A terminal data processing method for a smart transformer area terminal

Abstract

The present application relates to the field of data processing, and particularly relates to a terminal data processing method for an intelligent transformer area terminal, comprising: obtaining a target subsequence; obtaining a plurality of working condition data sequences of any preset working condition in history, respectively calculating the similarity of the target subsequence and each working condition data sequence under the same preset working condition to construct a similarity sequence, calculating the final similarity of the target subsequence and the preset working condition according to the similarity sequence, obtaining the final similarity of the target subsequence and all preset working conditions through iteration, and taking the preset working condition corresponding to the maximum value of the final similarity as the matching working condition corresponding to the target subsequence; calculating the abnormality degree of the target subsequence and the matching working condition, and processing the target subsequence according to the abnormality degree. The present application can improve the accuracy of voltage data judged by the terminal.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing. More particularly, the present application relates to a terminal data processing method for a smart transformer area terminal. BACKGROUND

[0002] The smart transformer area terminal has the function of electric energy monitoring, which is of great significance to the stable, safe and efficient operation of the power system. The content of electric energy monitoring includes detection and identification of power quality disturbance, the purpose of which is to improve and enhance power quality. Whether the voltage data is abnormal is one of the important indicators of power quality disturbance. In the traditional electric energy monitoring technology, whether the voltage is abnormal is judged by judging the amplitude fluctuation of the voltage data, but in actual application, there are cases where the voltage data is abnormally fluctuating but the voltage is normal. Therefore, the judgment result of abnormal voltage in the traditional technology is not accurate. SUMMARY

[0003] The main purpose of the embodiments of the present application is to provide a terminal data processing method for a smart transformer area terminal, aiming to improve the accuracy of the smart transformer area terminal in judging abnormal voltage.

[0004] To achieve the above-mentioned purpose, the embodiments of the present application provide a terminal data processing method for a smart transformer area terminal, which comprises: acquiring a target sub-sequence, the target sub-sequence being intercepted from a pre-processed voltage data sequence; acquiring a plurality of working condition data sequences of any one of the historical preset working conditions, respectively calculating the similarity of the target sub-sequence and each working condition data sequence under the same preset working condition to construct a similarity sequence, calculating the final similarity of the target sub-sequence and the preset working condition according to the similarity sequence, obtaining the final similarity of the target sub-sequence and all preset working conditions by traversal, and taking the preset working condition corresponding to the maximum value of the final similarity as the matching working condition of the target sub-sequence; calculating the abnormality degree of the target sub-sequence and the matching working condition, and processing the target sub-sequence according to the abnormality degree.

[0005] In some embodiments, calculating the final similarity of the target sub-sequence and the preset working condition comprises: calculating the sequence length difference of the target sub-sequence and any working condition data sequence under the same preset working condition, taking the sequence length difference as a weight, weighting the similarity of the target sub-sequence and any working condition data sequence to obtain a weighted result; obtaining the weighted result of each working condition data sequence under the same preset working condition by traversal, and summing all weighted results to obtain the final similarity.

[0006] In some embodiments, calculating the final similarity of the target sub-sequence and the preset working condition comprises: summing the similarities in the similarity sequence to obtain the final similarity.

[0007] In some embodiments, the calculating the abnormality degree comprises: calculating a standard deviation value of the target sub-sequence and the similarity sequence of the matching working condition, taking the final similarity of the target sub-sequence and the matching working condition as a feature similarity, and normalizing a reciprocal of a product of the standard deviation value and the feature similarity to obtain the abnormality degree.

[0008] In some embodiments, the obtaining the target sub-sequence comprises: dynamically and continuously sampling the voltage data sequence by a sliding window; calculating a change feature value in the sliding window to construct a change feature value sequence; traversing the change feature value sequence to take an inflection point in the change feature value sequence as a segmentation point; and taking voltage data between two adjacent segmentation points as a target sub-sequence.

