A device monitoring method and system based on chromaticity map mapping

Abstract

The present invention relates to the technical field of GIS device monitoring, and in particular to a device monitoring method and system based on chromaticity map mapping. The method includes the following steps: collecting real-time vibration information of the GIS device, and performing time-frequency conversion analysis on the real-time vibration information to obtain a number of real-time sub-vibration characteristics; analyzing each real-time sub-vibration characteristic, and converting and generating a real-time chromaticity map based on the analysis results; analyzing the real-time chromaticity map based on the first operating state analysis model to determine the real-time device component state of the GIS device. By analyzing the information of the real-time vibration, a variety of parameter information is configured in the real-time chromaticity map, and the real-time chromaticity map is used as input data, and the corresponding real-time device component state is determined by the first operating state analysis model. The present invention improves the accuracy and comprehensiveness of component operating state recognition by performing vibration time-frequency conversion and chromaticity map mapping on the GIS device.

Description

Technical Field

[0001] The present invention relates to the technical field of GIS equipment monitoring, and in particular to a device monitoring method and system based on chromaticity map mapping. Background Art

[0002] Gas insulated switchgear (GIS) in substations has been widely used in the power field due to its advantages such as small floor area, high system integration, and strong operation safety and reliability. According to relevant data statistics, 44% of GIS equipment failures are caused by mechanical defects. Therefore, in order to ensure the safe operation of GIS, it is of great significance to carry out research on on-line monitoring and intelligent diagnosis technologies for potential mechanical defects of GIS. However, since the vibration signals of GIS contain rich state information of the equipment and also have the advantage of being free from electromagnetic interference, good test results have been obtained in identifying the mechanical state of GIS based on vibration signal analysis. However, conventional feature extraction methods based on vibration signals have good identification effects for monitoring objects with simple structures and single working modes, but for GIS equipment with large and complex structures and diverse types of mechanical defects, it is usually difficult to effectively characterize different states of GIS, which may cause missed and misjudged defects, and the safety of conventional contact-based vibration monitoring methods is relatively low. Summary of the Invention

[0003] Based on this, it is necessary to provide a device monitoring method and system based on chromaticity map mapping to solve at least one of the above technical problems.

[0004] To achieve the above object, a device monitoring method based on chromaticity map mapping, the method includes the following steps:

[0005] Step S1: Collect the real-time vibration information of the GIS equipment, and perform time-frequency conversion analysis on the real-time vibration information by using complex Gabor-Morlet wavelet transform, and obtain a real-time sub-vibration feature set;

[0006] Step S2: Perform single feature analysis on the real-time sub-vibration feature set to generate a single feature analysis result; perform chromaticity map mapping on the real-time sub-vibration feature set according to the single feature analysis result to generate a real-time chromaticity map;

[0007] Step S3: Construct a first operating state analysis model; perform component state analysis on the real-time chromaticity map based on the first operating state analysis model to obtain real-time device component state data of the GIS equipment.

[0008] The present invention can obtain the characteristics of vibration signals in terms of time and frequency by collecting the real-time vibration information of GIS devices and performing time-frequency conversion analysis using complex Gabor-Morlet wavelet transform. Such analysis helps to understand the vibration patterns, frequency components, and change trends of the devices, providing basic data for subsequent condition assessment. By performing time-frequency conversion analysis on the real-time vibration information, a real-time sub-vibration feature set can be obtained. These feature sets contain information on the vibration components within different frequency ranges, making further analysis and determination more comprehensive and accurate. By performing single-feature analysis on the real-time sub-vibration feature set, key vibration feature parameters such as peak value, frequency, amplitude, etc. can be extracted. The results of these single-feature analyses are used to evaluate the operating state of the device and provide a basis for subsequent condition determination. According to the results of the single-feature analysis, chromaticity map mapping is performed on the real-time sub-vibration feature set, which can visually display the vibration characteristics. The chromaticity map can intuitively show the intensity and frequency distribution of the vibration, facilitating the rapid observation and analysis of the vibration state of the device. According to the analysis results and feature extraction from the first to the fourth steps, a first operating state analysis model can be constructed. This model is based on machine learning, deep learning, or other related technologies and is used to determine the operating state of the device and identify abnormal conditions. Based on the first operating state analysis model, component state analysis is performed on the real-time chromaticity map, which can correspond the vibration characteristics to the different component states of the device, thereby obtaining the real-time device component state data of the GIS device. These data can be used for device health monitoring, fault prediction, and maintenance decision-making, etc. Therefore, the present invention improves the accuracy and comprehensiveness of component operating state recognition by performing vibration time-frequency conversion and chromaticity map mapping on GIS devices.

[0009] In this specification, a device monitoring system based on chromaticity map mapping is provided for performing the device monitoring method based on chromaticity map mapping described above. The device monitoring system based on chromaticity map mapping includes:

[0010] A vibration information analysis module, configured to collect the real-time vibration information of GIS devices, perform time-frequency conversion analysis on the real-time vibration information using complex Gabor-Morlet wavelet transform, and perform time-frequency conversion analysis on the real-time vibration information to obtain a real-time sub-vibration feature set;

[0011] A chromaticity map generation module, configured to perform single-feature analysis on the real-time sub-vibration feature set to generate single-feature analysis results; perform chromaticity map mapping on the real-time sub-vibration feature set according to the single-feature analysis results to generate a real-time chromaticity map;

[0012] An operating state analysis module, configured to construct a first operating state analysis model; perform component state analysis on the real-time chromaticity map based on the first operating state analysis model to obtain the real-time device component state data of the GIS device.

[0013] The beneficial effects of the present invention are as follows: By collecting the real-time vibration information of GIS equipment and using complex Gabor-Morlet wavelet transform for time-frequency conversion analysis, efficient processing and analysis of vibration signals can be achieved. This helps to promptly detect abnormal vibration conditions of the equipment and give early warnings of possible faults. Conducting single-feature analysis on the real-time sub-vibration feature set and generating a chromaticity map can visually display the vibration characteristics and state changes of the equipment. This helps engineers and operators more quickly understand the operating state of the equipment for fault diagnosis and maintenance. By constructing the first operating state analysis model, systematic component state analysis of the real-time chromaticity map can be carried out to further extract the real-time equipment component state data of the equipment. This helps to establish an equipment state monitoring and prediction model to achieve intelligent management and optimized maintenance of the equipment state. Based on the above analysis results, real-time monitoring and analysis of the operating state of GIS equipment can be realized, promptly detecting abnormal states and potential faults of the equipment, thereby giving early warnings and taking corresponding maintenance measures to reduce the equipment failure rate and extend the equipment life. By implementing the above analysis and prediction models, precise maintenance of GIS equipment can be achieved, avoiding unnecessary maintenance and downtime, reducing maintenance costs, and improving the reliability and stability of the equipment. Brief Description of the Drawings

[0014] Figure 1 It is a schematic diagram of the step flow of a device monitoring method based on chromaticity map mapping;

[0015] Figure 2 is Figure 1 A detailed implementation step flow schematic diagram of chromaticity map mapping of the real-time sub-vibration feature set according to the single-feature analysis result in step S2;

[0016] Figure 3 is Figure 1 A detailed implementation step flow schematic diagram of constructing the first operating state analysis model in step S3 in;

[0017] The realization, functional characteristics and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the drawings. Detailed Embodiments

[0018] The technical methods of the present invention patent will be clearly and completely described below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0019] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus the repeated description thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0020] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0021] To achieve the above object, please refer to Figures 1 to 3 , a device monitoring method based on chromaticity map mapping, the method comprising the following steps:

[0022] Step S1: Collect the real-time vibration information of the GIS device, and perform time-frequency conversion analysis on the real-time vibration information by using the complex Gabor-Morlet wavelet transform, and perform time-frequency conversion analysis on the real-time vibration information to obtain a real-time sub-vibration feature set;

[0023] Step S2: Perform single feature analysis on the real-time sub-vibration feature set to generate a single feature analysis result; perform chromaticity map mapping on the real-time sub-vibration feature set according to the single feature analysis result to generate a real-time chromaticity map;

[0024] Step S3: Construct a first operating state analysis model; perform component state analysis on the real-time chromaticity map based on the first operating state analysis model to obtain real-time device component state data of the GIS device.

