A method, system and device for monitoring and analyzing power grid assets

Abstract

The present invention relates to the field of data processing technology, and in particular to a method, system and device for monitoring and analyzing power grid assets. The method comprises the steps of: obtaining multidimensional data points of power grid assets; obtaining the difference between any data point in any dimension and the mean of all data points, calculating the degree of outlier of the data point according to the absolute value of the difference, calculating the relative position of the data point according to the difference between the data point and the median of the data of the dimension, using the data point to segment all data of the dimension to which it belongs, obtaining two sub-sample data, calculating the degree of preference of using the data point as a segmentation threshold according to the degree of outlier and relative position of the two sub-sample data, using the data point with the largest degree of preference as the segmentation threshold, constructing a complete isolated tree, and obtaining the abnormal scores of all multidimensional data points, and issuing early warnings to staff according to the size of the abnormal scores. The present invention provides more accurate and timely early warnings for power grid asset data.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a power grid asset monitoring and analysis method, system and equipment. Background Art

[0002] With the continuous development of power grid construction and the rapid development of the power industry, power grid asset monitoring has also become a key link for power companies. Power grid assets refer to various equipment and facilities owned by power companies for the transmission and distribution of electric energy, including power transmission and distribution equipment, transformers, switchgear, relay protection equipment, line facilities, substation facilities, etc. Power grid assets are important means of production and infrastructure for power companies. By monitoring the operating data of power grid assets, the operation and maintenance of power grid equipment can be standardized, the operating risks of the power grid can be reduced, and the safety and stability of the power grid can be improved.

[0003] When processing power grid asset data, the need for anomaly detection is particularly important because abnormal data may indicate potential equipment failure or maintenance needs. The isolation forest algorithm is an effective anomaly detection algorithm that "isolates" abnormal points by constructing an isolation tree. For example, in the patent "Power grid fault diagnosis method and system based on cloud platform and improved isolation forest" with invention publication number CN116150692A, it is pointed out that by constructing an isolation tree to calculate the height anomaly score and density anomaly score of any multidimensional data point to determine whether the data point is abnormal data.

[0004] In the process of constructing the isolation tree, the segmentation threshold at each layer of the isolation tree is a randomly selected segmentation threshold. Since the power grid asset data may contain various types of distributions, including but not limited to normal distribution, skewed distribution or distribution with multi-peak characteristics, the randomly selected segmentation threshold is not sufficient to cope with this complexity, that is, the randomly selected segmentation threshold will affect the accuracy of monitoring the power grid asset data using the isolation forest algorithm, thereby resulting in inaccurate analysis results of the power grid asset data and untimely warnings. Summary of the invention

[0005] The purpose of the present invention is to provide a power grid asset monitoring and analysis method and system to solve the technical problems of inaccurate power grid asset data analysis results and untimely early warning.

[0006] In a first aspect, the present invention provides a method for monitoring and analyzing power grid assets, which adopts the following technical solution:

[0007] Obtain multi-dimensional data points of power grid assets, including: equipment operating voltage, current, and temperature;

[0008] Calculate the degree of outlier of each data point: for any dimension, obtain the difference between any data point in the dimension and the mean of all data points, and calculate the degree of outlier of any data point according to the absolute value of the difference and the variance of all data points. The degree of outlier is positively correlated with the absolute value of the difference and negatively correlated with the variance;

[0009] Calculate the relative position of each data point: for any dimension of any multidimensional data point, calculate the relative position of the data point according to the difference between the data point and the median of the dimension data and the standard deviation of all data points in the dimension, and the relative position is positively correlated with the difference and negatively correlated with the standard deviation;

[0010] Calculate the threshold preference of each data point, obtain the segmentation threshold, and construct a complete isolated tree: use any data point in any dimension to segment all the data that need to be segmented in any layer of the isolation tree to obtain two sub-sample data, and calculate the preference of using the data point as the segmentation threshold based on the difference in the mean, standard deviation and relative position standard deviation of the outlier degree of the two sub-sample data. The preference is positively correlated with the difference. Use the data point with the greatest preference as the segmentation threshold of the isolation tree in this layer, repeat the above steps, and construct a complete isolated tree;

[0011] Abnormal score calculation and early warning: According to the average path length of the data point in each isolated tree, the abnormal score of all multidimensional data points is calculated, and the abnormal score is used to determine whether to issue an early warning to the staff.

