A method and terminal for detecting fluctuations in station monitoring indicators

Abstract

This invention discloses a method and terminal for detecting fluctuations in monitoring indicators at a facility. The method involves acquiring historical monitoring indicator data sequences from the facility and dividing these sequences into multiple sub-sequences in chronological order. For each sub-sequence, the method calculates the average trend value, the variance of the residual values, and the growth rate of the residual variance, thereby determining reference ranges for the average trend value, the variance of the residual values, and the growth rate of the residual variance. Finally, the method acquires the latest monitoring indicator data sequence from the facility and determines whether the average trend value, the variance of the residual values, or the growth rate of the residual variance in the latest data sequence exceeds the corresponding reference ranges to confirm whether fluctuations exist in the latest monitoring indicator data sequence. This invention enables multi-faceted and accurate determination of fluctuations in the monitoring indicator data of a facility, facilitating timely early warning by maintenance personnel and providing reliable protection for facility safety.

Description

Technical Field

[0001] This invention relates to the field of electronic information technology, and in particular to a method and terminal for detecting fluctuations in monitoring indicators at a field station. Background Technology

[0002] Existing photovoltaic-storage-charging-testing stations, as part of the infrastructure for electric vehicle charging testing, are expanding their coverage. To ensure that these stations can provide a stable power supply for electric vehicles, it is currently essential to improve the safety management level of these stations and guarantee their safe operation.

[0003] However, the existing data distribution of the safety monitoring indicators of the stations is either single-peak or multi-peak, and the data changes are relatively unstable. Therefore, how to effectively identify the waveform of the station monitoring indicator data is the key to judging whether the station is operating abnormally. Summary of the Invention

[0004] The technical problem to be solved by this invention is to propose a method and terminal for detecting fluctuations in monitoring indicators at a facility, which can effectively identify fluctuations in the monitoring indicator data of the facility, so as to make timely early warnings and provide protection for facility safety.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0006] A method for detecting fluctuations in monitoring indicators at a facility includes the following steps:

[0007] S1. Obtain the historical monitoring indicator data sequence of the station within a first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order.

[0008] S2. Calculate the average trend value, variance of residual value, and growth rate of variance of residual value for each subsequence of the index data according to the time series decomposition algorithm;

[0009] S3. Based on the average trend value of all the indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values.

[0010] S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations.

[0011] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows:

[0012] A monitoring terminal for detecting fluctuations in station indicators includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps:

[0013] S1. Obtain the historical monitoring indicator data sequence of the station within a first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order.

[0014] S2. Calculate the average trend value, variance of residual value, and growth rate of variance of residual value for each subsequence of the index data according to the time series decomposition algorithm;

[0015] S3. Based on the average trend value of all the indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values.

[0016] S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations.

[0017] The beneficial effects of this invention are as follows: It provides a method and terminal for detecting fluctuations in station monitoring indicators. Based on the historical monitoring indicator data sequence of the station, it calculates and generates reference ranges for the mean trend value, variance of residual values, and growth rate of residual variance for fluctuation detection. Then, during detection, it acquires the latest monitoring indicator data sequence and determines whether the mean trend value, variance of residual values, and growth rate of residual variance of the latest monitoring indicator data sequence exceed the corresponding reference ranges. This enables multi-angle and accurate judgment of whether the monitoring indicator data of the station has fluctuated, so that operation and maintenance personnel can make timely early warnings and provide reliable protection for station safety. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating the steps of a method for detecting fluctuations in station monitoring indicators according to the present invention;

[0019] Figure 2This is a flowchart illustrating the method for judging the average value anomaly of a trend value in a site monitoring index fluctuation detection method according to the present invention.

[0020] Figure 3 This is a flowchart illustrating the residual variance anomaly judgment method of a station monitoring index fluctuation detection method according to the present invention.

[0021] Figure 4 This is a flowchart illustrating the abnormal variance growth rate of residual values ​​in a method for detecting fluctuations in station monitoring indicators according to the present invention.

[0022] Figure 5 This is a flowchart of a method for detecting fluctuations in station monitoring indicators according to the present invention, which involves judging the variance uniformity anomaly of residual values.

[0023] Figure 6 This is a system block diagram of a station monitoring index fluctuation detection terminal according to the present invention.