[0025] The present invention can obtain the characteristics of vibration signals in time and frequency by collecting the real-time vibration information of GIS devices and performing time-frequency conversion analysis using complex Gabor-Morlet wavelet transform. Such analysis helps to understand the vibration patterns, frequency components, and change trends of the devices, providing basic data for subsequent condition assessment. By performing time-frequency conversion analysis on the real-time vibration information, a real-time sub-vibration feature set can be obtained. These feature sets contain information on vibration components in different frequency ranges, making further analysis and determination more comprehensive and accurate. By performing single-feature analysis on the real-time sub-vibration feature set, key vibration feature parameters such as peak value, frequency, amplitude, etc. can be extracted. The results of these single-feature analyses are used to evaluate the operating state of the devices and provide a basis for subsequent condition determination. According to the results of the single-feature analysis, chromaticity map mapping is performed on the real-time sub-vibration feature set, which can visually display the vibration characteristics. The chromaticity map can intuitively show the intensity and frequency distribution of the vibration, helping to quickly observe and analyze the vibration state of the devices. According to the analysis results and feature extraction from the first to the fourth steps, a first operating state analysis model can be constructed. This model is based on machine learning, deep learning, or other related technologies and is used to determine the operating state of the devices and identify abnormal situations. Based on the first operating state analysis model, component state analysis is performed on the real-time chromaticity map, which can correspond the vibration characteristics to different component states of the devices, thereby obtaining the real-time device component state data of the GIS devices. These data can be used for device health monitoring, fault prediction, and maintenance decision-making, etc. Therefore, the present invention improves the accuracy and comprehensiveness of component operating state recognition by performing vibration time-frequency conversion and chromaticity map mapping on GIS devices.

[0026] In an embodiment of the present invention, referring to Figure 1 as described, it is a schematic diagram of the step flow of a device monitoring method based on chromaticity map mapping according to the present invention. In this example, the device monitoring method based on chromaticity map mapping includes the following steps:

[0027] Step S1: Collect the real-time vibration information of GIS devices, and perform time-frequency conversion analysis on the real-time vibration information using complex Gabor-Morlet wavelet transform, and perform time-frequency conversion analysis on the real-time vibration information to obtain a real-time sub-vibration feature set;

[0028] In the embodiments of the present invention, by using sensor devices such as vibration sensors or accelerometers, which are installed at key parts of the GIS device, the vibration information of the GIS device is collected in real time. The collection frequency needs to be high enough to ensure real-time monitoring and capture of the device vibration. The collected real-time vibration information is subjected to time-frequency conversion analysis using the complex Gabor-Morlet wavelet transform. The Gabor-Morlet wavelet transform is a commonly used time-frequency analysis method that can provide information in both the time and frequency domains and is suitable for the analysis of non-stationary signals. After the time-frequency conversion analysis, vibration characteristics such as frequency and amplitude are extracted from the obtained time-frequency image. According to the distribution and change trend of the vibration characteristics, a real-time sub-vibration characteristic set is constructed. More specifically, the vibration signal y = {y1, y2,... y n} of the GIS device is subjected to time-frequency domain conversion:

[0029]

[0030] In the formula, a and b respectively represent the scale and displacement factors, represents the wavelet basis function, is the conjugate of ψ(·).

[0031] Then there is the frequency domain expression of WT(a, b):

[0032]

[0033] In the formula, respectively represent the wavelet transforms of the time domain signal / f(ω) and ψ(ω), ω represents the frequency, represents the imaginary number.

[0034] Step S2: Perform single feature analysis on the real-time sub-vibration characteristic set to generate a single feature analysis result; perform chromaticity map mapping on the real-time sub-vibration characteristic set according to the single feature analysis result to generate a real-time chromaticity map;

[0035] In the embodiments of the present invention, for each vibration feature in the real-time sub-vibration feature set, single-feature analysis is performed. These features include frequency, amplitude, energy, etc. Statistics of each vibration feature can be calculated, such as mean, standard deviation, maximum value, minimum value, etc., as well as other relevant indicators, such as spectral peak, etc. The results obtained from the single-feature analysis are integrated to form the single-feature analysis results. These results can be presented in the form of tables, charts, etc. for subsequent analysis and comparison. For each feature, some thresholds or reference ranges can be set to determine whether it is within the normal range, thereby making a preliminary assessment of the equipment status. The vibration features in the real-time sub-vibration feature set are mapped to a chromaticity map. The chromaticity map presents the changes of vibration features in time and frequency in the form of an image, usually represented by a heat map. After mapping the vibration features to the chromaticity map, the changes of vibration features over time and frequency can be visually observed, facilitating the discovery of abnormal vibration patterns and trends.

[0036] Step S3: Construct a first operating state analysis model; based on the first operating state analysis model, perform component state analysis on the real-time chromaticity map to obtain real-time equipment component state data of the GIS equipment.

[0037] In the embodiments of the present invention, by collecting historical data and equipment operating conditions, including vibration feature sets, component state data, etc. under normal and abnormal states. Select appropriate machine learning or statistical modeling methods, such as support vector machine (SVM), decision tree, neural network, etc., to construct a first operating state analysis model. During the model construction process, appropriate features should be selected, appropriate model parameters should be set, and model evaluation and optimization should be carried out to improve the accuracy and generalization ability of the model. The real-time chromaticity map is used as the input data of the model, and component state analysis is performed through the first operating state analysis model. During the component state analysis process, the model can evaluate the states of each component of the GIS equipment according to the vibration features in the real-time chromaticity map to determine whether it is in a normal operating state. If it is found that the state of a certain component is abnormal, corresponding real-time equipment component state data is generated, and an alarm or suggestions for further inspection and maintenance measures are put forward.

[0038] Preferably, the chromaticity map mapping of the real-time sub-vibration feature set according to the single-feature analysis result in step S2 includes:

[0039] Perform fast Fourier transform on the real-time sub-vibration feature set to generate a vibration audio high-frequency spectrum map;

[0040] Extract frequency components from the vibration audio high-frequency spectrum map to obtain vibration frequency components; perform chromaticity order mapping on the vibration frequency components and perform normalization processing on the mapping results to generate vibration chromaticity order data;

[0041] Accumulate the corresponding band amplitudes of the vibration frequency components according to the vibration chromaticity order data, so as to obtain a real-time chromaticity map.

[0042] In the present invention, by performing a fast Fourier transform on the real-time sub-vibration feature set, the vibration signal can be transformed from the time domain to the frequency domain to obtain a vibration audio frequency spectrum diagram. This spectrum diagram can show the energy distribution of the vibration signal at different frequencies, which helps to analyze the frequency components and the main frequency regions of the vibration. Extracting the frequency components from the vibration audio frequency spectrum diagram can obtain the main frequency components in the vibration signal. These frequency components correspond to different vibration modes or vibration sources in the vibration signal. By extracting these components, the characteristics and sources of the vibration can be better understood. Mapping the vibration frequency components to chromaticity order levels can map the frequency information of the vibration to different color levels on the chromaticity map. Such a mapping can visually represent the distribution of the vibration frequency, which helps to observe the strength and relative relationship of different frequency components. Normalization processing can ensure the comparability and consistency of the chromaticity maps at different time points. Through chromaticity order mapping and normalization processing, vibration chromaticity order data can be generated. These data can be used for further analysis and visualization, such as showing the distribution of the vibration frequency through a heat map, or representing the intensity and change trend of the vibration through color coding. According to the vibration chromaticity order data, the corresponding band amplitudes of the vibration frequency components can be accumulated to obtain a real-time chromaticity map. This chromaticity map can visually display the intensity and changes of different frequency components at different time points, providing an important reference for equipment status analysis and fault diagnosis.