[0012] The effect is that the present invention calculates the discreteness and relative position of any data point in any dimension, and obtains the segmentation threshold according to the discreteness and relative position distribution of all data points in the dimension, so as to make the segmentation result of each layer of isolated trees as accurate as possible, and further make the anomaly detection result of the isolated forest more accurate, so as to make the system's analysis result of power grid asset data more accurate.

[0013] The outlier degree satisfies the expression:

[0014] ;

[0015] In the formula, A i,j represents the degree of outlier of the jth data point in the i-th dimension; x i,j Represents the data value of the jth data point in the i-th dimension; represents the mean of all data points in the i-th dimension; n i represents the total number of data points in the i-th dimension; j represents the ordinal number of the data point; norm() represents the standard normalization function.

[0016] The effect is that the present invention calculates the outlier degree of any data point in any dimension by the absolute value of the difference between the average level of all data in the dimension, and combines the standard deviation of all data in the dimension as the credibility of the outlier degree of the data point, thereby making the calculation result of the outlier degree of the data point more accurate.

[0017] The relative position satisfies the expression:

[0018] ;

[0019] In the formula, Q i,j represents the relative position of the jth data point in the i-th dimension; x i,j Represents the data value of the jth data point in the i-th dimension; Represents the median of all data points in the i-th dimension; n i represents the total number of data points in the i-th dimension; j represents the ordinal number of the data point; represents the mean of all data points in the i-th dimension; tanh() represents the hyperbolic tangent function.

[0020] The effect is that the present invention calculates the relative position of any data point in any dimension by the difference between the data point and the mid-latitude of the dimension and the standard deviation of all data points in the dimension, thereby making the calculation result of the relative position of the data point more accurate, and further making the subsequent calculation result of the preference degree of taking any data point as the segmentation threshold based on the relative position calculation of all data points more accurate.

[0021] The method of segmenting sample data of any dimension by using an isolation forest algorithm to obtain two sub-sample data includes the following steps:

[0022] Obtain all data points that need to be segmented, use any one of the data points to segment all the data points that need to be segmented to segment the data points, use data points that are greater than or equal to the value of the data point as the first sub-sample data, and use data points that are less than the value of the data point as the second sub-sample data.

[0023] The preference degree satisfies the expression:

[0024] ;

[0025] Y i,j Indicates the preference for taking the jth data point in the i-th dimension as the segmentation threshold; Represents the mean outlier degree of all data points in the first subsample data; Represents the mean outlier degree of all data points in the second subsample data; The standard deviation of the outlier degree of all data points in the first subsample data; The standard deviation of the outlier level of all data in the second subsample data; Represents the standard deviation of the relative positions of all data points in the first subsample data; Represents the standard deviation of the relative positions of all data points in the second subsample data; exp() represents an exponential function with the natural constant e as the base.

[0026] The effect is that by analyzing the difference in the average level of outliers of the two sub-sample data segments, the preference for using the data point as the segmentation threshold is obtained, so that the segmentation threshold of the isolated tree at this layer is selected more accurately, and the anomaly detection result of the isolation forest is further made more accurate.

[0027] The anomaly score satisfies the expression:

[0028] ;

[0029] Where S(X) represents the anomaly score of the multidimensional data point X; E(h(X)) represents the average path length of the multidimensional data point X in all isolated trees; N represents the total number of samples in the isolated tree; C(N) is a constant representing the expected value of the path length.

[0030] The average path length comprises:

[0031] Obtain the path lengths of the multidimensional data point X from the root node in all isolated trees, and take the average of the path lengths of the multidimensional data point X from the root node in all isolated trees as the average path length E(h(X)) of the multidimensional data point X.