[0024] Label Explanation:

[0025] 1. A terminal for detecting fluctuations in monitoring indicators at a monitoring station; 2. A memory; 3. A processor. Detailed Implementation

[0026] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0027] Please refer to Figures 1 to 5 A method for detecting fluctuations in station monitoring indicators, comprising the following steps:

[0028] S1. Obtain the historical monitoring indicator data sequence of the station within a first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order.

[0029] S2. Calculate the average trend value, variance of residual value, and growth rate of variance of residual value for each subsequence of the index data according to the time series decomposition algorithm;

[0030] S3. Based on the average trend value of all the indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values.

[0031] S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations.

[0032] As can be seen from the above description, the beneficial effects of the present invention are as follows: Based on the historical monitoring index data sequence of the station, a reference range for the mean trend value, a reference range for the variance of the residual value, and a reference range for the growth rate of the variance of the residual value are calculated and generated for fluctuation detection. Then, during detection, the latest monitoring index data sequence is obtained, and the average trend value, the variance of the residual value, and the growth rate of the variance of the residual value of the latest monitoring index data sequence are determined according to whether they exceed the corresponding reference range. This enables multi-angle and accurate judgment of whether the monitoring index data of the station has fluctuated, so that operation and maintenance personnel can make timely early warnings and provide reliable protection for the safety of the station.

[0033] Furthermore, step S4 also includes:

[0034] Subtracting the pre-set calibration index data sequence from the latest monitoring index data sequence yields the index residual sequence.

[0035] The residual sequence of the indicator is divided into two residual subsequences according to the time sequence. The variance consistency results of the two residual subsequences are calculated to determine whether there is any fluctuation in the latest monitoring indicator data sequence.

[0036] As can be seen from the above description, in addition to the average value of the trend value, the variance of the residual value, and the growth rate of the variance of the residual value, the variance consistency judgment result of the residual is also introduced. This makes the overall detection method not only use historical data as a reference, but also combine it with the comparison of ideal calibration index data sequence, so as to more comprehensively and accurately judge whether there is fluctuation in the monitoring index data.

[0037] Furthermore, step S4 also includes:

[0038] Establish an indicator calibration model, using the historical monitoring indicator data sequence as the model training set;

[0039] The indicator calibration model is trained using the model training set, and the calibration indicator data sequence is generated using the trained indicator calibration model.

[0040] As described above, the indicator calibration model is used to quickly generate calibration indicator data sequences. At the same time, the indicator calibration model is trained by combining historical monitoring indicator data sequences to improve the accuracy of the data provided by the indicator calibration model and its applicability to the field, so as to make the subsequent judgment results of the variance consistency of the residuals more reliable.

[0041] Further, step S2 includes:

[0042] At least two indicator data blocks are selected from the indicator data subsequence in chronological order;

[0043] Calculate and obtain the variance growth rate V of the residual values ​​of the index data subsequence based on the variance of the residual values ​​of two adjacent index data blocks. std :

[0044] V std =(std cur -std pre ) / t;

[0045] Among them, std pre std represents the variance of the residual values ​​of the previous data block of the aforementioned indicator. cur The variance of the residual value of the next indicator data block is represented by t, and the time interval between the recording times of the indicator data of the two adjacent indicator data blocks is represented by t.

[0046] As can be seen from the above description, when calculating the variance growth rate, the index data blocks contained in the index data subsequence are selected in chronological order. By measuring the variance change of the residual values ​​of the two consecutive index data blocks, the variance growth rate of the index data subsequence can be accurately calculated.

[0047] Furthermore, step S1 also includes:

[0048] A preset sequence length threshold and duration threshold are used to remove the indicator data subsequences whose sequence length is less than or equal to the sequence length threshold and whose sequence duration is less than or equal to the duration threshold.

[0049] As can be seen from the above description, before using the indicator data subsequence, preset sequence length threshold and duration threshold are used as filtering conditions to filter out some indicator data subsequences that are not of reference significance, thereby improving the accuracy of fluctuation judgment.

[0050] Please refer to Figure 6 A monitoring terminal 1 for detecting fluctuations in station indicators includes a memory 2, a processor 3, and a computer program stored in the memory 2 and executable on the processor 3. When the processor 3 executes the computer program, it performs the following steps:

[0051] S1. Obtain the historical monitoring indicator data sequence of the station within a first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order.

