Switch cabinet insulation fault characteristic gas on-line monitoring method and switch cabinet insulation fault characteristic gas on-line monitoring system

Abstract

The invention relates to the technical field of switch cabinet on-line monitoring, in particular to a switch cabinet insulation fault characteristic gas on-line monitoring method and system, and the method comprises the steps: collecting environment parameters and gas concentration data in a switch cabinet through a nano-material sensor, transmitting the environment parameters and gas concentration data to an edge computing device for preprocessing, and uploading the data to a cloud server; performing deep analysis on the preprocessed environmental parameters and gas concentration data by using a machine learning model, identifying the difference between a normal operation state and different types of insulation fault modes, and predicting the possible insulation fault risk in the future by using a fault prediction algorithm; high-precision detection of various characteristic gases is realized, the nano material sensor can identify trace fault characteristic gases, distinguish different types of fault gas components, provide a reliable early warning, capture tiny change signals at the initial stage of an insulation fault, and improve the detection accuracy of the insulation fault. And the insulation fault detection sensitivity of the switch cabinet is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of switch cabinet online monitoring, and in particular to a method and system for online monitoring of insulation fault characteristic gas in a switch cabinet. Background Art

[0002] As a key equipment in the power unit, the stability and safety of the switchgear's operation directly affect the reliability of the entire power grid. With the continuous development and upgrading of power equipment, the requirements for the insulation performance of the switchgear are also increasing. Traditionally, the detection of switchgear insulation faults mainly relies on regular maintenance and offline testing. However, this method has obvious limitations. It cannot monitor the equipment status in real time and can often only take remedial measures after the fault occurs. In recent years, with the advancement of sensor technology, data processing technology and communication technology, the online monitoring method of switchgear insulation fault characteristic gas has gradually become a research hotspot.

[0003] Although the existing online monitoring technology for characteristic gases of switchgear insulation faults has made certain progress, it still has some shortcomings. First, existing monitoring equipment usually uses a single type of sensor to measure gas concentration, which is difficult to fully reflect the complex situation of insulation faults. Most current monitoring equipment lacks intelligent management functions and cannot achieve remote monitoring and automatic diagnosis, which increases the workload of operation and maintenance personnel. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for online monitoring of characteristic gases of switch cabinet insulation faults to solve the problem that detection of switch cabinet insulation faults mainly relies on regular maintenance and offline testing and cannot monitor the equipment status in real time.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for online monitoring of characteristic gases of insulation faults in switchgear, comprising: installing a nanomaterial sensor and applying an adaptive sensitivity adjustment mechanism to optimize the accuracy of monitoring by the nanomaterial sensor;

[0008] Use sensors and optimized nanomaterial sensors to collect environmental parameters and gas concentration data inside the switchgear and send them to edge computing devices for preprocessing;

[0009] Use a deep neural network (DNN) model to deeply analyze pre-processed environmental parameter and gas concentration data, identify the differences between normal operating conditions and different types of insulation failure modes, and establish a failure mode database;

[0010] Use the fault prediction algorithm to calculate the switchgear fault prediction value and classify the insulation fault risk level that will occur in the future in combination with the failure mode differences;

[0011] Based on the failure mode database, the maintenance strategy is optimized using reinforcement learning mechanism to dynamically evaluate the health status of power equipment and formulate maintenance plans.

[0012] As a preferred solution of the method for online monitoring of characteristic gases of switchgear insulation faults described in the present invention, the application of the adaptive sensitivity adjustment mechanism to optimize the accuracy of nanomaterial sensor monitoring includes installing nanomaterial sensors at key positions of the switchgear according to the design of the switchgear and the locations of common fault points; integrating the adaptive sensitivity adjustment mechanism into the nanomaterial sensor, and calculating the sensitivity adjustment factor to optimize the sensitivity of the nanomaterial sensor.