[0009] In some embodiments, the calculating the change feature value comprises: calculating an absolute value of a difference between any two adjacent voltage data in the sliding window to construct an absolute value sequence, calculating a mean value of the absolute value sequence, and obtaining a maximum value in the absolute value sequence; in response to an absolute value in the absolute value sequence being greater than the mean value, accumulating a change frequency by one; taking a ratio of the maximum value to the mean value as a significant value; taking a ratio of the accumulated value of the change frequency to the length of the sliding window as a duration value; and taking a product of the significant value and the duration value as the change feature value.

[0010] In some embodiments, the processing the target sub-sequence according to the abnormality degree value comprises: in response to the abnormality degree being not less than a preset threshold, marking the target sub-sequence as abnormal and sending an alarm signal; and in response to the abnormality degree being less than the preset threshold, marking the target sub-sequence as normal.

[0011] In some embodiments, the voltage data sequence is subjected to padding processing and filtering and denoising.

[0012] Advantages of the present application:

[0013] The present application calculates the similarity of a target sub-sequence and each working condition data sequence of any preset working condition, and constructs a similarity sequence therefrom, calculates a final similarity of the target sub-sequence and the corresponding preset working condition according to a plurality of similarities in the similarity sequence, and then takes the preset working condition with the maximum final similarity as a matching working condition of the target sub-sequence, at this time, the present application performs a first calculation between the target sub-sequence and the matching working condition; after determining the matching working condition, the abnormality degree of the target sub-sequence and the matching working condition is calculated, and finally the target sub-sequence is processed according to the calculation result, at this time, the present application performs a second calculation between the target sub-sequence and the matching working condition; compared with the traditional technology of judging whether the voltage is abnormal by only judging the abnormal amplitude fluctuation of the voltage data, the present application improves the accuracy of the terminal in judging the abnormal voltage. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 is a flow chart of steps S1-step S3 in a terminal data processing method for a smart terminal. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0016] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0017] Referring to Figure 1 A terminal data processing method for a smart terminal includes steps S1-step S3, as follows:

[0018] Step S1: Obtain a target sub-sequence, which is cut from the pre-processed voltage data sequence.

[0019] It should be noted that the voltage data sequence refers to the voltage data sequence to be analyzed, which can be obtained by collecting the voltage data sequence. For example, the collection module of the smart terminal collects the voltage data sequence from the electric energy meter according to the preset collection task, and stores the collected voltage data sequence. The collection task can be to collect the voltage data sequence of a certain time period. In some embodiments, the voltage data sequence can be sorted in chronological order.

[0020] It should be noted that the pre-processed voltage data sequence can be cut into multiple segments, and one of them can be selected as the target sub-sequence.

[0021] Step S2: Obtain a plurality of working condition data sequences of any one of the preset working conditions, respectively calculate the similarity between the target sub-sequence and each working condition data sequence under the same preset working condition to construct a similarity sequence, calculate the final similarity between the target sub-sequence and the preset working condition according to the similarity sequence, and obtain the final similarity between the target sub-sequence and all preset working conditions by iteration. The maximum value of the final similarity corresponds to the matching working condition of the target sub-sequence.

[0022] It should be noted that the working condition data sequence of the preset working condition is obtained from the historical voltage data. The preset working condition can include some peak power consumption periods, trough power consumption periods and other special working conditions during the historical power consumption period, and can also include normal power consumption conditions and other abnormal power consumption conditions. Correspondingly, each preset working condition contains a plurality of working condition data sequences corresponding thereto. For example, the preset working condition is the peak power consumption period, which contains five working condition data sequences under this preset working condition.

[0023] It should be added that the intelligent distribution terminal can extract historical operating data as needed after acquiring it.

[0024] Step S3: Calculate the degree of anomaly between the target subsequence and the matching working condition, and process the target subsequence according to the degree of anomaly.