[0052] S2. Calculate the average trend value, variance of residual value, and growth rate of variance of residual value for each subsequence of the index data according to the time series decomposition algorithm;

[0053] S3. Based on the average trend value of all the indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values.

[0054] S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations.

[0055] As can be seen from the above description, the beneficial effects of the present invention are as follows: Based on the historical monitoring index data sequence of the station, a reference range for the mean trend value, a reference range for the variance of the residual value, and a reference range for the growth rate of the variance of the residual value are calculated and generated for fluctuation detection. Then, during detection, the latest monitoring index data sequence is obtained, and the average trend value, the variance of the residual value, and the growth rate of the variance of the residual value of the latest monitoring index data sequence are determined according to whether they exceed the corresponding reference range. This enables multi-angle and accurate judgment of whether the monitoring index data of the station has fluctuated, so that operation and maintenance personnel can make timely early warnings and provide reliable protection for the safety of the station.

[0056] Furthermore, step S4 also includes:

[0057] Subtracting the pre-set calibration index data sequence from the latest monitoring index data sequence yields the index residual sequence.

[0058] The residual sequence of the indicator is divided into two residual subsequences according to the time sequence. The variance consistency results of the two residual subsequences are calculated to determine whether there is any fluctuation in the latest monitoring indicator data sequence.

[0059] As can be seen from the above description, in addition to the average value of the trend value, the variance of the residual value, and the growth rate of the variance of the residual value, the consistency judgment result of the residual is also introduced. This makes the overall detection method not only use historical data as a reference, but also combine it with the comparison of the ideal calibration index data sequence, so as to more comprehensively and accurately judge whether there is fluctuation in the monitoring index data.

[0060] Furthermore, step S4 also includes:

[0061] Establish an indicator calibration model, using the historical monitoring indicator data sequence as the model training set;

[0062] The indicator calibration model is trained using the model training set, and the calibration indicator data sequence is generated using the trained indicator calibration model.

[0063] As described above, the indicator calibration model is used to quickly generate calibration indicator data sequences. At the same time, the indicator calibration model is trained by combining historical monitoring indicator data sequences to improve the accuracy of the data provided by the indicator calibration model and its applicability to the field, so as to make the subsequent judgment results of the variance consistency of the residuals more reliable.

[0064] Further, step S2 includes:

[0065] At least two indicator data blocks are selected from the indicator data subsequence in chronological order;

[0066] Calculate and obtain the variance growth rate V of the residual values ​​of the index data subsequence based on the variance of the residual values ​​of two adjacent index data blocks. std :

[0067] V std =(std cur -std pre ) / t;

[0068] Among them, std pre std represents the variance of the residual values ​​of the previous data block of the aforementioned indicator. cur The variance of the residual value of the next indicator data block is represented by t, and the time interval between the recording times of the indicator data of the two adjacent indicator data blocks is represented by t.

[0069] As can be seen from the above description, when calculating the variance growth rate, the index data blocks contained in the index data subsequence are selected in chronological order. By measuring the variance change of the residual values ​​of the two consecutive index data blocks, the variance growth rate of the index data subsequence can be accurately calculated.

[0070] Furthermore, step S1 also includes:

[0071] A preset sequence length threshold and duration threshold are used to remove the indicator data subsequences whose sequence length is less than or equal to the sequence length threshold and whose sequence duration is less than or equal to the duration threshold.

[0072] As can be seen from the above description, before using the indicator data subsequence, preset sequence length threshold and duration threshold are used as filtering conditions to filter out some indicator data subsequences that are not of reference significance, thereby improving the accuracy of fluctuation judgment.

[0073] Please refer to Figures 1 to 4 Embodiment 1 of the present invention is as follows:

[0074] A method for detecting fluctuations in monitoring indicators at a facility includes the following steps:

[0075] S1. Obtain the historical monitoring indicator data sequence of the station within the first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order.