[0013] As a preferred solution of the method for online monitoring of characteristic gases of insulation faults in switch cabinets described in the present invention, the method comprises: using sensors and optimized nanomaterial sensors to collect environmental parameters and gas concentration data in the switch cabinet, including using temperature, humidity sensors and nanomaterial sensors to collect environmental parameters and various gas concentration data in the switch cabinet in real time, and converting them into electrical signals; uploading the environmental parameters and various gas concentration data to the edge computing device through a secure channel and preprocessing them; and sending the preprocessed environmental parameters and various gas concentration data to the cloud server.

[0014] As a preferred solution of the switchgear insulation fault characteristic gas online monitoring method described in the present invention, the in-depth analysis of the preprocessed environmental parameters and gas concentration data using a deep neural network (DNN) model includes training the DNN model using environmental parameters and gas concentration historical data sets according to the deep neural network (DNN) model, and minimizing the loss function using a backpropagation algorithm and an Adam optimizer; and inputting the preprocessed environmental parameters and various gas concentration data into the trained deep neural network (DNN) model for in-depth analysis.

[0015] As a preferred solution of the switchgear insulation fault characteristic gas online monitoring method of the present invention, wherein: the identification of the difference between the normal operating state and different types of insulation fault modes and the establishment of a fault mode database include using a K-means clustering algorithm to perform a preliminary analysis of environmental parameters and historical gas concentration data, identifying different operating states and potential fault modes, and assigning environmental parameter and gas concentration data points to the nearest cluster center by minimizing intra-cluster variance; using principal component analysis (PCA) technology to extract key features; using a support vector machine (SVM) to distinguish between the normal operating state and different types of insulation fault modes; and storing the fault modes in the fault mode database based on the distinction results;

[0016] Among them, failure modes include the changing trend of specific gas concentration, temperature, humidity and maintenance measures;

[0017] Among them, the key features include time series features and spectrum features.

[0018] As a preferred solution of the switchgear insulation fault characteristic gas online monitoring method described in the present invention, the method comprises: using a fault prediction algorithm to calculate the switchgear fault prediction value, and dividing the future insulation fault risk level in combination with the failure mode difference, including using environmental parameters and gas concentration historical data to train a long short-term memory network LSTM model to capture the long-term dependency in time series data, and using a logistic regression model to evaluate the failure probability at a future time point; based on the output of the LSTM model, further predicting the development speed and possible time point of the fault, and generating a fault prediction report according to the LSTM model and the fault development speed.

[0019] As a preferred solution of the method for online monitoring of characteristic gas of insulation fault of switch cabinet described in the present invention, the method comprises: dynamically evaluating the health status of power equipment and formulating a maintenance plan by optimizing the maintenance strategy based on the fault mode database using a reinforcement learning mechanism, including obtaining the health status of the switch cabinet based on the difference between the actual value and the ideal value of the monitoring parameter under normal operating state of the switch cabinet; optimizing the maintenance strategy based on the health status of the equipment and the relevant fault information in the fault mode database using a reinforcement learning mechanism, tailoring a maintenance plan for the equipment, regularly updating the health status and maintenance plan, and collecting feedback information after the actual maintenance is performed.

[0020] In a second aspect, the present invention provides an online monitoring system for insulation fault characteristic gas of a switch cabinet, comprising:

[0021] The data acquisition module uses nanomaterial sensors to collect real-time environmental parameters and gas concentration data inside the switch cabinet;

[0022] The data prediction module pre-processes the data received from the nanomaterial sensor, including denoising and outlier detection;

[0023] A database building module uses a support vector machine (SVM) to distinguish between normal operating conditions and different types of insulation failure modes and stores the results in a failure mode database;

[0024] The fault prediction module further predicts the development speed and possible time point of the fault based on the output of the LSTM model;

[0025] The health management module tailors a management and maintenance plan for each device based on the device's health status score and classification results, combined with relevant information in the failure mode database.

[0026] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for online monitoring of characteristic gas of insulation fault of switch cabinet as described in the first aspect of the present invention is implemented.

[0027] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for online monitoring of characteristic gas of insulation fault of a switch cabinet as described in the first aspect of the present invention is implemented.