[0032] The warning to the staff according to the abnormal score includes:

[0033] A preset abnormal warning threshold is set. If the average abnormality level of all multidimensional data points is greater than or equal to the preset warning threshold, a warning will be issued to the staff.

[0034] The effect is: to warn the staff through the average level of abnormality of all multi-dimensional data points, to avoid the noise data collected by the sensor affected by the external environment from being detected as abnormal data, thereby triggering an early warning, making the early warning result more accurate.

[0035] In a second aspect, the present invention provides a power grid asset monitoring and analysis system, which adopts the following technical solution:

[0036] A power grid asset monitoring and analysis system, the system is used to execute the above-mentioned power grid asset monitoring and analysis method, comprising:

[0037] The power grid asset multi-dimensional data point acquisition module is used to obtain multi-dimensional information of power grid asset data points. The dimensions of power grid asset data points include: equipment operating voltage, current, and temperature;

[0038] The outlier degree calculation module of each data point is used to obtain the difference between any data point and the mean of all data points in any dimension, and calculate the outlier degree of any data point according to the absolute value of the difference and the variance of all data points. The outlier degree is positively correlated with the absolute value of the difference and negatively correlated with the variance.

[0039] The relative position calculation module of each data point is used to calculate the relative position of any dimension of any multidimensional data point according to the difference between the data point and the median of the dimension data and the standard deviation of all data points in the dimension, wherein the relative position is positively correlated with the difference and negatively correlated with the standard deviation;

[0040] Complete isolation tree construction module: used to segment all the data that need to be segmented in any layer of the isolation tree with any data point in any dimension to obtain two sub-sample data, and calculate the preference of the data point as the segmentation threshold according to the difference in the mean, standard deviation and relative position standard deviation of the outlier degree of the two sub-sample data. The preference is positively correlated with the difference, and the data point with the greatest preference is used as the segmentation threshold of the isolation tree in this layer. Repeat the above steps to construct a complete isolation tree;

[0041] Abnormal score calculation and early warning module: It is used to calculate the abnormal scores of all multidimensional data points according to the average path length of the data points in each isolated tree, and determine whether to issue an early warning to the staff according to the abnormal score.

[0042] In a third aspect, the present invention provides a power grid asset monitoring and analysis device, which adopts the following technical solution:

[0043] A device for monitoring and analyzing power grid assets comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned power grid asset monitoring and analysis method is implemented.

[0044] By adopting the above technical solution, the above-mentioned power grid asset monitoring and analysis method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. The present invention calculates the outlier degree and relative position of any data point in any dimension, and analyzes the difference in the outlier degree and relative position of two data samples after segmenting the dimensional data using any data point, to obtain the preferred degree of using the data point as the segmentation threshold of the isolated tree in this layer, thereby avoiding the problem of inaccurate abnormal results caused by the randomly selected segmentation threshold in each layer of the traditional isolated tree.

[0047] 2. The present invention can greatly reduce the number of isolated trees that need to be constructed by selecting the optimal segmentation threshold for each layer for the isolated tree, thereby saving anomaly detection time and further reducing warning time, thereby making the warning result more timely and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of a method in a power grid asset monitoring and analysis method according to an embodiment of the present invention.

[0049] Figure 2 It is a system module diagram of the present invention.

[0050] Figure 3 It is a device structure diagram of the present invention.

[0051] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0053] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0054] The embodiment of the present invention discloses a method for monitoring and analyzing power grid assets, referring to Figure 1 , comprising steps S1 to S5:

[0055] S1: Obtain multi-dimensional data points of power grid assets;

[0056] It should be noted that the information contained in the grid asset data is relatively extensive, and this data is used to ensure the stable operation and efficient management of the grid. The grid asset data is the real-time monitoring of the operating parameters of the grid equipment, such as voltage, current, and temperature. By monitoring the operating parameters of the equipment and judging the performance and operating status of the equipment, it can help staff optimize the operation of the grid and reduce unexpected power outages caused by equipment failures.