[0076] In this embodiment, the first preset duration can be selected in hours, with a value of M; preferably, M is greater than or equal to 36; when dividing the index data subsequences, a preset time interval can be used as the division basis, and the subsequences are divided according to time order, preferably 2 hours; simultaneously with the division, a preset sequence length threshold and a duration threshold are set, and index data subsequences with a sequence length less than or equal to the sequence length threshold and a sequence duration less than or equal to the duration threshold are removed. Specifically, in conjunction with the aforementioned preferred scheme, the sequence length threshold can be selected from 230 to 250 data points, preferably 240 data points, while the duration threshold can be selected from 1.5 hours to 2.5 hours, preferably 2 hours.

[0077] S2. Calculate the average trend value, variance of residual values, and growth rate of variance of residual values ​​for each index data subsequence based on the time series decomposition algorithm.

[0078] In this embodiment, the preferred time series decomposition algorithm is the STL algorithm. STL (Seasonal-Trend Decomposition Procedure based on Loess): In time series analysis, the STL method is used to decompose time series data into three parts: seasonal, trend, and residual.

[0079] S3. Based on the average trend value of all indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values.

[0080] In this embodiment, specifically within each indicator data subsequence, the average trend value is calculated by sliding and selecting data according to a sliding window length of a third preset duration and a requirement that the number of data points within the window exceeds a preset number. The third preset duration can be selected from 2 to 4 hours, preferably 3 hours, and the preset number of points can be selected from 35 to 45 points, preferably 40 points. In other words, the average trend value corresponding to each indicator data subsequence can be one or more. Then, the average trend values ​​calculated from all sequences are merged into one sequence, and a reference range for the average trend value is calculated based on the GESD algorithm.

[0081] Among them, the GESD algorithm (Generalized Extreme Studentized Deviate) is an existing statistical method used to detect outliers in univariate datasets that follow an approximately normal distribution.

[0082] In each subsequence of indicator data, the data is selected and the variance of the residual value is calculated by sliding and selecting data according to the third preset time period and the number of data in the window is required to exceed the preset number of points. In other words, the variance of the residual value corresponding to each subsequence of indicator data can be one or more. Then, the variances of the residual values ​​calculated from all sequences are merged into one sequence, and the reference range of the residual variance is calculated based on the GESD algorithm.

[0083] In this embodiment, within each index data subsequence, at least two index data blocks are selected by sliding and selecting the data blocks according to the same principle: the sliding window length is a third preset time period and the number of data points within the window exceeds a preset number. The variance growth rate V of the residual values ​​of the index data subsequence is calculated based on the variance of the residual values ​​of two adjacent index data blocks. std :

[0084] V std =(std cur -std pre ) / t;

[0085] Among them, std pre The variance of the residual values ​​of the previous indicator data block, std cur represents the variance of the residual value of the next indicator data block, and t represents the time interval between the recording times of the indicator data of the two adjacent indicator data blocks.

[0086] Finally, the variance growth rate of the residual values ​​calculated from all sequences is merged into one sequence, and the reference range of the variance growth rate of the residual values ​​is calculated based on the GESD algorithm.

[0087] S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations.

[0088] In this embodiment, the second preset duration can be selected in hours, with a value of N; preferably, N is greater than or equal to 4. The trend value of the latest monitoring indicator data sequence is extracted based on the STL algorithm, and then the trend value of the last two hours within the second preset duration is extracted. The average value of the trend value is calculated, and then it is determined whether the average value of the trend value exceeds the reference range of the trend value mean. Whether the variance of the residual value exceeds the reference range of the residual value variance is determined similarly. When determining whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate, the data from the last three hours within the second preset duration is used to calculate the variance values ​​of the three residual values ​​in hours. The growth rate of the variance of the residual values ​​between each pair of adjacent data points is calculated, and the average of the two growth rates is used as the growth rate of the residual value variance of the latest monitoring indicator data sequence to determine whether it exceeds the reference range of the residual value variance growth rate. When any one of the above three out-of-range conditions is met, the latest monitoring indicator data sequence is output as having fluctuations.

[0089] Please refer to Figure 5 Embodiment two of the present invention is as follows:

[0090] A method for detecting fluctuations in station monitoring indicators, based on the above embodiment one, further includes step S4 to improve the comprehensiveness and accuracy of fluctuation judgment: extracting a specified number of original indicator sequences from the latest monitoring indicator data sequence, subtracting a pre-set calibration indicator data sequence from the original indicator sequence to obtain an indicator residual sequence, wherein the specified number is preferably 60; dividing the indicator residual sequence into two residual subsequences according to time order, calculating and judging whether there is fluctuation in the latest monitoring indicator data sequence based on the variance consistency results of the two residual subsequences.