[0043] As an example of the present invention, refer to Figure 2 As shown, in this example, the chromaticity map mapping of the real-time sub-vibration feature set according to the single feature analysis result includes the following steps:

[0044] Step S21: Perform a fast Fourier transform on the real-time sub-vibration feature set to generate a vibration audio frequency spectrum diagram;

[0045] In the embodiment of the present invention, by obtaining the real-time sub-vibration feature set from step S1, including the time-domain data or spectrum data of the vibration signal. Using the fast Fourier transform (FFT) algorithm, the real-time sub-vibration feature set is transformed from the time domain to the frequency domain. FFT is an efficient algorithm that can transform a time-domain signal into a frequency-domain signal and provide the spectrum information of the signal. Plot the frequency-domain data obtained by the FFT transform into a spectrum diagram, where the horizontal axis represents the frequency and the vertical axis represents the amplitude or energy of the vibration signal. The spectrum diagram can visually display the distribution of the vibration signal at different frequencies, that is, the audio frequency spectrum diagram of the vibration signal. More specifically, a pitch calculation formula can also be used to calculate the pitch of the real-time sub-vibration feature set. The calculation formula is as follows:

[0046]

[0047] Wherein, p represents pitch, and its relationship with frequency is p = 69 + 12log2(f / 440), k ∈ [0, N - 1] represents the index number of frequency components, and N represents the signal length.

[0048] Step S22: Extract frequency components from the vibration tone high-frequency spectrogram to obtain vibration frequency components; perform chromatic order mapping on the vibration frequency components, and normalize the mapping result to generate vibration chromatic order data;

[0049] In the embodiment of the present invention, in the vibration tone high-frequency spectrogram, the main vibration frequency components are extracted by methods such as peak detection or spectral analysis. A threshold can be set or an adaptive algorithm can be used to identify the frequency components with significant amplitudes as the main components of the vibration frequency. Perform chromatic order mapping on the extracted vibration frequency components to map them to different positions on the chromatic circle. Chromatic order mapping is a commonly used method for converting frequency components into corresponding pitch orders, usually using logarithmic or linear mapping methods. Normalize the mapping result to map the vibration frequency components to a unified range for subsequent data processing and analysis. The normalization process can be achieved through linear transformation or other mathematical methods to ensure that the values of all frequency components are within a certain range. The specific corresponding mapping algorithm is as follows.

[0050] χ(k) = 12[log2(f s / N×k / f ref )] / 12;

[0051] Wherein, fs represents the signal sampling frequency, and f ref represents the reference frequency, which is the frequency value of a lower group of pitch A in the chromatic scale.

[0052] Therefore, the chromatic vector ξ ve is:

[0053]

[0054] Wherein, c represents a variable, and ξ ve (c) reflects the energy magnitude of each scale component belonging to the frequency band.

[0055] Furthermore, normalize the chromatic vector to obtain:

[0056] ξ norm (c) = ξ ve (c) / Pξ ve (c)P

[0057] Step S23: Accumulate the corresponding frequency band amplitudes of the vibration frequency components according to the vibration chromatic order data to obtain a real-time chromatic spectrogram.

[0058] In the embodiments of the present invention, by accumulating the vibration frequency components and the corresponding band amplitudes according to the vibration chromaticity order data. For each frequency band, traverse the vibration chromaticity order data, and accumulate the amplitudes of the vibration frequency components falling within the frequency band. The accumulated band amplitude is used as the data of the real-time chromaticity map. The real-time chromaticity map is an image with the frequency band as the horizontal axis and the vibration amplitude as the vertical axis, and is used to represent the vibration energy distribution on different frequency bands.

[0059] Preferably, constructing the first operating state analysis model in step S3 includes:

[0060] Obtain the historical vibration information and historical maintenance logs of the GIS device;

[0061] Perform time-frequency conversion analysis on the historical vibration information to obtain a number of historical sub-vibration feature sets, and perform chromaticity map mapping on the historical sub-vibration feature sets to generate a historical chromaticity map;

[0062] Perform maintenance time series analysis on the historical maintenance logs to determine the historical equipment component operating state data at different time nodes;

[0063] Based on different time nodes, perform time-corresponding association on the historical chromaticity map and the historical equipment component operating state data to obtain a time-associated historical chromaticity spectrum;

[0064] Divide the time-associated historical chromaticity spectrum and the historical equipment component operating state data into data sets to generate a model training set and a model test set;

[0065] Use the LeNet-5 neural network algorithm to train the model training set to generate a first operating state training model; use the model test set to perform model optimization iteration on the first operating state training model, thereby generating a first operating state analysis model.

[0066] The present invention obtains the historical vibration information and historical maintenance logs of GIS equipment. These data contain the vibration conditions and maintenance records of the equipment over a past period. By analyzing these historical data, the operation mode, vibration characteristics, and maintenance conditions of the equipment can be identified, providing a basis for subsequent analysis and modeling. Performing time-frequency conversion analysis on the historical vibration information can obtain several historical sub-vibration feature sets. By performing chromaticity map mapping on these feature sets, historical chromaticity maps can be generated. These maps can reflect the vibration characteristics and frequency distribution of the equipment at different time points, helping to understand the historical vibration state of the equipment. Performing maintenance time-series analysis on the historical maintenance logs can determine the historical equipment component operation state data at different time nodes. These data record the maintenance and repair conditions of the equipment at different times and can be used to judge the health status and maintenance history of the equipment. Correlating the historical chromaticity maps and the historical equipment component operation state data in terms of time can obtain a time-correlated historical chromaticity spectrogram. Such a spectrogram can correspond the vibration characteristics and equipment state information, providing a more comprehensive historical data analysis result. Dividing the time-correlated historical chromaticity spectrogram and the historical equipment component operation state data into data sets can generate a model training set and a model testing set. These data sets are used to train and evaluate the first operation state analysis model. Using the LeNet-5 neural network algorithm to train the model training set can generate the first operation state training model. Optimizing and iterating the training model through the model testing set can generate a more accurate and reliable first operation state analysis model. This model can be used for real-time equipment state analysis and fault prediction.

[0067] As an example of the present invention, refer to Figure 3 As shown, the steps of constructing the first operation state analysis model in this example include:

[0068] Step S31: Obtain the historical vibration information and historical maintenance logs of the GIS equipment;

[0069] In the embodiment of the present invention, by collecting the historical vibration information of the GIS equipment, including vibration signal data collected by sensors such as vibration sensors or accelerometers. These vibration information should contain the vibration data during the operation of the equipment, covering the vibration characteristics at different time periods and different operation states. Collect the historical maintenance logs of the GIS equipment, recording the maintenance and repair conditions of the equipment, including information such as repair dates, repair contents, and replaced parts. These maintenance logs can be recorded and archived in the form of equipment maintenance records, maintenance reports, or maintenance databases.

[0070] Step S32: Perform time-frequency conversion analysis on the historical vibration information to obtain several historical sub-vibration feature sets, and perform chromaticity map mapping on the historical sub-vibration feature sets to generate historical chromaticity maps;

[0071] In the embodiments of the present invention, through time-frequency conversion analysis of historical vibration information, a method similar to step S1 can be adopted, such as using algorithms like complex Gabor-Morlet wavelet transform. The historical vibration information is converted into time-frequency domain data to obtain several historical sub-vibration feature sets. Chromaticity map mapping is performed on each historical sub-vibration feature set to map the vibration features onto the chromaticity map, so as to visually display the time-frequency features of the historical vibration information. The chromaticity map can reflect the distribution of the vibration signal at different frequencies and times, which helps to discover the laws and trends of historical vibrations. The chromaticity maps of all historical sub-vibration feature sets are integrated and summarized to obtain a historical chromaticity map. Statistical analysis and visualization processing can be performed on the historical chromaticity map to reveal the vibration features and operating states of the GIS device in different historical periods.

[0072] Step S33: Perform overhaul time-series analysis on the historical overhaul logs to determine the historical device component operating state data at different time nodes;

[0073] In the embodiments of the present invention, by collecting past device overhaul logs, these logs include device maintenance records, fault reports, repair records, etc. These logs are usually recorded in text form and also contain timestamps and other relevant information. Data cleaning and sorting are performed on the collected historical overhaul logs to ensure the integrity and accuracy of the data. This involves operations such as removing duplicate data, filling in missing values, and unifying the time format. The cleaned historical overhaul log data is converted into time-series data, that is, a data sequence arranged in chronological order. The data at each time point should include the operating state of the device components, such as normal, under maintenance, faulty, etc. Time-series analysis techniques are used to analyze the data to determine the operating states of the device components at different time nodes. Common time-series analysis methods include time series models, sliding window analysis, periodic analysis, etc.