[0057] Specifically, sensors are arranged, a fixed collection cycle is set, the voltage, current, and temperature data of the power grid equipment are collected, and the data of different dimensions collected at the same time are used as a multi-dimensional data point. The preset collection cycle is 2S, which is not specifically limited in the present invention and can be adjusted by the implementer according to the actual situation.

[0058] At this point, the acquisition of multi-dimensional data points of power grid assets is completed.

[0059] S2: Calculate the outlier degree of any data point in any dimension;

[0060] It should be noted that when using the isolation forest algorithm to monitor power grid asset data, the segmentation threshold of each layer in the isolation tree is preset to achieve the construction of the isolation tree, and then the isolation forest is constructed. Among them, the selection of the segmentation threshold of each layer in the isolation tree is very important. When using the segmentation threshold to segment the sample data, the data that is more different from the majority of the data is considered to be abnormal data, and it is expected to separate this part of the abnormal data. Since the difference between abnormal data and normal data is large, the present invention calculates the degree of outlier of any data point in any dimension to facilitate the selection of the subsequent segmentation threshold.

[0061] In one embodiment, the expression for calculating the outlier degree of any data point in any dimension is as follows:

[0062] ;

[0063] In the formula, A i,j represents the degree of outlier of the jth data point in the i-th dimension; x i,j Represents the data value of the jth data point in the i-th dimension; represents the mean of all data points in the i-th dimension; n i represents the total number of data points in the i-th dimension; j represents the ordinal number of the data point; norm() represents the standard normalization function.

[0064] in, Indicates the difference between the data point and the average level of all data points in the dimension where the data point is located. The larger the value, the higher the degree of outlier the data point is.

[0065] It indicates the degree of concentration of the values ​​of all data points in the dimension where the data point is located. The smaller the value is, the more concentrated the values ​​of all data points in the dimension are, and the more credible the outlier degree of the data point is.

[0066] At this point, the outlier degree calculation of any data point in any dimension is completed.

[0067] S3: Calculate the relative position of any data point in any dimension;

[0068] It should be noted that the isolation forest algorithm identifies outliers by constructing multiple isolated trees. Each isolated tree segments the data sample by randomly selecting a dimension and a corresponding segmentation threshold, and gradually "isolates" a single data point. By calculating the relative position of the data point, the isolation forest can optimize the construction process of the isolation tree so that each segmentation can isolate the data point as effectively as possible. If a data point is relatively isolated in a certain dimension, the data point will be isolated earlier, thereby improving the efficiency of the entire algorithm. Therefore, the present invention needs to calculate the relative position of any data point in any dimension.

[0069] In one embodiment, for any data point in any dimension, the relative position of the data point in the dimension is calculated according to the value of the data point and the distribution of the values ​​of all data in the dimension, and the calculation formula is as follows:

[0070] ;

[0071] In the formula, Q i,j represents the relative position of the jth data point in the i-th dimension; x i,j Represents the data value of the jth data point in the i-th dimension; Represents the median of all data points in the i-th dimension; n i represents the total number of data points in the i-th dimension; j represents the ordinal number of the data point; represents the mean of all data points in the i-th dimension; tanh() represents the hyperbolic tangent function.

[0072] in, Indicates the relative position of the data point to the data sample. If its value is negative, it means that the data point is at a relatively forward position in the data sample. If its value is positive, it means that the data point is at a relatively backward position in the data sample. Indicates the density of the data sample. The smaller the value, the denser the data sample, and the higher the credibility of the relative position of the data point to the data sample. Conversely, the lower the credibility Used to The value of is mapped in the interval (0,1).

[0073] At this point, the calculation of the relative position of any data point in any dimension is completed.

[0074] S4: For data of any dimension, calculate the preference of taking any data in the data sample as the segmentation threshold, obtain the segmentation threshold, and construct an isolation tree;

[0075] It should be noted that when setting the segmentation threshold of the isolated forest pair, it is expected to separate the more abnormal data, and the setting of the threshold will affect the segmentation effect. A better threshold can help staff better identify abnormal power grid asset data. In the power grid asset data, the difference between abnormal data and normal data is large, that is, the relative position of the abnormal point in the data set will be greatly different from the normal data. Therefore, it is necessary to classify the data set according to the relative positions of all data points.