[0025] It should be further explained that after determining the matching condition of the target sub-sequence, this invention also calculates the degree of anomalousness between the target sub-sequence and the corresponding matching condition. The degree of anomalousness is used to determine whether the accuracy of the matching condition of the target sub-sequence meets the qualification standard. By calculating the degree of anomalousness, this invention enables the intelligent distribution station terminal to determine the matching condition of the target sub-sequence more accurately.

[0026] Understandably, calculating the final similarity between the target subsequence and the preset working condition includes: calculating the sequence length difference between the target subsequence and any working condition data sequence under the same preset working condition; using the sequence length difference as a weight to weight the similarity between the target subsequence and any working condition data sequence to obtain a weighted result; iterating through the target subsequence to obtain the weighted result between each working condition data sequence under the same preset working condition; summing all weighted results to obtain the final similarity.

[0027] It should be further noted that the similarity calculation method for the target subsequence and each working condition data sequence under the same preset working condition can be the Framing distance method. The Framing distance method prioritizes the geometric similarity between the two sequences, normalizing the inverse of the Framing distance between the two output vectors to obtain the similarity value. When using the Framing distance method to calculate the similarity between the target subsequence and a working condition data sequence, the difference in length between them can cause deviations in the similarity calculation, thus affecting the accuracy of the final similarity calculation between the target subsequence and the preset working condition. Therefore, the impact of different sequence lengths on the similarity must be considered when using the Framing distance method.

[0028] For example, when using the Framing distance method to calculate the similarity between a target subsequence and a working condition data sequence, the final similarity calculation formula is as follows:

[0029]

[0030] In formula (1), For the first The number of working condition data sequences for each preset working condition; For the first The first of the preset working conditions Section working condition data sequence and the first The similarity of the target subsequences; the first target sub-sequence belongs to the first preset working condition; the difference between the sequence length of the first target sub-sequence and the first preset working condition data; the final similarity between the first target sub-sequence and the first preset working condition; the difference between the sequence length of the first target sub-sequence and the first preset working condition data; the final similarity between the first target sub-sequence and the first preset working condition; the difference between the sequence length of the first target sub-sequence and the first preset working condition data; the final similarity between the first target sub-sequence and the first preset working condition; represents an exponential function.

[0031] For the above formula (1), it needs to be pointed out that, is calculated by the Frasni distance method, the higher the contribution degree of the current first target sub-sequence to the identification of the first preset working condition is, that is, the greater the possibility of the first target sub-sequence belonging to the first preset working condition is. the higher the contribution degree of the current first target sub-sequence to the identification of the first preset working condition is, that is, the greater the possibility of the first target sub-sequence belonging to the first preset working condition is. the higher the contribution degree of the current first target sub-sequence to the identification of the first preset working condition is, that is, the greater the possibility of the first target sub-sequence belonging to the first preset working condition is. the higher the contribution degree of the current first target sub-sequence to the identification of the first preset working condition is, that is, the greater the possibility of the first target sub-sequence belonging to the first preset working condition is. the higher the contribution degree of the current first target sub-sequence to the identification of the first preset working condition is, that is, the greater the possibility of the first target sub-sequence belonging to the first preset working condition is.

[0032] For the above formula (1), it needs to be pointed out that, when the Frasni distance method is used to calculate the similarity between the target sub-sequence and the working condition data sequence, the sequence length difference between the target sub-sequence and the working condition data sequence directly affects the similarity accuracy of the two. The greater the length difference between the two data sequences, the greater the Frasni distance, and the greater the similarity value deviation, the lower the confidence of the similarity, so that the contribution made when calculating the final similarity is also lower. Therefore, on this basis, the present application weights the similarity between the target sub-sequence and the working condition data sequence according to the sequence length difference between the target sub-sequence and the working condition data sequence, so as to reduce the error caused by the Frasni distance method when calculating the similarity.

[0033] For the above formula (1), it needs to be pointed out that, can be set as needed.