[0091] In this embodiment, an indicator calibration model is established, using historical monitoring indicator data sequences as the model training set. The training set is used to train the indicator calibration model, which then generates calibration indicator data sequences. The historical monitoring indicator data sequences are segmented into multiple subsequences with consecutive time points and a length of 60 points. The model features are the station ID (did) and the subsequence his, with the training objective being the subsequence his. The input and output are kept consistent, forming the model training set. During training, the following methods are primarily used: Figure 5 The training of the convolutional model shown is explained below:

[0092] Convolutional Models: Convolutional models are a type of deep learning neural network architecture primarily used for tasks involving data with a grid structure, such as image and audio processing. Its core component is the convolutional layer, which uses filters (also called convolutional kernels) to perform convolution operations on the input data.

[0093] Embedding: In computer science, embedding refers to the technique of mapping high-dimensional data to a low-dimensional space. In Natural Language Processing (NLP), embedding usually refers to word embedding, which maps words to low-dimensional dense vectors in a real-valued vector space. These vectors contain semantic information and can capture the semantic relationships between words.

[0094] Concatenate: In deep learning, "concatenate" usually refers to connecting two or more tensors together along a certain dimension.

[0095] Conv1d: Conv1D is a one-dimensional convolutional neural network (CNN) layer, typically used to process data with a sequential structure, such as time series data, text data, or audio data.

[0096] In deep learning, "Dense" refers to a fully connected layer, also known as a densely connected layer. Each neuron in a fully connected layer is connected to all neurons in the layer above it, and each connection has a weight. This means that each neuron in a fully connected layer receives the outputs of all neurons in the previous layer, which are then weighted and summed before being processed by an activation function.

[0097] Timedistributed: In deep learning, "TimeDistributed" is a technique for processing sequential data, often used in conjunction with recurrent neural networks (RNNs) or convolutional neural networks (CNNs). This technique applies a neural network layer to each time step of the sequential data to enable parallel processing of the entire sequence.

[0098] In this embodiment, the variance analysis of the two residual subsequences is specifically verified based on the Levene test, and the p-value of the test is obtained as the variance consistency result. The Levene test is a statistical method used to test the homogeneity of variance among multiple data samples. Its basic idea is to determine whether the differences between the data groups are significant by comparing their variances. In this judgment, the p-value is checked to see if it is less than 0.00001. If it is, it indicates abnormal variance consistency, and the latest monitoring indicator data series shows fluctuations; if it is not, it indicates normal variance consistency, and the latest monitoring indicator data series does not show fluctuations.

[0099] Please refer to Figure 6 Embodiment two of the present invention is as follows:

[0100] A station monitoring index fluctuation detection terminal 1 includes a memory 2, a processor 3, and a computer program stored in the memory 2 and capable of running on the processor 3. When the processor 3 executes the computer program, it implements a station monitoring index fluctuation detection method according to Embodiment 1 or 2.

[0101] In summary, the present invention provides a method and terminal for detecting fluctuations in station monitoring indicators. Based on the historical monitoring indicator data sequence of the station, it calculates and generates reference ranges for the mean trend value, variance of residual values, and growth rate of residual variance for fluctuation detection. Then, during detection, it acquires the latest monitoring indicator data sequence and checks whether the mean trend value, variance of residual values, and growth rate of residual variance of the latest monitoring indicator data sequence exceed the corresponding reference ranges. Simultaneously, it introduces a calibration indicator data sequence to detect the variance consistency of the residual values ​​of the latest monitoring indicator data sequence. This enables multi-angle and accurate judgment of whether the station's monitoring indicator data has fluctuated, allowing maintenance personnel to make timely early warnings and provide reliable protection for station safety.