[0076] It should be further explained that sensor failure will lead to a certain degree of abnormal data points, which will lead to the presence of abnormal data points in the collected data, and since the possibility of sensor failure is small, the number of abnormal data points collected is small. Therefore, the present invention segments the sample data by taking the value of any data point as the segmentation threshold, and calculates the preference of the data point as the segmentation threshold according to the relative position distribution of the segmented sample data.

[0077] In one embodiment, for any data point in any dimension, the data sample is segmented using the data point, and the data point with a value greater than or equal to the data point is used as the first sub-sample data, and the data point with a value less than the data point is used as the second sub-sample data.

[0078] The preference for using this data point as the segmentation threshold is calculated based on the relative position distribution of the data points in the two sub-sample data. The calculation formula is as follows:

[0079] ;

[0080] Y i,j Indicates the preference for taking the jth data point in the i-th dimension as the segmentation threshold; Represents the mean outlier degree of all data points in the first subsample data; Represents the mean outlier degree of all data points in the second subsample data; The standard deviation of the outlier degree of all data points in the first subsample data; The standard deviation of the outlier level of all data in the second subsample data; Represents the standard deviation of the relative positions of all data points in the first subsample data; Represents the standard deviation of the relative positions of all data points in the second subsample data; exp() represents an exponential function with the natural constant e as the base.

[0081] in, represents the average level of outliers in the first subsample data. Indicates the confidence level of the outlier degree of the first sub-sample data. The smaller its value is, the closer the outlier degree of the sub-sample data is, and the higher the confidence level of its average level is. It indicates the concentration degree of the relative position in the sub-sample data. The smaller its value is, the more concentrated the relative position of the sub-sample data is. The preference for using this data point as the segmentation threshold is calculated by calculating the difference in the average level of the outlier degree of the two sub-sample data. If the difference between the two is larger, it means that the data point can better separate the two sub-sample data, that is, the larger the difference between the two, the higher the preference for using this data point as the segmentation threshold.

[0082] At this point, the calculation of the preference of taking any data point as the segmentation threshold is completed.

[0083] In one embodiment, for the segmentation threshold of any layer in any isolated tree, the data point with the highest degree of preference is selected as the segmentation threshold of the layer, the sample data is segmented, and the two sub-sample data obtained are used as new sample data respectively, and the above steps are repeated to select the segmentation threshold of the next layer, and the segmentation is continued, thereby completing the construction of the isolated tree and further completing the construction of the isolation forest.

[0084] It should be noted that in the process of constructing an isolation tree, each isolated tree is data under the same dimension, and the isolation forest contains multiple isolated trees.

[0085] S5: Calculate the anomaly score of each multidimensional data point and issue an early warning based on the size of the anomaly score;

[0086] In one embodiment, the path length of any multidimensional data point in all isolated trees is obtained, thereby obtaining the mean value of the path length of the multidimensional data point in all isolated trees. The abnormal score of the multidimensional data point is calculated according to the mean value of the path length of the multidimensional data point in all isolated trees, and the calculation formula is as follows:

[0087] ;

[0088] Where S(X) represents the anomaly score of the multidimensional data point X; E(h(X)) represents the average path length of the multidimensional data point X in all isolated trees; N represents the total number of samples in the isolated tree; C(N) is a constant representing the expected value of the path length.

[0089] It should be noted that C(N) represents the expected value of the path length, which is not specifically limited in the present invention and can be determined by the implementer according to the actual situation. For example, , where N represents the total number of samples in the isolation tree; H(N-1) represents the sum of the first N-1 terms of the harmonic sequence.

[0090] In one embodiment, the anomaly scores of all multidimensional data points are obtained, and the average of the anomaly scores of all multidimensional data points is calculated. If the anomaly scores of all multidimensional data points are greater than a preset warning threshold T, an alarm is issued to the staff, and the power grid asset equipment needs to be repaired.