[0028] The beneficial effects of the present invention are: the highly sensitive and selective nanomaterial sensor realizes high-precision detection of multiple characteristic gases. This nanomaterial sensor can not only identify trace amounts of fault characteristic gases, but also has high selectivity and can effectively distinguish different types of fault gas components. Its function is to provide early warning and can capture tiny change signals at the early stage of insulation faults, thereby improving the sensitivity of switchgear insulation fault detection, allowing operation and maintenance personnel to take preventive measures before the fault expands, thereby avoiding the occurrence of major accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 Flowchart of a method for online monitoring of characteristic gas of insulation fault in a switch cabinet according to an embodiment.

[0031] Figure 2 Schematic diagram of fault mode recognition of a method for online monitoring of characteristic gas for insulation faults in a switch cabinet in an embodiment.

[0032] Figure 3 Schematic diagram of fault prediction of a method for online monitoring of characteristic gas of insulation fault in a switch cabinet in an embodiment.

[0033] Figure 4 Schematic diagram of fault health management of an online monitoring method for characteristic gas of insulation faults in a switch cabinet in an embodiment. DETAILED DESCRIPTION

[0034] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0035] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0036] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0037] Example 1, with reference to Figures 1-4 , is an embodiment of the present invention, which provides a method for online monitoring of characteristic gas of insulation fault in a switch cabinet, comprising:

[0038] S1: Install nanomaterial sensors and apply adaptive sensitivity adjustment mechanism to optimize the accuracy of nanomaterial sensor monitoring.

[0039] It should be noted that the design drawings of the switch cabinet were obtained from the design department and analyzed to understand the internal structure layout, the specific locations of the cable connections and the high-voltage disconnectors. According to the locations of common fault points, the installation points of the nanomaterial sensors were set near the cable connections and the high-voltage disconnector areas. The actual positioning was carried out inside the switch cabinet according to the location of the installation area and marked with a marker.

[0040] Lay the data transmission cables of the nanomaterial sensors along the pre-planned paths, ensuring that the lines are neat and do not affect the operation of other components. Use cable ties and cable troughs to secure the cables during laying to ensure that all connections are firm and reliable and that there are no loose cables or other equipment damage.

[0041] Fix the nanomaterial sensor in the previously marked position and secure it with screws. Connect the power cord and data transmission cable of the nanomaterial sensor to the corresponding interfaces. After ensuring that all connectors are properly connected and not loose, connect the nanomaterial sensor to the entire online monitoring data acquisition unit to ensure that it can work together with other components to form a complete monitoring network.

[0042] Furthermore, the installed nanomaterial sensor is adaptively adjusted for sensitivity, and the three influencing factors of temperature, humidity, and electromagnetic interference are considered. In order to ensure that the nanomaterial sensor can maintain the best working state under different environments, the sensitivity adjustment factor is calculated and expressed as:

[0043]

[0044] Among them, S r represents the sensitivity adjustment factor, T represents the absolute temperature value, H represents the relative humidity, and E represents the electromagnetic interference intensity;

[0045] Furthermore, by calculating the square root of the sum of the squares of these three variables, a value reflecting the degree of environmental impact on the performance of the nanomaterial sensor is obtained. The smaller the sensitivity adjustment factor, the greater the external interference, and the need to increase the sensitivity of the nanomaterial sensor. Conversely, the sensitivity should be reduced to reduce the false alarm rate. By continuously optimizing the sensitivity adjustment strategy of the nanomaterial sensor, the monitoring performance of the nanomaterial sensor has been significantly improved.

[0046] S2: Use sensors and optimized nanomaterial sensors to collect environmental parameters and gas concentration data inside the switch cabinet and send them to the edge computing device for preprocessing.

