameter, t - n represents the time series information of the power grid data, and the time interval between adjacent historical moments is The time interval between the t - n and the current moment is

[0096] x i(t - n) represents the data value of the i-th power grid index in the power grid data x(t - n), where i ∈ [1, 6], and the 1st - 6th power grid indexes are the voltage, current, power, frequency, temperature of power grid equipment, and equipment fault records in sequence;

[0097] Perform preprocessing and standardization processing on the power grid historical data sequence.

[0098] The preprocessing process of the power grid data includes:

[0099] Obtain the timing information of the missing power grid data, calculate the distance between the timing information of the missing power grid data and the timing information of the non - missing power grid data, select M pieces of timing information of the non - missing power grid data with the closest distance, and extract the power grid data corresponding to the selected timing information; the distance between the timing information is the absolute value of the difference between the timing information.

[0100] Calculate the timing information weight of the extracted power grid data, weight the extracted power grid data, and use the weighted result as the power grid data interpolation and completion result of the timing information of the missing power grid data. Use the power grid historical data sequence of the non - missing power grid data as the interpolated and completed power grid historical data sequence. The calculation formula for the power grid data interpolation and completion result is:

[0101]

[0102] Where:

[0103] y represents the power grid data interpolation and completion result of the timing information of the missing power grid data, x m represents the m - th extracted power grid data, Inf m represents the power grid data x m 's timing completion parameter, represents the timing information weight of the power grid data x m , where m ∈ [1, M];

[0104] time(x m ) represents the timing information of the power grid data x m , time represents the timing information of the missing power grid data, std represents the standard deviation of the distance between the timing information of the M extracted power grid data and the timing information of the missing power grid data, and ε represents the standard deviation control coefficient;

[0105] Perform data denoising processing on the interpolated and completed power grid historical data sequence to obtain the preprocessed power grid historical data sequence. The data denoising method is median filtering.

[0106] Perform standardization processing on the power grid data in the preprocessed historical power grid data sequence to obtain the standardized power grid data, and sort the standardized power grid data in the order of time series information to obtain the standardized historical power grid data sequence. The standardization process is as follows:

[0107] Obtain the power grid data in the preprocessed historical power grid data sequence, and perform standardization processing on the data values of different power grid indicators in the obtained power grid data. The standardization processing method for the data values under the voltage, current, power, frequency, and temperature indicators is normalization processing. The normalization processing method is the min-max normalization algorithm based on the preset maximum value and preset minimum value of the power grid indicator. The standardization processing method for the equipment failure record indicator uses {0,1} coding. If there is an equipment failure record in the power grid data, the standardization processing result of the equipment failure record indicator is 1; otherwise, the standardization processing result of the equipment failure record indicator is 0;

[0108] The standardized historical power grid data sequence is:

[0109]

[0110] Where:

[0111] represents the standardization processing result of the power grid data x(t - n), represents the data value x i (t - n) of the standardization processing result.

[0112] S2: Perform periodic prediction on the standardized historical power grid data sequence to obtain the standardized power grid prediction data sequence.

[0113] Perform time-frequency conversion on the standardized historical power grid data sequence to obtain the frequency amplitude distribution of the standardized historical power grid data sequence under different power grid indicators. The time-frequency conversion process is as follows:

[0114] Split the standardized historical power grid data sequence into standardized index data value sequences under different power grid indicators. The standardized index data value sequence under the i-th power grid indicator is

[0115]

[0116] Perform frequency domain transformation on the standardized index data value sequence to obtain the frequency domain representation result of the time series sequence. The frequency domain representation result of the standardized index data value sequence is:

[0117]

[0118] Where:

[0119] Y i s represents the sequence of standardized index data values in the frequency-domain representation result of the sth frequency component, where j represents the imaginary unit;

[0120] Based on the frequency-domain representation result, calculate the amplitude of the sequence of standardized index data values at the sth frequency component, as the frequency amplitude distribution of the sequence of standardized index data values, and the amplitude of the sequence of standardized index data values at the sth frequency component is

[0121]

[0122] where:

[0123] Re(Y i s ) represents the real part of the frequency-domain representation result Y i s and Im(Y i s ) represents the imaginary part of the frequency-domain representation result Y i s ;

[0124] Extract the periodic information of the sequence of standardized index data values from the frequency amplitude distribution, and the periodic information extraction process of the sequence of standardized index data values is as follows:

[0125] Obtain the frequency amplitude distribution of the sequence of standardized index data values and extract the order of the frequency components whose amplitudes are higher than the preset amplitude threshold. The order of the sth frequency component is s;

[0126] Convert the extracted orders into expected cycle lengths in turn, and select the shortest expected cycle length as the periodic information L of the sequence of standardized index data values, where the expected cycle length corresponding to the order s is i

[0127] Enhance the periodic characteristics of the sequence of standardized index data values based on the periodic information of the grid index.