[0034] For the above formula (1), it needs to be pointed out that, the greater the sequence length difference between the target sub-sequence and the working condition data sequence, the smaller the corresponding weight is, the smaller the corresponding weight is, the smaller the corresponding weight is, the smaller the corresponding weight is, the smaller the corresponding weight is,

[0035] Exemplarily, under one preset working condition, five working condition data sequences are provided: the similarity between the target sub-sequence and the first working condition data sequence is calculated by using the Franshiche distance method, the above steps are repeated to calculate the similarity between the target sub-sequence and the other four working condition data sequences, and a similarity sequence is constructed; the sequence difference between the target sub-sequence and the first working condition data sequence is calculated, the inverse number of the sequence difference is taken as the exponent of the natural base e to form a power, and the power is taken as the weight of the similarity between the target sub-sequence and the first working condition data sequence; the similarity between the target sub-sequence and the first working condition data sequence is weighted by using the weight to obtain the product of the weight and the similarity between the target sub-sequence and the first working condition data sequence; the above steps are repeated to calculate the weighted similarities between the target sub-sequence and the other four working condition data sequences; and the five weighted similarities are summed to obtain the final similarity between the target sub-sequence and the preset working condition.

[0036] Exemplarily, there are 10 preset working conditions, and the final similarities between the target sub-sequence and the 10 preset working conditions are calculated according to the above example, and the preset working condition corresponding to the maximum value of the final similarities is taken as the matching working condition of the target sub-sequence.

[0037] It can be understood that calculating the final similarity between the target sub-sequence and the preset working condition includes summing the similarities in the similarity sequence to obtain the final similarity.

[0038] It should be noted that the DTW (Dynamic Time Warping) algorithm can be used to calculate the similarity between the target sub-sequence and one working condition data sequence, and the DTW algorithm outputs the shortest path distance between two sequences, and therefore the reciprocal of the output result is normalized to obtain the similarity value.

[0039] Exemplarily, when the DTW algorithm is used to calculate the similarity between the target sub-sequence and one working condition data sequence, the calculation formula of the final similarity is as follows:

[0040]

[0041] wherein, is the number of working condition data sequences of the i-th preset working condition; is the similarity between the i-th target sub-sequence and the j-th working condition data sequence of the i-th preset working condition; is the final similarity between the i-th target sub-sequence and the i-th preset working condition.

[0042] ​​​​​​Exemplarily, under one preset working condition, five working condition data sequences are provided: the DTW algorithm is used to calculate the similarity of the target sub-sequence and the first working condition data sequence; the above steps are repeated to calculate the similarity of the target sub-sequence and the other four working condition data sequences respectively, and a similarity sequence is constructed; the five similarities in the similarity sequence are summed up to obtain the final similarity of the target sub-sequence and the preset working condition.

[0043] Exemplarily, there are 10 preset working conditions, and the final similarity of the target sub-sequence and the 10 preset working conditions is calculated according to the above example respectively, and the preset working condition corresponding to the maximum value of the final similarity is taken as the matching working condition of the target sub-sequence.

[0044] It can be understood that the calculation of the abnormality degree includes: calculating the standard deviation value of the similarity sequence of the target sub-sequence and the matching working condition, taking the final similarity of the target sub-sequence and the matching working condition as the characteristic similarity, and performing normalization processing on the negative number of the product of the standard deviation value and the characteristic similarity to obtain the abnormality degree.

[0045] It should be noted that the standard deviation is used to describe the dispersion degree of data, and the smaller the standard deviation, the more concentrated the data is; the larger the standard deviation, the more dispersed the data is. In the present application, the larger the standard deviation value, the more concentrated the multiple similarities of the target sub-sequence and the multiple working condition data sequences of the matching working condition are, so the matching working condition is more accurate; at the same time, the contribution degree of the similarity sequence of the target sub-sequence and the matching working condition in the subsequent calculation of the abnormality degree using the final similarity is also higher.