[0047] It should be noted that the optimized nanomaterial sensor array is started and the sampling frequency is set, for example, once per minute, to ensure that small changes in temperature, humidity, and gas concentration can be captured. The temperature, humidity, and gas concentration data are collected in real time through the temperature sensor, humidity sensor, and optimized nanomaterial sensor. The relationship between the voltage output by the optimized nanomaterial sensor and the measured temperature, humidity, and gas concentration data is calculated to obtain an electrical signal, which is expressed as:

[0048] V=k·X+b

[0049] Where X represents the actual measured value such as temperature and humidity, k and b represent the calibration coefficients of the nanosensor, and V represents the electrical signal;

[0050] Furthermore, the TLS encryption protocol is used to ensure that the temperature, humidity, and gas concentration data will not be leaked during the transmission from the optimized nanomaterial sensor to the edge computing device. The collected temperature, humidity, and gas concentration data are uploaded to the edge computing device through the above-mentioned secure channel. A low-pass filter is applied to eliminate random noise in the signal. The outliers identified by statistical methods are eliminated using the Z-score method. When the absolute value of the Z-score of a measurement value exceeds a certain value, the value is considered an outlier and is eliminated, which is expressed as:

[0051]

[0052] Among them, μ represents the sample mean, σ represents the standard deviation, and x i Indicates the measured value, Z i represents the standard score of a data point, Z represents the standard score, and i represents the data point;

[0053] Furthermore, the pre-processed temperature, humidity and gas concentration data are packaged into a JSON format that is easy to transmit, and the data packets are sent to the cloud server via the HTTP protocol.

[0054] S3. Use a deep neural network (DNN) model to perform in-depth analysis on the pre-processed environmental parameters and gas concentration data, identify the differences between normal operating conditions and different types of insulation failure modes, and establish a failure mode database.

[0055] It should be noted that a DNN model structure consisting of an input layer, multiple hidden layers and an output layer is used. Each hidden layer uses a ReLU activation function to increase the nonlinear expression capability. The input layer receives preprocessed temperature, humidity and gas concentration data, and the output layer predicts whether there is an insulation fault and its type.

[0056] The temperature, humidity and gas concentration data collected by the temperature sensor, humidity sensor and optimized nanomaterial sensor are divided into training set, test set and validation set. The DNN model weights are randomly initialized using a standard normal distribution with a mean of 0 and a standard deviation of 1. The temperature, humidity and gas concentration data collected by the optimized nanomaterial sensor are input into the initialized DNN model for training.

[0057] Input the preprocessed temperature, humidity, and gas concentration data into the trained DNN model to obtain the current temperature, humidity, and gas concentration data. Based on whether the current temperature, humidity, and gas concentration data indicate normal operation or insulation failure, the activation values ​​of each layer of the DNN model are checked to distinguish between normal and faulty states.

[0058] Furthermore, the optimal number of clusters k is determined based on the elbow method. For example, by plotting the total intra-cluster variance graph under different k values, the k value at the inflection point is found as the optimal number of clusters. K data points are selected as the initial cluster centers. Based on the distance between each temperature, humidity, and gas concentration data point and each cluster center, the temperature, humidity, and gas concentration data points are assigned to the center of the nearest cluster, and the average value of all temperature, humidity, and gas concentration data points in the cluster is used as the new cluster center. Based on K-means clustering, PCA technology is used to extract time series features and spectral features. The mean and standard deviation of the time series data in the time series features are calculated, expressed as:

[0059]

[0060] Among them, C represents the average value of the entire time series data, N represents the total number of data points in the time series, and x t Represents the data value at time point t, R represents the standard deviation of the time series data, and t represents the time point;

[0061] Use the spectrum feature to perform Fourier transform on the time series data and extract the characteristic power spectrum density in the frequency domain, which is expressed as:

[0062]

[0063] Among them, X k represents the frequency domain, e represents the complex exponential function;

[0064] Furthermore, these features can capture long-term trends and periodic changes in the data, which helps to accurately identify the differences in fault phenomena between normal operating conditions and different types of fault modes. The time series features and spectral features extracted using the PCA technology are used to train a support vector machine (SVM) model. The SVM model maximizes the intervals between different categories by finding a hyperplane and classifies different insulation faults according to the fault phenomena. According to the classification results of the SVM model, including partial discharge, overheating fault, arc fault, and moisture intrusion, each fault type is stored in the fault mode database. Each fault mode entry should contain different gas concentration change trends and fault development speeds.