[0074] Step S34: Based on different time nodes, perform time-corresponding association between the historical chromaticity map and the historical device component operating state data to obtain a time-associated historical chromaticity spectrogram;

[0075] In the embodiments of the present invention, time-corresponding association is performed between the historical chromaticity map data and the historical device component operating state data. This can be achieved through timestamps or time intervals to ensure that there is corresponding chromaticity map and device component operating state data for each time point or time period. Based on the time-associated data, a time-associated historical chromaticity spectrogram is generated. This can be visualizing the chromaticity map data and the device operating state data on the same graph, or a data structure representing the relationship between chromaticity and device state obtained through statistical analysis methods.

[0076] Step S35: Divide the historical chromaticity spectrograms associated with time and the historical operating status data of device components into datasets to generate a model training set and a model test set;

[0077] In the embodiments of the present invention, the division ratio of the training set and the test set is determined according to the actual situation. Usually, the training set occupies most of the data, while the test set occupies a smaller proportion. For example, a common division ratio is that 80% of the data is used for training and 20% of the data is used for testing. Randomly divide the historical dataset associated with time into a training set and a test set according to the determined ratio. Ensure that the time order is maintained during the division to avoid the chaos of time series data. If the number of samples of different device states in the historical data is unbalanced, some strategies can be considered to balance the number of samples of different categories in the training set and the test set when dividing the dataset, so as to avoid the model being biased towards the category with a larger number of samples. Ensure that each sample in the training set and the test set is correctly labeled with the corresponding operating status of the device component, and these labels will be used as the targets for model training and testing.

[0078] Step S36: Use the LeNet-5 neural network algorithm to train the model training set to generate a first operating status training model; use the model test set to optimize and iterate the first operating status training model to generate a first operating status analysis model.

[0079] In the embodiments of the present invention, a neural network model is constructed according to the network structure of LeNet-5. LeNet-5 includes two convolutional layers, two pooling layers and three fully connected layers, and is suitable for processing image data.

[0080] Compile the constructed LeNet-5 model and select appropriate loss functions, optimizers and evaluation metrics. The loss function can choose the cross-entropy loss function, the optimizer can choose the Adam optimizer, and the evaluation metric can choose accuracy, etc. Use the prepared training data to train the LeNet-5 model. During the training process, continuously adjust the model parameters through the backpropagation algorithm to minimize the loss function. Use a validation set other than the training set to evaluate the trained model, and evaluate the performance of the model on the validation set, including metrics such as accuracy, precision, and recall. Save the trained first operating status training model to disk for subsequent use. The calculation formulas for the convolutional layer and the pooling layer of the specific LeNet-5 neural network algorithm are as follows:

[0081]

[0082] In the formula, f(g) represents the activation function, and the ReLU function can be selected, X i,jDenote the input element at the $i$-th row and $j$-th column, $\sigma$ denote the convolutional kernel element, $m$ and $n$ respectively denote the size of the convolutional kernel, $\delta$ denote the error offset. $D(g)$ denotes downsampling, and $Y_{i,j}$ denotes the element in the pooling region.

[0083] Preferably, the obtaining of the historical vibration information and historical maintenance log of the GIS device includes:

[0084] Apply a preset fault to the GIS device, and for each application of the preset fault, use a vibration sensor to collect the historical vibration information data of the GIS device;

[0085] Perform fault maintenance on the GIS device based on the historical vibration information data to obtain historical maintenance data;

[0086] Use the cloud platform to upload the data log of the historical maintenance data, so as to obtain the historical maintenance log.

[0087] In the present invention, by applying a preset fault and using a vibration sensor to collect the historical vibration information data of the GIS device, the vibration characteristics under real fault conditions can be simulated. These historical vibration information data provide a detailed record of the vibration patterns and characteristics of the device under different fault states. By analyzing these data, the vibration characteristics corresponding to various fault modes can be identified, providing a basis for subsequent fault maintenance and condition analysis. Performing fault maintenance based on the historical vibration information data can obtain historical maintenance data by repairing the device and replacing components, etc. These data record the maintenance and repair records of the device at different time points, including information such as replaced components, repair processes, and repair results. By analyzing these data, the maintenance history and maintenance situation of the device can be understood, providing a basis for the health assessment and prediction of the device. Using the cloud platform to upload the data log of the historical maintenance data can centrally store and manage the historical maintenance records. By uploading the historical maintenance log, data backup and sharing can be realized, facilitating subsequent data analysis and query. The cloud platform also provides data processing and visualization functions, making the historical maintenance log more convenient for analysis and application.

[0088] In the embodiments of the present invention, for GIS devices, a preset fault scenario is designed and implemented, such as simulating the damage or abnormality of a certain component of the device. Sensor devices such as vibration sensors or accelerometers are installed on the GIS device to monitor the vibration of the device in real time. After each preset fault is applied, historical vibration information data of the GIS device is collected by the sensor. These data will record the vibration characteristics of the device in the fault state and can be used for subsequent fault diagnosis and prediction. Based on the collected historical vibration information data, the vibration characteristics and patterns of the GIS device in the fault state are analyzed. According to the vibration characteristics, fault diagnosis and location are carried out to determine the faulty components or problems existing in the device. Corresponding maintenance and repair work is carried out to repair or replace the damaged device components to restore the normal operation state of the device. The historical maintenance data is uploaded to the cloud platform for centralized management and storage. The advantages of cloud storage can be utilized to back up, archive, and analyze the historical maintenance data to achieve long-term preservation and effective utilization of the data. A data logging system is established on the cloud platform to ensure the integrity, traceability, and security of the historical maintenance data, facilitating subsequent data query and analysis work.

[0089] Preferably, the model optimization iteration of the first operating state training model using the model test set includes:

[0090] A real-time vibration acquisition timeline is established based on the real-time vibration information of the GIS device, and time points are marked on the real-time vibration acquisition timeline based on a preset time interval to generate re-detection points;

[0091] Extract the time-point vibration information from the real-time vibration acquisition timeline according to adjacent re-detection points to obtain the first vibration information extraction data and the second vibration information extraction data; perform repeated vibration segment discrimination on the first vibration information extraction data and the second vibration information extraction data. When it is determined that there is no repeated vibration information segment, chromaticity map mapping is performed according to the first vibration information extraction data and the second vibration information extraction data of the non-repeated information segment to generate the first chromaticity map and the second chromaticity map;

[0092] Perform map joint analysis on the first chromaticity map and the second chromaticity map to generate a joint analysis map group;

[0093] Perform comparison analysis on each chromaticity map of the joint analysis map group, construct a vibration comparison chromaticity map based on the comparison analysis results, and perform vibration accuracy feature analysis on the vibration comparison chromaticity map to generate vibration chromaticity map accuracy feature data;

[0094] Based on the vibration chromaticity map accuracy feature data, construct a model optimization strategy for the first operating state training model to generate a model optimization strategy;

[0095] According to the model optimization strategy, the first running state training model is iteratively optimized using the model test set to generate the first running state analysis model.

[0096] In the present invention, a real-time vibration acquisition timeline is established, and time points are marked on the timeline according to a preset time interval, so as to generate re-detection points. These re-detection points are used to extract vibration information from the real-time vibration acquisition timeline for subsequent analysis and modeling. According to adjacent re-detection points, vibration information is extracted from the real-time vibration acquisition timeline to obtain first vibration information extraction data and second vibration information extraction data. The discrimination of repeated vibration segments is performed on this information, and the non-repeated information segments are extracted for chromaticity map mapping to generate a first chromaticity map and a second chromaticity map. These chromaticity maps reflect the vibration characteristics and frequency distribution of the equipment, providing a basis for subsequent analysis and comparison. The chromaticity maps are jointly analyzed to generate a group of jointly analyzed maps. Then, through the comparison and analysis of each chromaticity map, a vibration comparison chromaticity map can be constructed for the analysis of vibration precision characteristics. These characteristic data can provide more accurate and detailed vibration information, providing a basis for model optimization. Based on the vibration chromaticity map precision characteristic data, a model optimization strategy can be constructed. These strategies can include measures in aspects such as feature selection, model parameter adjustment, and algorithm optimization, aiming to improve the accuracy and robustness of the model. Then, the first running state training model is iteratively optimized using the model test set. By continuously adjusting and updating the model, a more accurate and reliable first running state analysis model is generated.

