A method for analyzing abnormal operation of microgrid system

Abstract

This application discloses a method for analyzing operational anomalies in a microgrid system, comprising the following steps: obtaining operational data of the microgrid system and removing noise; dividing the operational data into several segments and generating graphs, performing similarity comparisons on each graph to obtain similarity relationships; hierarchically classifying the operational data based on the similarity relationships; based on the hierarchical classification results and the segment division results, conducting an overall evaluation of the operational data within the preset time period and determining whether operational anomalies exist; and performing cluster analysis on all segments to determine whether any nodes have operational anomalies. This application denoises the data of the microgrid system before dividing and visualizing it. By analyzing similarities to quickly troubleshoot anomalies, the application can effectively improve data analysis efficiency. The analysis results can then be used to optimize and improve the efficiency of the microgrid system's electrical energy conversion and scheduling, enabling better application in power systems.

Description

Technical Field

[0001] The present application relates to the field of power operation technology, and in particular to a method for analyzing operation anomalies of a microgrid system. Background Art

[0002] Power imbalances in microgrid power systems and fluctuations in the grid itself can lead to instability in the microgrid's power supply system, impacting the safe operation of microgrid equipment and, in severe cases, damaging equipment and causing economic losses. Energy storage systems, however, can store energy and distribute it appropriately when needed, thereby reducing the impact of microgrid equipment on the grid, improving the quality of microgrid power supply systems, ensuring the safe and reliable operation of equipment, and enhancing the reliability and safety of the power system.

[0003] Therefore, microgrids are becoming an increasingly important component of power systems. By storing energy during low-load periods and releasing it during peak load periods, microgrids can effectively reduce the grid's energy demand during peak load periods, improving the environmental and economic benefits of the power system. Furthermore, microgrids can provide grid frequency control. When the grid frequency experiences anomalies, microgrids can coordinate energy storage and release with the main grid, contributing to the smooth operation of the power system.

[0004] Currently, microgrid operation involves monitoring multiple data points, such as the capacity, status, and output power of each energy storage plant. This data can be used to assess the health and performance of the microgrid. However, the sheer volume of collected data makes comprehensive analysis inconvenient. Existing analysis methods rely solely on simple data classification and filtering to obtain the desired results. This results in slow processing and difficulty in analyzing large amounts of data.

[0005] Therefore, a method for analyzing the operation anomaly of a microgrid system is needed to solve a large number of data processing problems during the operation analysis process, denoise the data, and analyze the denoised data by setting a hierarchy. This can improve the efficiency and accuracy of the analysis, thereby improving the stability of the microgrid system. Summary of the Invention

[0006] This application mainly provides a method for analyzing the operation anomaly of a microgrid system to solve the problems of slow data analysis speed and low efficiency during the operation of the microgrid system.

[0007] To solve the above technical problems, a technical solution adopted by the present application is to provide a method for analyzing abnormal operation of a microgrid system, comprising the following steps:

[0008] S10: Acquire the operating data of the microgrid system and remove noise;

[0009] S20: Dividing the operation data into a plurality of sections, generating charts based on the operation data of different sections, and performing similarity comparison on the charts to obtain a similarity relationship;

[0010] S30: hierarchically classifying the operation data within a preset time period according to the similarity relationship;

[0011] S40: Based on the hierarchical classification result and the segment division result, an overall evaluation is performed on the operation data within the preset time, and it is determined whether there is an operation abnormality within the preset time;

[0012] S50: Perform cluster analysis on all the segments to determine whether any node has operational anomalies.

[0013] In a possible implementation manner, after the step of obtaining the operating data of the microgrid system and performing noise removal, the method further includes:

[0014] S11: Perform integrity check on the operating data, identify and mark missing values;

[0015] S12: Filling or re-collecting the operating data based on the missing values.

[0016] In a possible implementation manner, after the step of obtaining the operating data of the microgrid system and performing noise removal, the method further includes:

[0017] S13: removing abnormal data based on the mean and standard deviation of the operating data.

[0018] In one possible implementation, the step of dividing the operating data into a plurality of segments, generating charts based on the operating data of different segments, and performing similarity comparison on the charts to obtain a similarity relationship includes:

[0019] S21: Dividing the operation data into sections based on preset division parameters;

[0020] S22: Selecting different types of charts based on the characteristics of the division parameters.

[0021] In a possible implementation, the step of dividing the operating data into a plurality of segments, generating charts based on the operating data of different segments, and performing similarity comparison on the charts to obtain similarity relationships further includes:

[0022] S23: Extract multi-dimensional features, construct feature vectors, and perform similarity comparison based on the feature vectors.

[0023] In a possible implementation, after the steps of dividing the operating data into a plurality of segments, generating charts based on the operating data of different segments, and performing similarity comparison on the charts to obtain similarity relationships, the method further includes:

[0024] S24: Based on analysis requirements and operational data characteristics, adjust the width of the segment division.