[0065] S4: Use the fault prediction algorithm to calculate the switchgear fault prediction value and classify the insulation fault risk level that will occur in the future in combination with the failure mode differences.

[0066] It should be noted that the use of the LSTM model is the same as above. The temperature, humidity and gas concentration data are divided into a training set, a test set and a validation set, and the training set is used to train the LSTM model; the validation set is input into the trained LSTM model, and the LSTM model outputs the temperature prediction value, humidity prediction value and gas concentration prediction value. At the same time, the temperature prediction value, humidity prediction value and gas concentration prediction value output by the LSTM model are passed as input to the logistic regression model. The logistic regression model predicts the probability of whether the switch cabinet will fail at a certain point in the future.

[0067] Furthermore, the probability of a switchgear failure at a certain point in the future is divided into different levels of failure risk. For example, a failure output probability less than 0.3 is considered low risk, a failure output probability between 0.3 and 0.7 is considered medium risk, and a failure output probability greater than 0.7 is considered high risk.

[0068] The rate of change of characteristic gas concentration is calculated using the difference method and is expressed as:

[0069]

[0070] Among them, t1 and t2 represent two consecutive time points; and Expressed as the corresponding gas concentration.

[0071] Furthermore, the rate of change in each time period is plotted into a graph to observe the increasing and decreasing trends of the rate of change over time. Combined with the fault development speed, the fault development speed trend is obtained. A detailed fault prediction report is formed based on the fault development speed trend, gas concentration trend and fault risk level.

[0072] S5: Based on the failure mode database, the maintenance strategy is optimized using reinforcement learning mechanism to dynamically evaluate the health status of power equipment and formulate maintenance plans.

[0073] It should be noted that the health status of the switchgear is obtained by comparing the actual values ​​of various monitoring parameters of the nanosensors under normal operating conditions with the ideal values. These monitored parameters include temperature, humidity and the concentrations of various gases, such as carbon monoxide (CO), carbon dioxide (CO2) and methane (CH4).

[0074] Furthermore, the electrical fault information and insulation fault information of the switchgear in the fault mode database are used to understand the failure modes and recommended maintenance measures of the switchgear in a healthy state. At the same time, based on the health status of the switchgear, combined with the fault mode database and fault prediction report, a maintenance plan is tailored for each device, including cleaning, adjustment, and replacement of parts. After each maintenance, feedback information on the actual maintenance is collected, including maintenance results and new problems discovered.

[0075] Furthermore, the effectiveness and accuracy of future maintenance plans can be adjusted and improved based on feedback information. In addition, a reinforcement learning mechanism optimizes maintenance strategies, using historical maintenance records and the health status of the switchgear to initialize the RL model's health status. By continuously inputting new maintenance feedback information into the RL model, the RL model can learn which maintenance measures are more effective under specific conditions. For example, for a certain type of insulation fault, using a certain cleaning and component replacement method is more effective than other methods. As more switchgear electrical fault information and insulation fault information is collected and analyzed, the RL model will gradually adapt to the specific operating environment and aging level of different equipment. This method combined with reinforcement learning enables switchgear insulation fault monitoring to move from simple fault detection to intelligent management.

[0076] In summary, the present invention achieves high-precision detection of multiple characteristic gases by installing a group of nanomaterial sensors inside the switchgear that are highly sensitive and selective to the characteristic gases generated by insulation faults. This nanomaterial sensor can not only identify trace amounts of fault characteristic gases, but also has a high degree of selectivity and can effectively distinguish different types of fault gas components. Its function is to provide early warning and can capture tiny change signals at the early stage of insulation faults, thereby improving the sensitivity of switchgear insulation fault detection. This allows operation and maintenance personnel to take preventive measures before the fault escalates, thereby avoiding the occurrence of major accidents.