[0046] It should be noted that the abnormality degree obtained by normalization processing is in the interval of 0 to 1.

[0047] Exemplarily, the calculation formula of the abnormality degree is as follows:

[0048]

[0049] Wherein, is the abnormality degree of the i-th segment, is a normalization function, is the standard deviation value of the i-th target sub-sequence and the similarity sequence corresponding to the i-th preset working condition, is the final similarity of the i-th target sub-sequence and the i-th preset working condition.

[0050] ​​​​​Exemplarily, there are 10 preset working conditions, and 5 working condition data sequences are set in one preset working condition. After calculating the final similarity according to the above example, the first preset working condition is taken as the matching working condition. The standard deviation value of the similarity sequence of the matching working condition is calculated, the final similarity of the target sub-sequence and the matching working condition is taken as the characteristic similarity, and the reciprocal of the product of the standard deviation value and the characteristic similarity is normalized to obtain the abnormality degree.

[0051] It should be noted that according to formula (1) and formula (2), the greater the multiple similarities of the target sub-sequence and the multiple working condition data sequences of the matching working condition, the greater the final similarity. According to formula (3), the greater the final similarity, the higher the abnormality degree. According to formula (3), the smaller the standard deviation value of the similarity sequence of the target sub-sequence and the matching working condition, the higher the abnormality degree. Therefore, the higher the abnormality degree, the more accurate the matching working condition of the target sub-sequence. After obtaining the matching working condition of the target sub-sequence, the abnormality degree is calculated, which represents the accuracy of the matching working condition of the target sub-sequence.

[0052] It can be understood that obtaining the target sub-sequence includes: dynamically and continuously sampling the voltage data sequence through the sliding window; calculating the change characteristic value in the sliding window to construct a change characteristic value sequence; traversing the change characteristic value sequence, and taking the inflection point in the change characteristic value sequence as a segmentation point; and taking the voltage data between two adjacent segmentation points as a target sub-sequence.

[0053] It should be noted that the starting point of the sliding window is the starting point of the voltage data sequence, and the end point of the sliding window is the end point of the voltage data sequence. The sliding window dynamically and continuously samples the voltage data sequence according to the preset window length and step length, and the length of the sliding window represents the number of data that can be accommodated in the window.

[0054] Exemplarily, the preset step length is 1, and the preset window length is 20. The change characteristic value is calculated once every time the sliding window slides one step. After the entire voltage data sequence is sampled by the sliding window, the change characteristic value sequence is constructed, the change characteristic value sequence is traversed, the inflection point in the change characteristic value sequence is taken as a segmentation point, and the voltage data in the voltage data sequence corresponding to the segmentation point is segmented. For example, the change characteristic value in the change characteristic value sequence first increases and then decreases, and the change characteristic value that starts to decrease is taken as an inflection point. The voltage data in the voltage data sequence corresponding to the inflection point is segmented. Through the above steps, two adjacent segmentation points are found, and the voltage data between the two adjacent segmentation points is taken as a target sub-sequence.

[0055] It should be noted that the present application discusses the process of determining the matching condition for a sub-sequence from the perspective of a sub-sequence, which is defined as a target sub-sequence for clarity. After segmentation, multiple sub-sequences can be obtained, and after determining the matching conditions of the sub-sequences respectively, if the matching conditions of adjacent sub-sequences are the same, the adjacent sub-sequences are merged. When calculating the abnormality degree thereafter, the sub-sequence corresponding to the maximum final similarity is determined, and the abnormality degree is calculated according to the similarity sequence of the sub-sequence and the final similarity of the sub-sequence.