[0091] Wherein, T is a preset warning threshold, which is not specifically limited in the present invention and can be adjusted by implementers according to actual conditions, for example, T=0.8.

[0092] At this point, the abnormality score is obtained, and the staff is warned according to the abnormality score.

[0093] Through the above steps, the abnormal detection of power grid asset equipment is completed, and the early warning of the staff is completed.

[0094] The present invention also discloses a power grid asset monitoring and analysis system. Figure 2 The system is used to execute a power grid asset monitoring and analysis method of the present invention, comprising:

[0095] The power grid asset multi-dimensional data point acquisition module is used to obtain multi-dimensional information of power grid asset data points. The dimensions of power grid asset data points include: equipment operating voltage, current, and temperature;

[0096] The outlier degree calculation module of each data point is used to obtain the difference between any data point and the mean of all data points in any dimension, and calculate the outlier degree of any data point according to the absolute value of the difference and the variance of all data points. The outlier degree is positively correlated with the absolute value of the difference and negatively correlated with the variance.

[0097] The relative position calculation module of each data point is used to calculate the relative position of any dimension of any multidimensional data point according to the difference between the data point and the median of the dimension data and the standard deviation of all data points in the dimension, wherein the relative position is positively correlated with the difference and negatively correlated with the standard deviation;

[0098] Complete isolation tree construction module: used to segment all the data that need to be segmented in any layer of the isolation tree with any data point in any dimension to obtain two sub-sample data, and calculate the preference of the data point as the segmentation threshold according to the difference in the mean, standard deviation and relative position standard deviation of the outlier degree of the two sub-sample data. The preference is positively correlated with the difference, and the data point with the greatest preference is used as the segmentation threshold of the isolation tree in this layer. Repeat the above steps to construct a complete isolation tree;

[0099] Abnormal score calculation and early warning module: It is used to calculate the abnormal scores of all multidimensional data points according to the average path length of the data points in each isolated tree, and determine whether to issue an early warning to the staff according to the abnormal score.

[0100] The present invention also discloses a device for monitoring and analyzing power grid assets, see Figure 3 , including: a processor and a memory, the memory storing computer program instructions, and when the computer program instructions are executed by the processor, a power grid asset monitoring and analysis method of the present invention is implemented.

[0101] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0102] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid storage cube, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0103] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0104] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for monitoring and analyzing power grid assets, characterized in that: Includes steps: Obtain multi-dimensional data points of power grid assets, including: equipment operating voltage, current, and temperature; Calculate the degree of outlier of each data point: For any dimension, obtain the difference between any data point in the dimension and the mean of all data points, and calculate the degree of outlier of any data point according to the absolute value of the difference and the variance of all data points. The degree of outlier is positively correlated with the absolute value of the difference and negatively correlated with the variance; the degree of outlier satisfies the expression: In the formula, A i,j represents the degree of outlier of the jth data point in the i-th dimension; x i,j Represents the data value of the jth data point in the i-th dimension; represents the mean of all data points in the i-th dimension; n i represents the total number of data points in the i-th dimension; j represents the ordinal number of the data point; norm() represents the standard normalization function; Calculate the relative position of each base point: For any dimension of any multidimensional data point, calculate the relative position of the data point based on the difference between the data point and the median of the dimension data and the standard deviation of all data points in the dimension. The relative position is positively correlated with the difference and negatively correlated with the standard deviation. The relative position satisfies the expression: In the formula, Q i,j represents the relative position of the jth data point in the i-th dimension; x i,j Represents the data value of the jth data point in the i-th dimension; Represents the median of all data points in the i-th dimension; n i represents the total number of data points in the i-th dimension; j represents the ordinal number of the data point; Represents the mean of all data points in the i-th dimension; tanh() represents the hyperbolic tangent function; Calculate the threshold preference of each data point, obtain the segmentation threshold, and construct a complete isolated tree: use any data point in any dimension to segment all the data that need to be segmented in any layer of the isolated tree to obtain two sub-sample data. Calculate the preference of using the data point as the segmentation threshold based on the difference in the mean, standard deviation, and standard deviation of the outlier degree of the two sub-sample data. The preference is positively correlated with the difference. Use the data point with the highest preference as the segmentation threshold of the isolated tree in this layer. Repeat the above steps to construct a complete isolated tree. The preference satisfies the expression: Y i,j Indicates the preference for taking the jth data point in the i-th dimension as the segmentation threshold; Represents the mean outlier degree of all data points in the first subsample data; Represents the mean outlier degree of all data points in the second subsample data; The standard deviation of the outlier degree of all data points in the first subsample data; The standard deviation of the outlier level of all data in the second subsample data; Represents the standard deviation of the relative positions of all data points in the first subsample data; Represents the standard deviation of the relative positions of all data points in the second sub-sample data; exp() represents an exponential function with the natural constant e as the base; Abnormal score calculation and early warning: According to the average path length of the data point in each isolated tree, the abnormal score of all multidimensional data points is calculated, and the abnormal score is used to determine whether to issue an early warning to the staff.