[0077] This embodiment also provides an online monitoring system for insulation fault characteristic gas in a switch cabinet, comprising:

[0078] The data acquisition module uses nanomaterial sensors to collect real-time environmental parameters and gas concentration data inside the switch cabinet;

[0079] The data prediction module pre-processes the data received from the nanomaterial sensor, including denoising and outlier detection;

[0080] A database building module uses a support vector machine (SVM) to distinguish between normal operating conditions and different types of insulation failure modes and stores the results in a failure mode database;

[0081] The fault prediction module further predicts the development speed and possible time point of the fault based on the output of the LSTM model;

[0082] The health management module tailors a management and maintenance plan for each device based on the device's health status score and classification results, combined with relevant information in the failure mode database.

[0083] This embodiment also provides a computer device, which is suitable for the case of an online monitoring method for characteristic gas of insulation fault in a switch cabinet, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the online monitoring method for characteristic gas of insulation fault in a switch cabinet as proposed in the above embodiment.

[0084] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0085] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for online monitoring of switchgear insulation fault characteristic gas as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0086] Example 2 is an embodiment of the present invention. In order to further verify the technical solution of the present invention, experimental simulation data of the file encryption method is provided.

[0087] To verify the effectiveness of an online monitoring method for switchgear insulation fault characteristic gases based on nanomaterial sensors, two existing technologies were selected as comparison benchmarks: one is a detection system based on traditional electrochemical sensors, and the other is a gas analyzer based on the infrared absorption principle (now). First, the design drawings of the medium-voltage switchgear for testing were obtained and the sensors were installed according to the design drawings. According to the design requirements, multiple nanomaterial sensors were set up near the cable connection and the high-voltage disconnector area to ensure that any possible insulation fault characteristic gases could be effectively captured. Specifically:

[0088] A total of 22 nanomaterial sensors (12 at cable connections and 10 in the high-voltage disconnector area) were installed near the cable connections and high-voltage disconnector areas to achieve comprehensive monitoring of potential insulation faults.

[0089] During the sensor installation process, we strictly followed the design requirements and used cable ties and cable ducts to secure the data transmission cables to ensure that all connectors were firm and reliable. After the hardware installation was completed, adaptive sensitivity adjustment was performed. Considering the impact of temperature, humidity, and electromagnetic interference on sensor performance, the sensitivity adjustment factor was calculated to optimize the sensor's working state in different environments.

[0090] Subsequently, the optimized nanomaterial sensor array was started, and the sampling frequency was set to once per minute to ensure real-time monitoring of changes in temperature, humidity, and gas concentration. The TLS encryption protocol was used to ensure the security of data transmission, and the collected data was pre-processed through edge computing devices, including steps such as removing outliers.

[0091] During the data analysis phase, a deep neural network model (DNN) is used to conduct in-depth analysis of the preprocessed data to identify the differences between normal operating conditions and different types of insulation failure modes. K-means clustering combined with PCA technology is used to extract time series and spectral features, and a support vector machine (SVM) model is trained for fault classification. Finally, a long short-term memory (LSTM) network is used to predict the risk level of future insulation faults, and a personalized maintenance plan is developed to achieve dynamic assessment of the health status of the switchgear.

[0092] The details are shown in Table 1 below:

[0093] Table 1 Performance comparison of switchgear insulation fault characteristic gas online monitoring system

[0094]

[0095] As can be seen from Table 1, the present invention outperforms the prior art 1 and the prior art 2 in many key indicators. For example, in terms of temperature and humidity measurement accuracy, the present invention achieves high accuracies of ±0.16°C and ±1.08%RH, respectively, exceeding the prior art 1 (±0.52°C, ±3.1%RH) and the prior art 2 (±0.34°C, ±2.2%RH). This shows that the present invention can provide more accurate readings under a wider range of environmental conditions, thereby improving the accuracy of fault detection.

[0096] In terms of fault detection rate, the present invention achieves a high detection rate of 94.7%. In comparison, the prior art 1 and prior art 2 have detection rates of 74.8% and 81.2%, respectively. This means that the present invention can more effectively identify potential insulation faults and reduce the risk of undetected faults.

[0097] Response time is crucial for promptly handling possible problems. The present invention shortens the response time to 14.9 seconds, while prior art 1 and prior art 2 are 58.3 seconds and 44.6 seconds respectively. The rapid response capability helps to take immediate action to prevent accidents from escalating.