[0056] It can be understood that the calculation of the change characteristic value includes: calculating the absolute value of the difference between any two adjacent voltage data in the sliding window, constructing an absolute value sequence, calculating the mean value of the absolute value sequence, and obtaining the maximum value in the absolute value sequence; in response to the absolute value in the absolute value sequence being greater than the mean value, the change frequency is accumulated once; the ratio of the maximum value to the mean value is taken as the significant value; the ratio of the accumulated value of the change frequency to the length of the sliding window is taken as the duration value; and the product of the significant value and the duration value is taken as the change characteristic value.

[0057] Exemplarily, the formula for calculating the change characteristic value is as follows:

[0058]

[0059] Wherein, is the absolute value of the difference between the i-th voltage data and the i+1-th voltage data in the sliding window with the starting point being is the maximum value of the absolute value of the difference between adjacent voltage data in the sliding window; is the mean value of the absolute value of the difference between all adjacent voltage data in the sliding window; is the significant value; is an indicator function, which is 1 when , otherwise 0; is the accumulated value of the change frequency; is the duration value, is the length of the sliding window, i.e. the number of voltage data that the sliding window can accommodate; is the change characteristic value of the sliding window.

[0060] It should be noted that the ratio can reflect the significant degree of the change of the voltage data in the sliding window, the larger the ratio is, the more significant the change of the voltage data is, and the more likely it is that the window contains a segmentation point.

[0061] It should be noted that represents that the change characteristic value of the sliding window is greater than the change characteristic value of the previous sliding window. ​​​The number of absolute values of differences between two adjacent voltage data in the sliding window as a starting point, that is, the cumulative value of the number of changes, is greater than the mean value; the mean value of the absolute values of the differences between all adjacent voltage data in the sliding window is taken as a threshold value, and the number of voltage data greater than the threshold value is counted, which can quantify the persistence of voltage data changes in the sliding window, and the more the number of voltage data greater than the threshold value, the stronger the persistence of voltage data changes in the sliding window, and the greater the possibility that the sliding window contains a segmentation point.

[0062] It needs to be noted that since the voltage data sequence to be analyzed may contain multiple working condition periods, it needs to be segmented and analyzed. The present application uses the sliding window technology to dynamically and continuously sample the voltage data sequence, which can take into account some situations affecting analysis, such as the voltage data sequence may contain noise during collection, and the instantaneous slight unstable voltage change does not represent a working condition change. Considering the above influences, the present application only segments under the condition of persistent and significant voltage change.

[0063] It can be understood that processing the target sub-sequence according to the abnormality degree value includes: in response to the abnormality degree being not less than a preset threshold value, marking the target sub-sequence as abnormal and sending an alarm signal; and in response to the abnormality degree being less than the preset threshold value, marking the target sub-sequence as normal.

[0064] It needs to be noted that the present application determines the matching working condition for the target sub-sequence according to the final similarity, and the matching working condition is the working condition to which the target sub-sequence belongs as judged by the intelligent transformer area terminal; then the present application calculates the abnormality degree, and if the abnormality degree is not less than a preset threshold value, it indicates that the matching working condition accuracy of the target sub-sequence is qualified, at this time, the intelligent transformer area terminal marks the target sub-sequence as abnormal and sends an alarm signal; if the abnormality degree is less than the preset threshold value, it indicates that the matching working condition accuracy of the target sub-sequence is unqualified, at this time, the target sub-sequence is marked as normal.

[0065] Exemplarily, the preset threshold value is 0.7, and the matching working condition of the target sub-sequence is a certain abnormal working condition, at this time: if the calculated abnormality degree is not less than 0.7, the target sub-sequence is marked as abnormal and an alarm signal is sent, so that the staff knows that the voltage is in a certain abnormal working condition and can make some fault troubleshooting preparations; if the calculated abnormality degree is less than 0.7, the target sub-sequence is marked as normal and no alarm signal is sent.