[0097] In an embodiment of the present invention, real-time vibration information is obtained by using a GIS device, and a real-time vibration acquisition timeline is established. According to a preset time interval, time points are marked on the timeline to generate re-detection points. Based on adjacent re-detection points, vibration information is extracted from the real-time vibration acquisition timeline to obtain first vibration information extraction data and second vibration information extraction data. The first vibration information extraction data and the second vibration information extraction data are discriminated for repeated vibration segments to determine non-repeated vibration information segments. The first vibration information extraction data and the second vibration information extraction data of the non-repeated information segments are used for chromaticity map mapping to generate a first chromaticity map and a second chromaticity map. The first chromaticity map and the second chromaticity map are subjected to map joint analysis to generate a joint analysis map group. Each chromaticity map in the joint analysis map group is subjected to comparison analysis, and a vibration comparison chromaticity map is constructed according to the comparison result. The vibration comparison chromaticity map is subjected to vibration accuracy feature analysis to extract relevant feature data. Based on the vibration chromaticity map accuracy feature data, a model optimization strategy is constructed. This may include measures in aspects such as feature selection, model parameter adjustment, and algorithm optimization. The first operating state training model is subjected to model optimization iteration using a model test set. According to the model optimization strategy, the model is adjusted and updated to generate a more accurate and reliable first operating state analysis model.

[0098] Preferably, the comparison analysis of each chromaticity map in the joint analysis map group and the vibration accuracy feature analysis of the vibration comparison chromaticity map include:

[0099] Establish a chromaticity acquisition point acquisition rule, and perform point chromaticity acquisition on the first chromaticity map and the second chromaticity map in the joint analysis map group according to the chromaticity acquisition point acquisition rule to obtain chromaticity map point chromaticity; record chromaticity information for the chromaticity map point chromaticity to generate chromaticity map acquisition point information data, where the chromaticity information record includes the chromaticity record and position record of the acquisition point, and the chromaticity map acquisition point information data includes the first chromaticity map acquisition point information data of the first chromaticity map and the second chromaticity map acquisition point information data of the second chromaticity map;

[0100] Perform chromaticity difference analysis on the chromaticity map acquisition point information data to generate a first chromaticity acquisition point difference amount and a second chromaticity acquisition point difference amount;

[0101] Compare the first chromaticity acquisition point difference amount and the second chromaticity acquisition point difference amount with a preset chromaticity acquisition difference threshold. If both the first chromaticity acquisition point difference amount and the second chromaticity acquisition point difference amount are less than the preset chromaticity acquisition difference threshold, then calculate the average value of the first chromaticity acquisition point difference amount and the second chromaticity acquisition point difference amount to obtain the average chromaticity acquisition point difference amount; use the average chromaticity acquisition point difference amount to perform corresponding chromaticity replacement on the blocks with chromaticity differences in the joint analysis map group to obtain the vibration comparison chromaticity map;

[0102] Based on the difference amount of the first acquisition point and the difference amount of the second acquisition point, perform vibration accuracy feature analysis on the vibration comparison chromaticity map to generate vibration chromaticity map accuracy feature data.

[0103] In the present invention, by establishing a chromaticity acquisition point acquisition rule, it is possible to ensure consistent point chromaticity acquisition for the first chromaticity map and the second chromaticity map in the combined analysis map group. This helps to ensure the accuracy and reliability of subsequent analysis. By performing chromaticity difference analysis on the chromaticity map acquisition point information data, the difference between the first chromaticity map and the second chromaticity map can be quantified. This helps to determine the blocks with differences and provides a basis for subsequent chromaticity replacement. By using the average value of the chromaticity difference amount to perform chromaticity replacement on the blocks with chromaticity differences in the combined analysis map group, a vibration comparison chromaticity map can be generated. Such processing can highlight the effect of vibration comparison and provide a clearer and more easily observable chromaticity map. Based on the difference amount of the first acquisition point and the difference amount of the second acquisition point, perform vibration accuracy feature analysis on the vibration comparison chromaticity map. This helps to further understand the characteristics and behaviors of vibration and generate relevant feature data. These feature data can provide important indicators for model optimization and performance evaluation.

[0104] In an embodiment of the present invention, by designing a chromaticity acquisition point acquisition rule, including the selection and interval of acquisition points. These rules should consider the importance of vibration characteristics to ensure that the acquired data can fully reflect the vibration situation. According to the chromaticity acquisition point acquisition rule, perform point chromaticity acquisition on the first chromaticity map and the second chromaticity map in the combined analysis map group. The acquired data should include chromaticity records and position records for subsequent analysis and comparison. Perform difference analysis on the chromaticity map acquisition point information data to calculate the difference amount of the first chromaticity acquisition point and the difference amount of the second chromaticity acquisition point. Compare the difference amount of the first chromaticity acquisition point and the difference amount of the second chromaticity acquisition point with a preset chromaticity acquisition difference threshold. If both difference amounts are less than the threshold, calculate the average value of the difference amounts and perform corresponding chromaticity replacement on the chromaticity blocks with larger differences to make the chromaticity more consistent. Based on the difference amount of the first acquisition point and the difference amount of the second acquisition point, perform vibration accuracy feature analysis on the vibration comparison chromaticity map, including the extraction and analysis of characteristics such as the amplitude, frequency, and period of vibration. According to the results of the vibration accuracy feature analysis, generate vibration chromaticity map accuracy feature data. These data can be used for subsequent model optimization and fault diagnosis.

[0105] Preferably, the establishment of the chromaticity acquisition point acquisition rule and the point chromaticity acquisition of the first chromaticity map and the second chromaticity map in the combined analysis map group according to the chromaticity acquisition point acquisition rule include:

[0106] Collect information on the sequence order of the acquisition points and the spacing between adjacent acquisition points in the combined analysis atlas group to obtain the acquisition point order data and the adjacent acquisition point spacing information data; sort the acquisition point order data and the adjacent acquisition point spacing information data to generate an acquisition point sequence;

[0107] Perform random scan mapping analysis on the first chromaticity atlas and the second chromaticity atlas in the combined analysis atlas group according to the acquisition point sequence, and record the analysis results to obtain the first chromaticity record data and the second chromaticity record data; sort the first chromaticity record data and the second chromaticity record data to generate a chromaticity reference sequence;

[0108] Perform several chromaticity scan mapping analyses on the remaining atlases in the combined analysis atlas group except for the first chromaticity atlas and the second chromaticity atlas according to the acquisition point sequence. And each time a scan mapping analysis is performed, change the position of the first acquisition point in the acquisition point sequence of the remaining atlases in the combined analysis atlas group except for the first chromaticity atlas and the second chromaticity atlas relative to the first chromaticity atlas and the second chromaticity atlas to generate several chromaticity comparison sequences;

[0109] Calculate the chromaticity similarity between each chromaticity comparison sequence and the chromaticity reference sequence to obtain the chromaticity comparison similarity degree; mark the sequence information corresponding to the chromaticity comparison similarity degree greater than the preset similarity threshold to generate a chromaticity acquisition point acquisition rule;

[0110] Perform point chromaticity acquisition on the first chromaticity atlas and the second chromaticity atlas in the combined analysis atlas group according to the chromaticity acquisition point acquisition rule to obtain the chromaticity atlas point chromaticity.

[0111] By collecting the order of the acquisition points and the information on the spacing between adjacent acquisition points in the combined analysis atlas group, the present invention can obtain an accurate layout and arrangement of the acquisition points. This helps to establish consistent acquisition rules, ensuring the consistency and comparability of subsequent chromaticity acquisition processes. By sorting the acquisition point sequence and performing random scan mapping analysis on the first chromaticity atlas and the second chromaticity atlas according to the sorted sequence, chromaticity information can be recorded and a reference sequence can be generated. Such processing helps to establish an accurate chromaticity reference benchmark, providing a basis for subsequent chromaticity comparison. By performing multiple chromaticity scan mapping analyses on atlases other than the first chromaticity atlas and the second chromaticity atlas and changing the position of the acquisition point sequence, multiple chromaticity comparison sequences can be generated. Such processing can increase the diversity and comprehensiveness of the comparison, improving the accuracy and reliability of chromaticity comparison. By calculating the chromaticity similarity between each chromaticity comparison sequence and the chromaticity reference sequence, the degree of similarity can be determined. According to a preset similarity threshold, the comparison sequences with a similarity degree higher than the threshold are marked, thereby determining the acquisition rules for chromaticity acquisition points. Such processing helps to screen out reliable acquisition points, improving the accuracy and consistency of chromaticity spectra. By performing point chromaticity acquisition on the first chromaticity atlas and the second chromaticity atlas according to the acquisition rules for chromaticity acquisition points, accurate point chromaticity data of the chromaticity spectra can be obtained. This provides basic data for subsequent analysis and processing, facilitating further vibration characteristic analysis and research.