[0025] In a possible implementation, the step of hierarchically classifying the operating data within a preset time period according to the similarity relationship includes:

[0026] S31: Classifying the operation data into multiple levels based on a similarity threshold.

[0027] In a possible implementation, the step of performing cluster analysis on all the segments to determine whether a node has an operational abnormality includes:

[0028] S51: performing cluster analysis on the segment based on the parent node in the hierarchical classification as the initial cluster center;

[0029] S52: Perform secondary cluster analysis based on multi-dimensional features.

[0030] In a possible implementation, after the step of performing cluster analysis on all the segments to determine whether a node has an operational abnormality, the following steps may be performed:

[0031] S60: Establish a historical anomaly database for storing anomaly data from past generations;

[0032] S61: performing a similarity comparison between the operating data and the abnormal data, and determining whether the operating data is abnormal.

[0033] In a possible implementation, after the step of performing cluster analysis on all the segments to determine whether a node has an operational abnormality, the following steps may be performed:

[0034] S70: Build an evaluation system based on operating conditions, similarity comparison and historical data.

[0035] The beneficial effects of the present application are as follows: Different from the prior art, the present application discloses a method for analyzing the operation anomalies of a microgrid system, which denoises the data of the microgrid and then divides and images it, and quickly detects anomalies by analyzing similarities, which can effectively improve the efficiency of data analysis. The analysis results can then be used to optimize and improve the efficiency of the microgrid system's power conversion and scheduling, so that it can be better used in the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. Those skilled in the art can also derive other drawings based on these drawings without inventive work, among which:

[0037] Figure 1 It is a flowchart of a method for analyzing abnormal operation of a microgrid system in one embodiment of the present application. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0039] The terms "first", "second" and "third" in the embodiments of the present application are only used for descriptive purposes and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second" and "third" may explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally also include steps or units that are not listed, or may optionally also include other steps or units inherent to these processes, methods, products or devices.

[0040] References to "embodiments" herein mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0041] See also Figure 1 , the embodiment of the present application includes: a method for analyzing abnormal operation of a microgrid system, comprising the steps of:

[0042] S10: Acquire the operating data of the microgrid system and remove noise;

[0043] S20: Divide the operation data into several sections, generate charts based on the operation data of different sections, perform similarity comparison on each chart, and obtain a similarity relationship;

[0044] S30: Classifying the operating data within a preset time period into hierarchical categories based on similarity relationships;

[0045] S40: Based on the hierarchical classification results and the segment division results, an overall evaluation is performed on the operation data within the preset time, and it is determined whether there is any operation abnormality within the preset time;

[0046] S50: Perform cluster analysis on all segments to determine whether any node has operational abnormalities.

[0047] Specifically, in step S10, operational data refers to the microgrid's operational status, parameters, and related parameters of its key equipment, such as the microgrid's output power, active power, the charge and discharge status of each energy storage station, charge and discharge power, battery status, and response time. Operational data can be collected and processed using various monitoring devices and sampling devices, and noise removal is performed. Noise removal involves removing significantly abnormal data from the collected operational data. This can be done by using specific metrics (such as signal-to-noise ratio and error) or by employing various filtering algorithms.

[0048] In step S20, the operating data is segmented based on time or status. The time can be hourly, daily, etc., and the status can be charging, discharging, or standby. Segmentation is performed to more accurately analyze data features.

[0049] The choice of chart can be based on the data type, such as using curve charts and waveform charts for continuous data, and scatter plots and matrices for state data. Similarity relationships are then obtained for the charts through methods such as graphic comparison, distance calculation, and dynamic time warping.

[0050] In step S30, thresholds are set for the similarity relationships, and hierarchical classification is performed based on the preset thresholds. This hierarchical classification can include multiple levels, such as a preliminary classification into normal and abnormal levels. The normal level can be further subdivided into stable operation and minor fluctuations, and the abnormal level can be further subdivided into mild and severe abnormalities. Specific classifications can be based on different thresholds or combined with special factors (such as specific equipment abnormalities and the status of each energy storage plant).

[0051] In S40, a comprehensive assessment system can be constructed, combining hierarchical classification results, similarity, historical data, and other factors for comprehensive evaluation. For example, the proportion of abnormal sections and similarity trend can be calculated to evaluate the overall operation of the microgrid system. If all sections exceeding the threshold are abnormal within a certain time period, a warning should be issued and further investigation should be carried out.

[0052] In S50, clustering algorithms are applied to different types of data. For example, K-means clustering can be used for steady-state data, with a K value of 3, corresponding to the three modes of charge, discharge, and standby. For transient data, DBSCAN clustering combined with outlier detection can be used for identification. Specifically, if a segment in the clustering results is more than twice the standard deviation away from the cluster center, the segment is identified as anomalous data.