[0102] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention's specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for detecting fluctuations in station monitoring indicators, characterized in that, Includes the following steps: S1. Obtain the historical monitoring indicator data sequence of the station within a first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order. S2. Calculate the average trend value, variance of residual value, and growth rate of variance of residual value for each subsequence of the index data according to the time series decomposition algorithm; S3. Based on the average trend value of all the indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values. S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations. Step S2 includes: At least two indicator data blocks are selected from the indicator data subsequence in chronological order; Calculate and obtain the variance growth rate V of the residual values ​​of the index data subsequence based on the variance of the residual values ​​of two adjacent index data blocks. std : V std =(std cur -std pre ) / t; Among them, std pre std represents the variance of the residual values ​​of the previous data block of the aforementioned indicator. cur The variance of the residual value of the next indicator data block is represented by t, and the time interval between the recording times of the indicator data of the two adjacent indicator data blocks is represented by t.

2. The method for detecting fluctuations in station monitoring indicators according to claim 1, characterized in that, Step S4 further includes: Subtracting the pre-set calibration index data sequence from the latest monitoring index data sequence yields the index residual sequence. The residual sequence of the indicator is divided into two residual subsequences according to the time sequence. The variance consistency results of the two residual subsequences are calculated to determine whether there is any fluctuation in the latest monitoring indicator data sequence.

3. The method for detecting fluctuations in station monitoring indicators according to claim 2, characterized in that, Step S4 further includes: Establish an indicator calibration model, using the historical monitoring indicator data sequence as the model training set; The indicator calibration model is trained using the model training set, and the calibration indicator data sequence is generated using the trained indicator calibration model.

4. The method for detecting fluctuations in station monitoring indicators according to claim 1, characterized in that, Step S1 further includes: A preset sequence length threshold and duration threshold are used to remove the indicator data subsequences whose sequence length is less than or equal to the sequence length threshold and whose sequence duration is less than or equal to the duration threshold.

5. A terminal for detecting fluctuations in monitoring indicators at a monitoring station, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: S1. Obtain the historical monitoring indicator data sequence of the station within a first preset time period, and divide the historical monitoring indicator data sequence into a preset number of indicator data subsequences according to the time order. S2. Calculate the average trend value, variance of residual value, and growth rate of variance of residual value for each subsequence of the index data according to the time series decomposition algorithm; S3. Based on the average trend value of all the indicator data subsequences, obtain the reference range of the mean trend value; based on the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the variance of the residual values; based on the growth rate of the variance of the residual values ​​of all the indicator data subsequences, obtain the reference range of the growth rate of the variance of the residual values. S4. Obtain the latest monitoring indicator data sequence of the station from the current time to the second preset time period, and determine whether the average value of the trend value of the latest monitoring indicator data sequence exceeds the reference range of the trend value mean, or whether the variance of the residual value exceeds the reference range of the residual value variance, or whether the growth rate of the residual value variance exceeds the reference range of the residual value variance growth rate. If so, output that the latest monitoring indicator data sequence has fluctuations. Step S2 includes: At least two indicator data blocks are selected from the indicator data subsequence in chronological order; Calculate and obtain the variance growth rate V of the residual values ​​of the index data subsequence based on the variance of the residual values ​​of two adjacent index data blocks. std : V std =(std cur -std pre ) / t; Among them, std pre std represents the variance of the residual values ​​of the previous data block of the aforementioned indicator. cur The variance of the residual value of the next indicator data block is represented by t, and the time interval between the recording times of the indicator data of the two adjacent indicator data blocks is represented by t.

6. A station monitoring indicator fluctuation detection terminal according to claim 5, characterized in that, Step S4 further includes: Subtracting the pre-set calibration index data sequence from the latest monitoring index data sequence yields the index residual sequence. The residual sequence of the indicator is divided into two residual subsequences according to the time sequence. The variance consistency results of the two residual subsequences are calculated to determine whether there is any fluctuation in the latest monitoring indicator data sequence.

7. A station monitoring indicator fluctuation detection terminal according to claim 6, characterized in that, Step S4 further includes: Establish an indicator calibration model, using the historical monitoring indicator data sequence as the model training set; The indicator calibration model is trained using the model training set, and the calibration indicator data sequence is generated using the trained indicator calibration model.

8. A station monitoring indicator fluctuation detection terminal according to claim 5, characterized in that, Step S1 further includes: A preset sequence length threshold and duration threshold are used to remove the indicator data subsequences whose sequence length is less than or equal to the sequence length threshold and whose sequence duration is less than or equal to the duration threshold.

Images