[0112] In an embodiment of the present invention, information on the order of acquisition points and the spacing between adjacent acquisition points in the combined analysis atlas group is collected. This can be accomplished through image processing techniques or manual annotation. The acquisition point order data and the information data on the spacing between adjacent acquisition points are sorted to obtain an ordered set of acquisition point sequences. According to the acquisition point sequence, random scan mapping analysis is performed on the first chromaticity atlas and the second chromaticity atlas in the combined analysis atlas group. During each scan mapping analysis, the analysis results are recorded to obtain the first chromaticity record data and the second chromaticity record data. The first chromaticity record data and the second chromaticity record data are sorted by chromaticity to obtain a chromaticity reference sequence. According to the acquisition point sequence, multiple chromaticity scan mapping analyses are performed on atlases other than the first chromaticity atlas and the second chromaticity atlas in the combined analysis atlas group. During each scan mapping analysis, the position of the first acquisition point in the acquisition point sequence of atlases other than the first chromaticity atlas and the second chromaticity atlas relative to the first chromaticity atlas and the second chromaticity atlas is changed to generate multiple chromaticity comparison sequences. The chromaticity similarity between each chromaticity comparison sequence and the chromaticity reference sequence is calculated to obtain the chromaticity comparison similarity degree. According to a preset similarity threshold, the chromaticity comparison sequences with a chromaticity comparison similarity degree greater than the threshold are marked, thereby generating the acquisition rules for chromaticity acquisition points. According to the acquisition rules for chromaticity acquisition points, point chromaticity acquisition is performed on the first chromaticity atlas and the second chromaticity atlas in the combined analysis atlas group to obtain the point chromaticity data of the chromaticity spectra.

[0113] Preferably, the vibration accuracy feature analysis of the vibration contrast chromaticity map based on the first collection point difference amount and the second collection point difference amount includes:

[0114] Calculating the vibration accuracy feature of the vibration contrast chromaticity map based on the first collection point difference amount and the second collection point difference amount by using the accuracy feature formula to obtain the vibration chromaticity map accuracy feature data;

[0115] The accuracy feature formula is specifically as follows:

[0116]

[0117] In the formula, J is the corresponding value of the accuracy feature between real-time chromaticity maps, k1 is the accuracy conversion coefficient, c1 is the first collection point position adjustment coefficient, x1 is the first collection point difference amount, Δx1 is the first collection point difference standard amount, c2 is the second collection point position adjustment coefficient, x2 is the second collection point difference amount, Δx2 is the second collection point difference standard amount, b1 is the first collection point chromaticity adjustment constant, and b2 is the second collection point chromaticity adjustment constant.

[0118] The present invention can calculate the vibration chromaticity map accuracy feature data according to the first collection point difference amount and the second collection point difference amount by using the accuracy feature formula. This feature value reflects the accuracy between vibration chromaticity maps and provides a quantitative evaluation index. In the accuracy feature formula, the difference amounts of the first collection point and the second collection point can be adjusted and weighted by setting the accuracy conversion coefficient and the position adjustment coefficients (c1 and c2). This helps to flexibly control the weight distribution of the accuracy feature calculation according to actual requirements and the importance of vibration features. The difference standard amounts (Δx1 and Δx2) and the chromaticity adjustment constants (b1 and b2) in the formula are used to standardize and adjust the difference amounts of the first collection point and the second collection point. Such processing can reasonably control and adjust the range of vibration features and chromaticity adjustment according to actual situations and data characteristics. The vibration chromaticity map accuracy feature data obtained by calculation can be further analyzed and compared. This helps to understand the relationship between vibration features, evaluate the accuracy and reliability of vibration contrast, and provide important reference indexes for subsequent model optimization and performance evaluation.

[0119] In the embodiment of the present invention, by using the given precision characteristic formula, according to the first acquisition point difference amount x1 and the second acquisition point difference amount x2, the precision characteristic value J of the vibration comparison chromaticity map is calculated. The characteristic value J reflects the precision characteristic corresponding value between the real-time chromaticity maps. The higher the characteristic value J, the higher the precision of the vibration comparison chromaticity map. Analyze the calculated characteristic value J, and judge the precision of the vibration comparison chromaticity map according to the size of the characteristic value. A higher characteristic value indicates a higher similarity between the chromaticity maps and better precision.

[0120] Preferably, in step S3, the component state analysis of the real-time chromaticity map based on the first operating state analysis model includes:

[0121] Import the real-time chromaticity map into the first operating state analysis model for GIS device state operation analysis to generate GIS device state operation analysis data;

[0122] Perform device anomaly identification on the GIS device state operation analysis data to generate GIS device anomaly identification data; use the GIS device anomaly identification data to classify the anomalies of the GIS device state operation analysis data to generate a GIS device anomaly fault classification result;

[0123] Perform closed-loop feedback control based on the GIS device anomaly fault classification result to generate a GIS closed-loop feedback control strategy; perform anomaly regulation on the GIS device anomaly identification data according to the GIS closed-loop feedback control strategy to perform normal operation monitoring of the GIS device.

[0124] In the present invention, by importing the real-time chromaticity map into the first operating state analysis model for analysis, the state operation analysis data of the GIS device can be obtained. These data provide a comprehensive evaluation of the current operating state of the GIS device, including the calculation results of various key parameters and indicators. By performing anomaly identification on the GIS device state operation analysis data, existing device anomaly situations can be detected. Further, classify the anomaly data to distinguish and classify different types of anomaly faults. This helps to discover device problems in advance and perform timely maintenance and processing. Based on the GIS device anomaly fault classification result, a closed-loop feedback control strategy can be established. This strategy can provide corresponding control schemes and operation guides according to different anomaly situations and fault types to ensure the normal operation of the GIS device. Such strategy generation is an intelligent decision based on the analysis results, which can improve the efficiency of device maintenance and management. Using the GIS device anomaly identification data for anomaly regulation can take corresponding measures and adjustments for the detected anomaly situations. This helps to improve the operating state of the device and avoid further faults. At the same time, by monitoring and analyzing the anomaly identification data, the normal operation of the GIS device can be monitored in real time, and potential problems can be discovered and solved in a timely manner.

[0125] In the embodiments of the present invention, by importing the real-time chromaticity map into the first operating state analysis model for GIS device state operation analysis, which involves technologies such as image processing, pattern recognition, and machine learning, to extract features from the chromaticity map and perform state analysis. Generate GIS device state operation analysis data, including device operation status and performance characteristics. Perform anomaly recognition on the GIS device state operation analysis data. This can be achieved by setting thresholds or using supervised learning methods to identify abnormal states. The anomaly recognition results include information such as the location, type, and degree of the anomaly, generating GIS device anomaly recognition data. Use the GIS device anomaly recognition data to classify the anomalies in the GIS device state operation analysis data, which involves using expert rules, machine learning algorithms, or deep learning models to classify the anomalies. The anomaly classification results include information such as the anomaly type, cause, and impact, generating GIS device anomaly fault classification results. Based on the GIS device anomaly fault classification results, design a closed-loop feedback control strategy. This includes determining the handling method of the anomaly, adjusting device parameters, or taking other measures to deal with the abnormal situation to ensure the normal operation of the device. According to the closed-loop feedback control strategy, perform anomaly regulation on the GIS device anomaly recognition data, including operations such as sending alarms, adjusting device parameters, starting standby devices, etc., and real-time monitoring of the device operation to ensure that the device can still operate normally under abnormal conditions.