[0066] Exemplarily, the preset threshold value is 0.7, and the matching working condition of the target sub-sequence is the peak electricity consumption period, at this time: if the calculated abnormality degree is not less than 0.7, the target sub-sequence is marked as abnormal and an alarm signal is sent, so that the staff knows that the voltage is in the peak electricity consumption period and can make some emergency preparations, such as increasing power supply equipment; if the calculated abnormality degree is less than 0.7, the target sub-sequence is marked as normal and no alarm signal is sent.

[0067] It can be understood that the preprocessing includes filling processing and filtering denoising on the voltage data sequence.

[0068] It should be noted that due to environmental interference or network congestion and other factors, the collected data may have missing values. In order to ensure the integrity and accuracy of the data and avoid affecting the subsequent calculation, the present application adopts a regression filling method to fill in the missing values. The regression filling method is a filling method based on data model, which can establish a model according to the known data, predict and fill in the missing values, so as to ensure the integrity of the data. Specifically, this method estimates the reasonable value of the missing data by analyzing the correlation between the data, using the regression model to ensure that the statistical characteristics of the filled data are consistent with the original data. The voltage data sequence may contain noise during the collection process, so it needs to be denoised, and the present application adopts a filtering denoising method.

[0069] It should be pointed out that the above describes the preferred embodiments of the embodiments of the application with reference to the drawings, and does not limit the scope of the embodiments of the application. For those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of protection of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A terminal data processing method for a smart transformer terminal, characterized in that, The method comprises the following steps: Dynamic continuous sampling is performed on the preprocessed voltage data sequence through a sliding window, the absolute value of the difference between any two adjacent voltage data in the sliding window is calculated, an absolute value sequence is constructed, the mean value of the absolute value sequence is calculated, the maximum value in the absolute value sequence is obtained, in response to the absolute value in the absolute value sequence being greater than the mean value, the number of changes is accumulated once, the ratio of the maximum value to the mean value is taken as a significant value, the ratio of the accumulated value of the number of changes to the length of the sliding window is taken as a persistence value, the product of the significant value and the persistence value is taken as a change feature value, a change feature value sequence is constructed, the inflection points in the change feature value sequence are taken as segmentation points, and the voltage data between two adjacent segmentation points is taken as a target subsequence; A plurality of working condition data sequences of any one of the historical preset working conditions are obtained, the Fréchet distance method or the DTW algorithm is used to calculate the similarity between the target subsequence and each working condition data sequence under the same preset working condition, respectively, to construct a similarity sequence, the sequence length difference between the target subsequence and any working condition data sequence under the same preset working condition is calculated, the sequence length difference is taken as a weight, and the similarity between the target subsequence and any working condition data sequence is weighted to obtain a weighted result; The weighted results of the target subsequence and each working condition data sequence under the same preset working condition are traversed, all the weighted results are summed to obtain a final similarity, the final similarities of the target subsequence and all preset working conditions are traversed, and the preset working condition corresponding to the maximum value of the final similarities is taken as a matching working condition of the target subsequence; The standard deviation value of the similarity sequence of the target subsequence and the matching working condition is calculated, the final similarity of the target subsequence and the matching working condition is taken as a feature similarity, the reciprocal of the product of the standard deviation value and the feature similarity is normalized to obtain an abnormality degree, and the target subsequence is processed according to the abnormality degree.

2. The terminal data processing method for the intelligent transformer area terminal according to claim 1, characterized in that, The calculation of the final similarity of the target subsequence and the preset working condition comprises: The similarities in the similarity sequence are summed to obtain the final similarity.

3. The terminal data processing method for the intelligent transformer terminal according to claim 1, characterized in that, Processing the target subsequence according to the abnormality degree value comprises: In response to the abnormality degree being not less than a preset threshold, the target subsequence is marked as abnormal and an alarm signal is sent; In response to the abnormality degree being less than a preset threshold, the target subsequence is marked as normal.

4. The terminal data processing method for the intelligent transformer area terminal according to claim 1, characterized in that, The preprocessing comprises: The voltage data sequence is filled and filtered to remove noise.

Images