[0053] In one embodiment, after the steps of obtaining the operating data of the microgrid system and performing noise removal, the method further includes:

[0054] S11: Perform integrity check on the running data, identify and mark missing values;

[0055] S12: Fill in or recollect running data based on missing values.

[0056] Specifically, targeted processing is performed based on the results of the integrity check. If there are few missing values, linear interpolation or spline interpolation can be used to fill them in. If there are many missing values, re-collection, data cropping, and overall estimation can be considered.

[0057] In one embodiment, after the steps of obtaining the operating data of the microgrid system and performing noise removal, the method further includes:

[0058] S13: Remove abnormal data based on the mean and standard deviation of the running data.

[0059] The mean and standard deviation of the operating data are calculated. Data that differs from the mean by more than a certain multiple (e.g., 2 in this example) is considered abnormal and removed. The threshold setting can be determined based on the operating conditions. For example, during stable operation, where fluctuations are small, the threshold can be lower, while during startup and shutdown, where fluctuations are large, the threshold can be higher.

[0060] In one embodiment, the steps of dividing the operating data into a plurality of segments, generating charts based on the operating data of different segments, and performing similarity comparison on the charts to obtain similarity relationships include:

[0061] S21: segmenting the operating data based on preset segmentation parameters;

[0062] S22: Select different types of charts based on the characteristics of the partitioning parameters.

[0063] Specifically, the segmentation parameters can include time or state. Time segmentation can be hourly, several hours, or daily, and state segmentation can be charging, discharging, or standby. After segmentation, the data within each segment is graphed. The graph selection can be determined based on the segmentation parameters and the characteristics of the data. For example, continuous data can use a curve graph or waveform graph, while state data can use a scatter plot or matrix.

[0064] In one embodiment, the step of dividing the operating data into a plurality of segments, generating charts based on the operating data of different segments, and performing similarity comparison on the charts to obtain similarity relationships further includes:

[0065] S23: Extract multi-dimensional features, construct feature vectors, and perform similarity comparison based on the feature vectors.

[0066] Specifically, a set of comprehensive similarity indicators is integrated through weight distribution to improve the accuracy and stability of similarity calculation.

[0067] First, multi-dimensional features such as time domain, frequency domain, and morphology are extracted to construct feature vectors. Then, each feature is standardized to eliminate the dimension. Weights are assigned to each feature using the entropy weight method or empirical method, and the final similarity is obtained through linear weighted fusion.

[0068] In this embodiment, time domain features, frequency domain features and morphological features are extracted in sequence. Time domain features include mean, variance, peak, skewness, kurtosis, etc. Frequency domain features include main frequency amplitude, total harmonic distortion (THD), energy proportion, etc. Morphological features include extreme point density, waveform slope, envelope area, etc. The above features are normalized by Z-score to eliminate dimensional differences, and then the entropy weight method is used to calculate the first j The entropy E of the features j ,

[0069]

[0070] in i It is i samples, n is the number of all samples, P ij After standardization i The sample in j The normalized probability value on the features. Then calculate j The weight of the feature w j ,

[0071]

[0072] The denominator of the fraction refers to the sum of the information utility values ​​of all features. m is the number of all features.

[0073] After obtaining the feature weights of all samples, linear weighted fusion is performed to obtain similarity. Alternatively, the final similarity can be obtained through nonlinear fusion using the attention mechanism in the neural network.

[0074] In one embodiment, after dividing the operating data into a plurality of segments, generating charts based on the operating data of different segments, and performing similarity comparison on the charts to obtain similarity relationships, the method further includes:

[0075] S24: Adjust the width of the segment division based on analysis requirements and operating data characteristics.

[0076] The width of the segment divisions can be flexible. For example, if higher accuracy is required for analysis, the segment width can be reduced, or a larger width can be used for non-critical segments to speed up calculations. Data characteristics can include the degree of data fluctuation and its correlation with time. For example, data with low fluctuations can be divided into larger segment widths, while data that fluctuates only at specific times can be divided into smaller segment widths at that time.

[0077] In one embodiment, the step of hierarchically classifying the operating data within a preset time period according to the similarity relationship includes:

[0078] S31: Classify the operation data into multiple levels based on a similarity threshold.

[0079] Specifically, similarity thresholds are set to perform hierarchical classification. In this embodiment, the hierarchical classification consists of two levels: the first level corresponds to a coarse classification, and the second level corresponds to a finer classification. The first level includes two categories: normal and abnormal. The normal category is further subdivided into stable operation and small fluctuations, and the abnormal category is further subdivided into minor anomalies and severe anomalies. The hierarchical classification is not fixed and can be fine-tuned by adjusting the thresholds based on needs and specific data conditions.