[0126] In this specification, a device monitoring system based on chromaticity map mapping is provided for performing the above-mentioned device monitoring method based on chromaticity map mapping. The device monitoring system based on chromaticity map mapping includes:

[0127] A vibration information analysis module, configured to collect real-time vibration information of the GIS device, and perform time-frequency conversion analysis on the real-time vibration information by using complex Gabor-Morlet wavelet transform, and perform time-frequency conversion analysis on the real-time vibration information to obtain a real-time sub-vibration feature set;

[0128] A chromaticity map generation module, configured to perform single-feature analysis on the real-time sub-vibration feature set to generate a single-feature analysis result; perform chromaticity map mapping on the real-time sub-vibration feature set according to the single-feature analysis result to generate a real-time chromaticity map;

[0129] An operating state analysis module, configured to construct a first operating state analysis model; perform component state analysis on the real-time chromaticity map based on the first operating state analysis model to obtain real-time device component state data of the GIS device.

[0130] The beneficial effects of the present invention are as follows: by collecting the real-time vibration information of GIS equipment and using complex Gabor-Morlet wavelet transform for time-frequency conversion analysis, efficient processing and analysis of vibration signals can be achieved. This helps to promptly detect abnormal vibration conditions of the equipment and give early warnings of possible faults. Conducting single-feature analysis on the real-time sub-vibration feature set and generating a chromaticity map can visually display the vibration characteristics and state changes of the equipment. This helps engineers and operators more quickly understand the operating state of the equipment for fault diagnosis and maintenance. By constructing the first operating state analysis model, systematic component state analysis of the real-time chromaticity map can be carried out to further extract the real-time equipment component state data of the equipment. This helps to establish an equipment state monitoring and prediction model to achieve intelligent management and optimized maintenance of the equipment state. Based on the above analysis results, real-time monitoring and analysis of the operating state of GIS equipment can be realized, promptly detecting abnormal states and potential faults of the equipment, thus giving early warnings and taking corresponding maintenance measures to reduce the equipment failure rate and extend the equipment life. By implementing the above analysis and prediction models, precise maintenance of GIS equipment can be achieved, avoiding unnecessary maintenance and downtime, reducing maintenance costs, and improving the reliability and stability of the equipment.

[0131] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be encompassed within the present invention.

[0132] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A device monitoring method based on chromaticity spectrum mapping, characterized in that: The following steps are involved: Step S1: collecting real-time vibration information of GIS equipment, and using complex Gabor-Morlet wavelet transform to perform time-frequency conversion analysis on the real-time vibration information to obtain a real-time sub-vibration feature set; Step S2: performing a single feature analysis on the real-time sub-vibration feature set to generate a single feature analysis result; performing chromaticity spectrum mapping on the real-time sub-vibration feature set according to the single feature analysis result to generate a real-time chromaticity spectrum; Step S3: constructing a first operating status analysis model; performing component status analysis on the real-time chromaticity spectrum based on the first operating status analysis model to obtain real-time device component status data of the GIS device; constructing the first operating status analysis model in step S3 includes: Obtain historical vibration information and historical maintenance logs of GIS equipment; Performing time-frequency conversion analysis on historical vibration information to obtain a number of historical sub-vibration feature sets, and performing chromaticity spectrum mapping on the historical sub-vibration feature sets to generate a historical chromaticity spectrum; Perform maintenance time sequence analysis on historical maintenance logs to determine the historical equipment component operation status data at different time nodes; Based on different time nodes, the historical chromaticity spectrum and the historical equipment component operation status data are temporally correlated to obtain a time-correlated historical chromaticity spectrum; The time-correlated historical chromaticity spectra and historical equipment component operation status data are divided into data sets to generate model training sets and model test sets; The model training set is trained using the LeNet-5 neural network algorithm to generate a first operating state training model; the first operating state training model is optimized and iterated using the model test set to generate a first operating state analysis model; the optimization and iteration of the first operating state training model using the model test set includes: Establish a real-time vibration collection timeline based on the real-time vibration information of the GIS equipment, and mark the time points of the real-time vibration collection timeline based on the preset time interval to generate re-detection points; Extracting vibration information at a time point on a real-time vibration collection timeline according to adjacent re-detection points to obtain first vibration information extraction data and second vibration information extraction data; performing repeated vibration segment discrimination on the first vibration information extraction data and the second vibration information extraction data, and when determining that there is no repeated vibration information segment, performing chromaticity spectrum mapping on the first vibration information extraction data and the second vibration information extraction data without the repeated information segment to generate a first chromaticity spectrum and a second chromaticity spectrum; Performing a joint analysis of the first chromaticity spectrum and the second chromaticity spectrum to generate a joint analysis spectrum group; Performing a comparison analysis on each chromaticity spectrum of the joint analysis spectrum group, and constructing a vibration contrast chromaticity spectrum based on the comparison analysis results, performing a vibration accuracy feature analysis on the vibration contrast chromaticity spectrum, and generating vibration chromaticity spectrum accuracy feature data; performing a comparison analysis on each chromaticity spectrum of the joint analysis spectrum group and performing a vibration accuracy feature analysis on the vibration contrast chromaticity spectrum include: Establishing a collection rule for chromaticity collection points, and collecting point chromaticity of the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group according to the collection rule to obtain the point chromaticity of the chromaticity spectrum; recording chromaticity information of the point chromaticity of the chromaticity spectrum to generate chromaticity spectrum collection point information data, wherein the chromaticity information record includes a chromaticity record and a position record of the collection point, and the chromaticity spectrum collection point information data includes first chromaticity spectrum collection point information data of the first chromaticity spectrum and second chromaticity spectrum collection point information data of the second chromaticity spectrum; Perform chromaticity difference analysis on the chromaticity spectrum acquisition point information data to generate a first chromaticity acquisition point difference amount and a second chromaticity acquisition point difference amount; Compare the difference amount of the first chromaticity acquisition point and the difference amount of the second chromaticity acquisition point with a preset chromaticity acquisition difference threshold; if the difference amount of the first chromaticity acquisition point and the difference amount of the second chromaticity acquisition point are both less than the preset chromaticity acquisition difference threshold, calculate the average value of the difference amount of the first chromaticity acquisition point and the second chromaticity acquisition point to obtain the average value of the difference amount of the chromaticity acquisition point; use the average value of the difference amount of the chromaticity acquisition point to perform corresponding chromaticity replacement on the blocks with different chromaticity in the joint analysis spectrum group, so as to obtain a vibration contrast chromaticity spectrum; Performing vibration accuracy feature analysis on the vibration contrast chromaticity spectrum based on the difference between the first acquisition point and the second acquisition point to generate vibration chromaticity spectrum accuracy feature data; Based on the vibration chromaticity spectrum accuracy feature data, a model optimization strategy is constructed for the first operating state training model to generate a model optimization strategy; The first operating state training model is optimized and iterated using the model test set according to the model optimization strategy, thereby generating a first operating state analysis model.

2. The device monitoring method based on chromaticity spectrum mapping according to claim 1 is characterized in that: In step S2, the chromaticity spectrum mapping of the real-time sub-vibration feature set according to the single feature analysis result includes: Perform fast Fourier transformation on the real-time sub-vibration feature set to generate a vibration pitch spectrum diagram; Extract frequency components from the vibration pitch spectrum to obtain vibration frequency components; perform chromaticity order mapping on the vibration frequency components, and normalize the mapping results to generate vibration chromaticity order data; The corresponding frequency band amplitudes of the vibration frequency components are accumulated according to the vibration chromaticity order data to obtain a real-time chromaticity spectrum.

3. The device monitoring method based on chromaticity spectrum mapping according to claim 1, characterized in that: The acquisition of historical vibration information and historical maintenance logs of GIS equipment includes: Apply preset faults to the GIS equipment, and after each preset fault is applied, use a vibration sensor to collect historical vibration information data of the GIS equipment; Perform fault maintenance on GIS equipment based on historical vibration information data to obtain historical maintenance data; The cloud platform is used to upload data logs of historical maintenance data to obtain historical maintenance logs.