[0128]

[0128] The periodic characteristic enhancement method is:

[0129] Obtain the standardized index data value sequence and the periodic information of the standardized index data value sequence, slice the standardized index data value sequence using the periodic information, and stack the slicing results to obtain the slicing matrix of the standardized index data value sequence as the periodic feature enhancement result;

[0130] As an embodiment of the present invention, the number of rows of the slicing matrix is the number of slices, the number of columns is the periodic information, and the standardized index data value sequence The number of slices is

[0131] Construct a power grid data prediction model to perform periodic prediction on the standardized power grid historical data sequence. The power grid data prediction model is of the Transformer structure, including an encoder and a decoder. The encoder is used to receive the slicing matrix of the standardized index data value sequence, perform convolutional processing and activation function processing on the slicing matrix, and perform residual connection on the processing result and the slicing matrix to obtain the encoded representation result of the slicing matrix. The decoder is used to perform periodic cross-attention calculation on the encoded representation result of the slicing matrix to obtain the periodic attention values of the standardized index data value sequence at different prediction times. The standardized index data value sequence The periodic attention value at the h-th prediction time t+h is:

[0132]

[0133] Where:

[0134] U i Represents the encoded representation result of the slicing matrix of the standardized index data value sequence, T represents transpose, Represents the query matrix applicable to the h-th prediction time in the standardized index data value sequence In, Represents the key matrix applicable to the h-th prediction time in the standardized index data value sequence In; As a preferred embodiment of the present invention, The number of rows of is the periodic information of the standardized index data value sequence In, The N+1+h mod L-th column in the matrix i Is a larger value, which can improve the capture ability of the periodic features related to the h-th prediction time, where mod is the remainder. If N+1+h mod L i Is 0, then N+1+h mod L i Is assigned the value of L i ;

[0135] Generate the standardized index prediction data values of the standardized index data value sequence at different prediction times based on the periodic attention values, and form the generated results into the standardized power grid prediction data sequence, the standardized index data value sequence The standardized index prediction data value at the h-th prediction time t+h is x * (t+h):

[0136]

[0137] Where:

[0138] softmax(·) is the softmax activation function, G represents the value matrix, and H+1 is the total number of prediction times;

[0139] The standardized power grid prediction data sequence is:

[0140] x * =(x * (t), x * (t+1),..., x * (t+h),..., x * (t+H))

[0141] x * (t+h)={x * (t+h)|i∈[1,6]}

[0142] Where:

[0143] x * (t+h) represents the standardized power grid prediction data at the h-th prediction time t+h.

[0144] S3: Use the power grid equipment operation life evaluation model to receive the standardized power grid historical data sequence, perform the power grid equipment operation life evaluation, and obtain the remaining operation life of the power grid equipment.

[0145] The power grid equipment operation life evaluation model includes a life distribution parameter evaluation module and a power grid equipment operation life calculation module. The life distribution parameter evaluation module is used to receive the standardized power grid historical data sequence, construct the likelihood function of the life distribution, solve the likelihood function, and obtain the life distribution parameters. The power grid equipment operation life calculation module is used to receive the life distribution parameters and calculate the remaining operation life of the power grid equipment by using the life distribution parameters and the standardized index data value sequence of the equipment failure record index;

[0146] The likelihood function of the life distribution constructed by the life distribution parameter evaluation module is F(λ1,λ2):

[0147]

[0148] Wherein:

[0149] λ1 and λ2 represent life distribution parameters, λ1 represents the life ratio parameter, and λ2 represents the life shape parameter; the life ratio parameter describes the stretching degree of the life distribution of power grid equipment, that is, the length of the life distribution, and the life shape parameter describes the peak position and tail behavior of the life distribution of power grid equipment; when the life shape parameter is less than 1, it means that power grid equipment will have failures in the early stage of use, when the life shape parameter is equal to 1, it means that power grid equipment will have failures in a random time period, and when the life shape parameter is greater than 1, it means that power grid equipment will have aging failures as the usage duration increases;

[0150] Take the logarithm of the likelihood function F(λ1, λ2) to obtain the log-likelihood function, and solve the first-order partial derivative of the log-likelihood function. Let the first-order partial derivative be 0 to obtain the equation of the life ratio parameter and the life shape parameter, and solve the equation to obtain the life ratio parameter and the life shape parameter;

[0151] The calculation formula for the remaining operating life of power grid equipment calculated by using the standardized index data value sequence of the life distribution parameter and the equipment failure record index is:

[0152]

[0153] Wherein:

[0154] V1 represents the original operating life of power grid equipment, V represents the remaining operating life of power grid equipment, and α represents the life control parameter.

[0155] S4: Generate a power grid security situation prediction and analysis result based on the remaining operating life of power grid equipment and the standardized power grid prediction data sequence.

[0156] The analysis process of the power grid security situation prediction and analysis result is as follows:

[0157] Extract the standardized power grid prediction data and the remaining operating life of power grid equipment at different prediction times, and calculate the security situation of power grid equipment at the prediction time. The security situation of power grid equipment at the hth prediction time is:

[0158]

[0159] Wherein:

[0160] P(t + h) represents the security situation of power grid equipment at the hth prediction time, ρ represents the preset standardized power grid data of normally operating power grid equipment, β represents the scale control parameter, ||·||1 represents the L1 norm, and V t+h represents the retention ratio of the operating life of power grid equipment at the hth prediction time;

[0161] The higher the security situation, the safer the power grid equipment. The average value of the security situations of all power grid equipment at the prediction moment is used as the power grid security situation of the power grid at the prediction moment.

[0162] Embodiment 2:

[0163] This solution conducts a comparative experiment on the power grid security situation analysis method of big data analysis, the ARIMA+SVM method, the LSTM+SVM method, the ARIMA+random forest method, and the LSTM+random forest method. ARIMA and LSTM are used for prediction, and SVM and the random forest method are used for security situation assessment. The results of the comparative experiment are shown in Table 1:

[0164] Table 1

[0165]

[0166] As shown in Table 1, the recall rate of the security situation assessment divides the test samples into positive classes and negative classes, and evaluates the recall rates in the positive classes and negative classes. The prediction accuracy of the LSTM method is higher than that of the ARIMA method and is close to the prediction method in the power grid security situation analysis method of big data analysis. The classification accuracy of the SVM method is higher than that of the random forest method. Both are weaker than the power grid security situation analysis method of big data analysis in numerical evaluation and are close to the power grid security situation analysis method of big data analysis in the classification effects of positive classes and negative classes.

[0167] It should be understood that the above embodiments are only for illustration and are not limited by this structure in the scope of the patent application.

[0168] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the term "including" or "comprising" or any other variant thereof in this article is intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements not only includes those elements but also includes other elements not expressly listed, or also includes elements inherent in such a process, device, article or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, device, article or method including the element.