[0080] In one embodiment, the step of performing cluster analysis on all segments to determine whether a node has an operational anomaly includes:

[0081] S51: Perform cluster analysis on the segments based on the parent node in the hierarchical classification as the initial cluster center;

[0082] S52: Perform secondary cluster analysis based on multi-dimensional features.

[0083] Specifically, we first perform coarse K-means clustering on the child nodes under each parent node. The initial number of clusters is determined based on the hierarchical tree, enabling rapid initial screening and narrowing the scope of detection. For the subsets after the coarse clustering, we perform secondary clustering in the time and frequency domains, using DBSCAN density clustering to identify differences in waveform morphology.

[0084] In one embodiment, after performing cluster analysis on all segments to determine whether a node has an operational anomaly, the following steps are included:

[0085] S60: Establish a historical anomaly database for storing anomaly data from past generations;

[0086] S61: performing a similarity comparison between the operating data and the abnormal data, and determining whether there is an abnormality in the operating data.

[0087] Specifically, cluster analysis of historical anomaly data allows for rapid determination of data anomalies. The anomaly database also facilitates subsequent data backtracking and verification, enabling rapid identification of anomaly causes and troubleshooting.

[0088] In one embodiment, after performing cluster analysis on all segments to determine whether a node has an operational anomaly, the following steps are included:

[0089] S70: Build an evaluation system based on operating conditions, similarity comparison and historical data.

[0090] Specifically, a weight matrix can be designed to differentiate the importance of different operating modes, such as increasing the weight of abnormalities during discharge and decreasing the weight of abnormalities during standby. An attenuation factor can be introduced to make recent abnormal data more influential and earlier abnormalities less influential.

[0091] Different from the existing technology, this embodiment can more efficiently and accurately calculate and analyze the operating status of the microgrid, obtain the cause of abnormalities, and implement real-time adjustments based on various factors, with higher adaptability and robustness.

[0092] The above description is merely an embodiment of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for analyzing abnormal operation of a microgrid system, characterized in that: Including steps: S10: Acquire the operating data of the microgrid system and remove noise; S20: Divide the operation data into several sections, generate charts based on the operation data of different sections, perform similarity comparison on the charts, and obtain similarity relationships, wherein the operation data is divided into sections based on the state, and different types of charts are selected based on the characteristics of the state; S30: hierarchically classifying the operation data within a preset time period according to the similarity relationship; S40: Based on the hierarchical classification result and the segment division result, an overall evaluation is performed on the operation data within the preset time, and it is determined whether there is an operation abnormality within the preset time; S50: performing cluster analysis on all the segments to determine whether any node has an operational anomaly; The similarity comparison step further comprises: S23: Extract multi-dimensional features of time domain, frequency and morphology, construct feature vectors, standardize and assign weights to each feature, and perform similarity comparison based on the feature vectors; wherein the step of assigning weights is performed by using the formula Get the entropy E of the i-th sample in the j-th feature j , n is the number of all samples, P ij After standardization i The sample in j The normalized probability value on the feature, and then use the formula calculate j The weight of the feature w j , m is the number of all features, and the weights of all the features are obtained, and then linear weighted fusion is performed to calculate the similarity.

2. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: After the steps of obtaining the operating data of the microgrid system and performing noise removal, the method further includes: S11: Perform integrity check on the operating data, identify and mark missing values; S12: Filling or re-collecting the operating data based on the missing values.

3. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: After the steps of obtaining the operating data of the microgrid system and performing noise removal, the method further includes: S13: removing abnormal data based on the mean and standard deviation of the operating data.

4. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: After the steps of dividing the operation data into a plurality of sections, generating charts based on the operation data of different sections, and performing similarity comparison on the charts to obtain similarity relationships, the method further includes: S24: Based on analysis requirements and operational data characteristics, adjust the width of the segment division.

5. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: The step of hierarchically classifying the operating data within a preset time period according to the similarity relationship includes: S31: Classifying the operation data into multiple levels based on a similarity threshold.

6. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: The step of performing cluster analysis on all the segments to determine whether a node has an operational abnormality includes: S51: performing cluster analysis on the segment based on the parent node in the hierarchical classification as the initial cluster center; S52: Perform secondary cluster analysis based on multi-dimensional features.

7. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: After the step of performing cluster analysis on all the segments and determining whether a node has an operational abnormality, the method further includes: S60: Establish a historical anomaly database for storing anomaly data from past generations; S61: performing a similarity comparison between the operating data and the abnormal data, and determining whether the operating data is abnormal.

8. The method for analyzing abnormal operation of a microgrid system according to claim 1, characterized in that: After the step of performing cluster analysis on all the segments and determining whether a node has an operational abnormality, the method further includes: S70: Build an evaluation system based on operating conditions, similarity comparison and historical data.

Images