4. The device monitoring method based on chromaticity spectrum mapping according to claim 1, characterized in that: The step of establishing a chromaticity acquisition point collection rule and performing point chromaticity acquisition on the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group according to the chromaticity acquisition point collection rule comprises: Collecting information on the sequence of collection points and the distance between adjacent collection points for the joint analysis atlas group to obtain collection point sequence data and adjacent collection point distance information data; sorting the collection point sequence data and adjacent collection point distance information data to generate a collection point sequence; Performing random scanning mapping analysis on the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group according to the acquisition point sequence, and recording the analysis results to obtain first chromaticity recording data and second chromaticity recording data; performing chromaticity sorting on the first chromaticity recording data and the second chromaticity recording data to generate a chromaticity reference sequence; Performing a plurality of chromaticity scanning and mapping analyses on the remaining spectra except the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group according to the collection point sequence, and in each scanning and mapping analysis, transforming the position of the first collection point in the collection point sequence of the remaining spectra except the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group relative to the first chromaticity spectrum and the second chromaticity spectrum, to generate a plurality of chromaticity comparison sequences; Calculate the chromaticity similarity of each chromaticity comparison sequence and the chromaticity reference sequence respectively to obtain the chromaticity comparison similarity degree; mark the chromaticity comparison sequence corresponding to the chromaticity comparison similarity degree greater than the preset similarity threshold with sequence information, thereby generating the chromaticity collection point collection rule; According to the chromaticity acquisition point acquisition rule, point chromaticity acquisition is performed on the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group to obtain the point chromaticity of the chromaticity spectrum.

5. The device monitoring method based on chromaticity spectrum mapping according to claim 1, characterized in that: The vibration accuracy characteristic analysis of the vibration contrast chromaticity spectrum based on the difference between the first acquisition point and the second acquisition point includes: Calculate the vibration accuracy characteristics of the vibration contrast chromaticity spectrum based on the difference between the first acquisition point and the second acquisition point using the accuracy characteristic formula to obtain the vibration chromaticity spectrum accuracy characteristic data; The accuracy characteristic formula is as follows: In the formula, is the accuracy feature correspondence value between the real-time chromaticity spectra, is the accuracy conversion factor, is the position adjustment coefficient of the first acquisition point, is the difference of the first acquisition point, is the standard value of the difference at the first collection point, is the position adjustment coefficient of the second acquisition point, is the difference of the second acquisition point, is the standard value of the difference at the second collection point, is the chromaticity adjustment constant of the first acquisition point, Chromaticity adjustment constant for the second acquisition point.

6. The device monitoring method based on chromaticity spectrum mapping according to claim 1, characterized in that: In step S3, performing component status analysis on the real-time chromaticity spectrum based on the first operating status analysis model includes: Importing the real-time chromaticity spectrum into the first operation status analysis model to perform GIS equipment status operation analysis and generate GIS equipment status operation analysis data; Perform equipment anomaly identification on GIS equipment status operation analysis data to generate GIS equipment anomaly identification data; use GIS equipment anomaly identification data to perform anomaly classification on GIS equipment status operation analysis data to generate GIS equipment abnormal fault classification results; Based on the abnormal fault classification results of GIS equipment, closed-loop feedback control is performed to generate a GIS closed-loop feedback control strategy; according to the GIS closed-loop feedback control strategy, abnormal regulation is performed on the GIS equipment abnormal identification data to perform normal operation monitoring of GIS equipment.

7. A device monitoring system based on chromaticity spectrum mapping, characterized in that: Used to execute the device monitoring method based on chromaticity spectrum mapping as claimed in claim 1, the device monitoring system based on chromaticity spectrum mapping comprises: The vibration information analysis module is used to collect the real-time vibration information of GIS equipment and use the complex Gabor-Morlet wavelet transform to perform time-frequency conversion analysis on the real-time vibration information to obtain the real-time sub-vibration feature set; A chromaticity spectrum generation module is used to perform a single feature analysis on the real-time sub-vibration feature set to generate a single feature analysis result; perform chromaticity spectrum mapping on the real-time sub-vibration feature set according to the single feature analysis result to generate a real-time chromaticity spectrum; The operation status analysis module is used to construct a first operation status analysis model; based on the first operation status analysis model, the component status analysis is performed on the real-time chromaticity spectrum to obtain the real-time device component status data of the GIS device; wherein the construction of the first operation status analysis model includes: Obtain historical vibration information and historical maintenance logs of GIS equipment; Performing time-frequency conversion analysis on historical vibration information to obtain a number of historical sub-vibration feature sets, and performing chromaticity spectrum mapping on the historical sub-vibration feature sets to generate a historical chromaticity spectrum; Perform maintenance time sequence analysis on historical maintenance logs to determine the historical equipment component operation status data at different time nodes; Based on different time nodes, the historical chromaticity spectrum and the historical equipment component operation status data are temporally correlated to obtain a time-correlated historical chromaticity spectrum; The time-correlated historical chromaticity spectra and historical equipment component operation status data are divided into data sets to generate model training sets and model test sets; The model training set is trained using the LeNet-5 neural network algorithm to generate a first operating state training model; the first operating state training model is optimized and iterated using the model test set to generate a first operating state analysis model; the optimization and iteration of the first operating state training model using the model test set includes: Establish a real-time vibration collection timeline based on the real-time vibration information of the GIS equipment, and mark the time points of the real-time vibration collection timeline based on the preset time interval to generate re-detection points; Extracting vibration information at a time point on a real-time vibration collection timeline according to adjacent re-detection points to obtain first vibration information extraction data and second vibration information extraction data; performing repeated vibration segment discrimination on the first vibration information extraction data and the second vibration information extraction data, and when determining that there is no repeated vibration information segment, performing chromaticity spectrum mapping on the first vibration information extraction data and the second vibration information extraction data without the repeated information segment to generate a first chromaticity spectrum and a second chromaticity spectrum; Performing a joint analysis of the first chromaticity spectrum and the second chromaticity spectrum to generate a joint analysis spectrum group; Performing a comparison analysis on each chromaticity spectrum of the joint analysis spectrum group, and constructing a vibration contrast chromaticity spectrum based on the comparison analysis results, performing a vibration accuracy feature analysis on the vibration contrast chromaticity spectrum, and generating vibration chromaticity spectrum accuracy feature data; performing a comparison analysis on each chromaticity spectrum of the joint analysis spectrum group and performing a vibration accuracy feature analysis on the vibration contrast chromaticity spectrum include: Establishing a collection rule for chromaticity collection points, and collecting point chromaticity of the first chromaticity spectrum and the second chromaticity spectrum in the joint analysis spectrum group according to the collection rule to obtain the point chromaticity of the chromaticity spectrum; recording chromaticity information of the point chromaticity of the chromaticity spectrum to generate chromaticity spectrum collection point information data, wherein the chromaticity information record includes a chromaticity record and a position record of the collection point, and the chromaticity spectrum collection point information data includes first chromaticity spectrum collection point information data of the first chromaticity spectrum and second chromaticity spectrum collection point information data of the second chromaticity spectrum; Perform chromaticity difference analysis on the chromaticity spectrum acquisition point information data to generate a first chromaticity acquisition point difference amount and a second chromaticity acquisition point difference amount; Compare the difference amount of the first chromaticity acquisition point and the difference amount of the second chromaticity acquisition point with a preset chromaticity acquisition difference threshold; if the difference amount of the first chromaticity acquisition point and the difference amount of the second chromaticity acquisition point are both less than the preset chromaticity acquisition difference threshold, calculate the average value of the difference amount of the first chromaticity acquisition point and the second chromaticity acquisition point to obtain the average value of the difference amount of the chromaticity acquisition point; use the average value of the difference amount of the chromaticity acquisition point to perform corresponding chromaticity replacement on the blocks with different chromaticity in the joint analysis spectrum group, so as to obtain a vibration contrast chromaticity spectrum; Performing vibration accuracy feature analysis on the vibration contrast chromaticity spectrum based on the difference between the first acquisition point and the second acquisition point to generate vibration chromaticity spectrum accuracy feature data; Based on the vibration chromaticity spectrum accuracy feature data, a model optimization strategy is constructed for the first operating state training model to generate a model optimization strategy; The first operating state training model is optimized and iterated using the model test set according to the model optimization strategy, thereby generating a first operating state analysis model.

Images