[0169] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described example methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above and includes several instructions for causing a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0170] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A power grid security situation analysis method for big data analysis, characterized in that The method includes: S1: Collect the historical data sequence of the power grid, and perform preprocessing and standardization processing on the historical data sequence of the power grid to obtain a standardized historical data sequence of the power grid. The preprocessing includes interpolation completion and data denoising. The historical data sequence of the power grid is the historical data sequence of power grid indicators, and the power grid indicators include the voltage, current, power, frequency, temperature of power grid equipment, and equipment fault records; S2: Perform periodic prediction on the standardized historical data sequence of the power grid to obtain a standardized predicted data sequence of the power grid. The standardized predicted data sequence of the power grid is the standardized predicted power grid data at multiple prediction times; Perform time-frequency conversion on the standardized historical power grid data sequence to obtain the frequency-amplitude distribution of the standardized historical power grid data sequence under different power grid indicators, and extract the periodic information of the standardized indicator data value sequence from the frequency-amplitude distribution. The standardized indicator data value sequence The process for extracting the periodic information is as follows: Obtain the sequence of the standardized index data values of the frequency amplitude distribution, and extract the order of the frequency components whose amplitudes are higher than the preset amplitude threshold, where the order of the s-th frequency component is s; Convert the extracted order into the expected cycle lengths in sequence, and select the shortest expected cycle length among them as the standardized index data value sequence of the periodic information , where the expected cycle length corresponding to the order s is ; Generate the standardized index predicted data values of the standardized index data value sequence at different prediction times based on the periodic attention values, and form the generated results into the standardized predicted data sequence of the power grid; S3: Use the power grid equipment operation life assessment model to receive the standardized historical data sequence of the power grid, and perform power grid equipment operation life assessment to obtain the remaining operation life of the power grid equipment; S4: Generate a power grid security situation prediction and analysis result based on the remaining operation life of the power grid equipment and the standardized predicted data sequence of the power grid. The power grid security situation prediction and analysis result is the power grid security situation at multiple prediction times; The analysis process of the power grid security situation prediction and analysis result is: Extract the standardized predicted power grid data and the remaining operation life of the power grid equipment at different prediction times, and calculate the security situation of the power grid equipment at the prediction time. The security situation of the power grid equipment at the h-th prediction time is: ； ； Where: Indicates the security situation of the power grid equipment at the h-th prediction moment, Represents the preset standardized power grid data of the power grid equipment operating normally, Indicates the scale control parameter, Indicates the L1 norm, Indicates the retention ratio of the operating life of the power grid equipment at the h-th prediction moment; Indicates the life distribution parameter, Indicates the life ratio parameter, Indicates the life shape parameter; Indicates the original operating life of the power grid equipment, Is the standardized power grid prediction data sequence; The higher the security situation, the safer the power grid equipment. The average value of the security situations of all power grid equipment at the prediction time is used as the power grid security situation at the prediction time; Perform periodic feature enhancement on the standardized index data value sequence based on the periodic information of the power grid indicators; The periodic feature enhancement method is: Obtain the standardized index data value sequence and the periodic information of the standardized index data value sequence, use the periodic information to slice the standardized index data value sequence, and stack the sliced results to obtain the sliced matrix of the standardized index data value sequence as the periodic feature enhancement result; Construct a power grid data prediction model to perform periodic prediction on the standardized historical data sequence of the power grid. The power grid data prediction model is a Transformer structure, including an encoder and a decoder. The encoder is used to receive the sliced matrix of the standardized index data value sequence, perform convolution processing and activation function processing on the sliced matrix, and perform residual connection on the processing result and the sliced matrix to obtain the encoded representation result of the sliced matrix. The decoder is used to perform periodic cross-attention calculation on the encoded representation result of the sliced matrix to obtain the periodic attention values of the standardized index data value sequence at different prediction times.

2. The power grid security situation analysis method for big data analysis according to claim 1, wherein, The grid equipment refers to various equipment used for constructing, operating, and maintaining power systems, including power generation equipment, power transmission equipment, power transformation equipment, power distribution equipment, and power consumption equipment. The grid historical data sequence is : ； ； Where: represents the power grid data at N historical moments, represents the power grid data at the nth historical moment, and t represents the time series information parameter, represents the time series information of the power grid data, and the time interval between adjacent historical moments is , the time interval from the current moment is ; Represents power grid data The data value of the i-th power grid index in , and the 1st to 6th power grid indexes are the voltage, current, power, frequency, temperature of the power grid equipment, and the equipment fault record in sequence; Perform preprocessing and standardization processing on the historical data sequence of the power grid.

3. The power grid security situation analysis method for big data analysis according to claim 2, characterized in that The preprocessing process of the power grid data includes: Obtain the time-series information lacking grid data, calculate the distance between the time-series information lacking grid data and the time-series information not lacking grid data, select M time-series information of the nearest non-lacking grid data, and extract the grid data corresponding to the selected time-series information; Calculate the time-series information weights of the extracted grid data, weight the extracted grid data, and use the weighted result as the grid data interpolation and completion result of the time-series information lacking grid data. Use the grid historical data sequence not lacking grid data as the interpolated and completed grid historical data sequence. The calculation formula for the grid data interpolation and completion result is: ； ； Where: Represents the interpolation and completion result of power grid data indicating the lack of timing information of power grid data, Represents the m-th extracted power grid data, Represents power grid data The timing completion parameter of, Represents power grid data The timing information weight of, ; Represents power grid data of the timing information, Represents the timing information of the missing power grid data, Represents the standard deviation of the distance between the timing information of the M extracted power grid data and the timing information of the missing power grid data, Represents the standard deviation control coefficient; Perform data denoising processing on the interpolated and completed grid historical data sequence to obtain the preprocessed grid historical data sequence. The data denoising method is median filtering.