2. A power grid asset monitoring and analysis method according to claim 1, characterized in that: The method of segmenting sample data of any dimension by using an isolation forest algorithm to obtain two sub-sample data includes the following steps: Obtain all data points that need to be segmented, use any one of the data points to segment all the data points that need to be segmented to segment the data points, use data points that are greater than or equal to the value of the data point as the first sub-sample data, and use data points that are less than the value of the data point as the second sub-sample data.

3. A power grid asset monitoring and analysis method according to claim 1, characterized in that: The anomaly score satisfies the expression: Where S(X) represents the anomaly score of the multidimensional data point X; E(h(X)) represents the average path length of the multidimensional data point X in all isolated trees; N represents the total number of samples in the isolated tree; C(N) is a constant representing the expected value of the path length.

4. A method for monitoring and analyzing power grid assets according to claim 3, characterized in that: The average path length comprises: Obtain the path lengths of the multidimensional data point X from the root node in all isolated trees, and take the average of the path lengths of the multidimensional data point X from the root node in all isolated trees as the average path length E(h(X)) of the multidimensional data point X.

5. A method for monitoring and analyzing power grid assets according to claim 1, characterized in that: The warning to the staff according to the abnormal score includes: A preset abnormal warning threshold is set. If the average abnormality level of all multidimensional data points is greater than or equal to the preset warning threshold, a warning will be issued to the staff.

6. A power grid asset monitoring and analysis system, characterized in that: The system is used to execute a power grid asset monitoring and analysis method according to any one of claims 1 to 5, comprising: The power grid asset multi-dimensional data point acquisition module is used to obtain multi-dimensional information of power grid asset data points. The dimensions of power grid asset data points include: equipment operating voltage, current, and temperature; The outlier degree calculation module of each data point is used to obtain the difference between any data point and the mean of all data points in any dimension, and calculate the outlier degree of any data point according to the absolute value of the difference and the variance of all data points. The outlier degree is positively correlated with the absolute value of the difference and negatively correlated with the variance. Each is a relative position calculation module of a base point: for calculating the relative position of any dimension of any multidimensional data point according to the difference between the data point and the median of the dimension data and the standard deviation of all data points in the dimension, wherein the relative position is positively correlated with the difference and negatively correlated with the standard deviation; Complete isolation tree construction module: used to segment all the data that need to be segmented in any layer of the isolation tree with any data point in any dimension to obtain two sub-sample data, and calculate the preference of the data point as the segmentation threshold according to the difference in the mean, standard deviation and relative position standard deviation of the outlier degree of the two sub-sample data. The preference is positively correlated with the difference, and the data point with the greatest preference is used as the segmentation threshold of the isolation tree in this layer. Repeat the above steps to construct a complete isolation tree; Abnormal score calculation and early warning module: It is used to calculate the abnormal scores of all multidimensional data points according to the average path length of the data points in each isolated tree, and determine whether to issue an early warning to the staff according to the abnormal score.

7. A device for monitoring and analyzing power grid assets, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a power grid asset monitoring and analysis method according to any one of claims 1 to 5 is implemented.

Images