4. The method for analyzing the power grid security situation through big data analysis according to claim 3, characterized in that, Perform standardization processing on the grid data in the preprocessed grid historical data sequence to obtain the standardized grid data. Sort the standardized grid data in the order of time-series information to obtain the standardized grid historical data sequence. The standardization processing process is: Obtain the grid data in the preprocessed grid historical data sequence, and perform standardization processing on the data values of different grid indicators in the obtained grid data. The standardization processing method for the data values under the voltage, current, power, frequency, and temperature indicators is normalization processing. The normalization processing method is the min-max normalization algorithm based on the preset maximum value and preset minimum value of the grid indicator. The standardization processing method for the equipment failure record indicator uses {0,1} coding. If there is an equipment failure record in the grid data, the standardization processing result of the equipment failure record indicator is 1, otherwise the standardization processing result of the equipment failure record indicator is 0; The standardized historical power grid data sequence is :[[-END]] ； ； Where: Represents the standardized processing result of power grid data , Represents the standardized processing result of the data value .

5. The method for analyzing the power grid security situation through big data analysis according to claim 4, characterized in that, The time-frequency conversion process is: Split the standardized grid historical data sequence into standardized index data value sequences under different grid indexes, where the standardized index data value sequence under the \(i\)-th grid index is : ； Perform a frequency-domain transformation on the standardized index data value sequence to obtain the frequency-domain representation result of the time series sequence, where the standardized index data value sequence The frequency-domain representation result is as follows: ； Where: representing the sequence of the standardized index data values the frequency-domain representation result of the sth frequency component, where j represents the imaginary unit; Based on the frequency-domain representation result, the amplitude values of the standardized index data value sequence at frequency components are calculated as the frequency amplitude distribution of the standardized index data value sequence. The amplitude of the standardized index data value sequence at the sth frequency component is : ； Where: represents the frequency-domain representation result the real part of represents the frequency-domain representation result the imaginary part of 6. The power grid security situation analysis method for big data analysis according to claim 5, characterized in that, The standardized index data value sequence The predicted data value of the standardized index at the h-th prediction moment t+h is : ； Where: To standardize the sequence of index data values The periodic attention value at the h-th prediction moment t+h; is the softmax activation function, represents the value matrix, and H + 1 is the total number of prediction times; The standardized power grid prediction data sequence is : ； ； Where: Represents the standardized power grid prediction data at the h-th prediction time t+h.

7. The power grid security situation analysis method for big data analysis according to claim 1, characterized in that The grid equipment operation life assessment model includes a life distribution parameter assessment module and a grid equipment operation life calculation module. The life distribution parameter assessment module is used to receive the standardized grid historical data sequence, construct the likelihood function of the life distribution, solve the likelihood function, and obtain the life distribution parameters. The grid equipment operation life calculation module is used to receive the life distribution parameters and calculate the remaining operation life of the grid equipment using the life distribution parameters and the standardized index data value sequence of the equipment failure record indicator; The likelihood function of the lifetime distribution constructed by the lifetime distribution parameter evaluation module is : ； Where: represents the life distribution parameter, represents the life ratio parameter, represents the life shape parameter; For the likelihood function Take the logarithm to obtain the log-likelihood function, solve the first-order partial derivatives of the log-likelihood function, set the first-order partial derivatives to 0 to obtain the equation of the life ratio parameter and the life shape parameter, and solve the equation to obtain the life ratio parameter and the life shape parameter; The calculation formula for calculating the remaining operation life of the grid equipment using the life distribution parameters and the standardized index data value sequence of the equipment failure record indicator is: ； Where: represents the original operating life of grid equipment represents the remaining operating life of grid equipment represents the life control parameter

Images