[0098] Regarding data transmission security, the present invention not only complies with the ISO / IEC 27001 standard, but also meets the requirements of NIST SP800-53, providing higher security than existing technologies. In addition, the maintenance cycle of the present invention is as long as 12 months, which exceeds the existing technology 1 (6 months) and the existing technology 2 (5 months), greatly reducing operation and maintenance costs.

[0099] In terms of cost, although the initial investment of the present invention is slightly higher than that of prior art 1 (￥35,200.58) and prior art 2 (￥49,850.74), its total cost of ownership is more economical due to its longer maintenance cycle and higher efficiency. Specifically, the cost of the present invention is ￥45,789.26. Although slightly higher than prior art 1, this investment is very worthwhile due to the high-level functions and long-term benefits it provides.

[0100] In summary, the present invention demonstrates significant advantages in both technical performance and economic benefits. By integrating an adaptive sensitivity adjustment mechanism, deep neural network analysis, and reinforcement learning-based maintenance optimization strategies, it provides a more intelligent, efficient, and economical solution for switchgear insulation fault monitoring. These improvements make power system operation safer and more reliable while also reducing unnecessary downtime and maintenance costs.

[0101] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for online monitoring of characteristic gases of insulation faults in switch cabinets, characterized by: include: Install nanomaterial sensors and apply adaptive sensitivity adjustment mechanisms to optimize the accuracy of nanomaterial sensor monitoring; Use sensors and optimized nanomaterial sensors to collect environmental parameters and gas concentration data inside the switchgear and send them to edge computing devices for preprocessing; Use a deep neural network (DNN) model to deeply analyze pre-processed environmental parameter and gas concentration data, identify the differences between normal operating conditions and different types of insulation failure modes, and establish a failure mode database; Use the fault prediction algorithm to calculate the switchgear fault prediction value and classify the insulation fault risk level that will occur in the future in combination with the failure mode differences; Based on the failure mode database, the maintenance strategy is optimized using reinforcement learning mechanism to dynamically evaluate the health status of power equipment and formulate maintenance plans.

2. The method for online monitoring of characteristic gas for insulation faults in switch cabinets according to claim 1, characterized in that: The application of the adaptive sensitivity adjustment mechanism to optimize the accuracy of nanomaterial sensor monitoring includes installing nanomaterial sensors at key locations of the switch cabinet based on the design of the switch cabinet and the locations of common fault points; integrating the adaptive sensitivity adjustment mechanism into the nanomaterial sensor, and calculating the sensitivity adjustment factor to optimize the sensitivity of the nanomaterial sensor.

3. The method for online monitoring of characteristic gas of insulation fault in switch cabinet according to claim 2, characterized in that: The use of sensors and optimized nanomaterial sensors to collect environmental parameters and gas concentration data in the switch cabinet includes using temperature, humidity sensors and nanomaterial sensors to collect environmental parameters and various gas concentration data in the switch cabinet in real time, and converting them into electrical signals; uploading the environmental parameters and various gas concentration data to the edge computing device through a secure channel and preprocessing them; and sending the preprocessed environmental parameters and various gas concentration data to the cloud server.

4. The method for online monitoring of characteristic gas of insulation fault in switch cabinet according to claim 3, characterized in that: The use of a deep neural network (DNN) model to perform in-depth analysis on preprocessed environmental parameters and gas concentration data includes training the DNN model using environmental parameters and gas concentration historical data sets according to the deep neural network (DNN) model, minimizing the loss function using a backpropagation algorithm and an Adam optimizer; and inputting the preprocessed environmental parameters and various gas concentration data into the trained deep neural network (DNN) model for in-depth analysis.

5. The method for online monitoring of characteristic gas of insulation fault in switch cabinet according to claim 4, characterized in that: The identification of differences between normal operating states and different types of insulation failure modes and the establishment of a failure mode database includes using a K-means clustering algorithm to perform a preliminary analysis of environmental parameter and historical gas concentration data, identifying different operating states and potential failure modes, and assigning environmental parameter and gas concentration data points to the nearest cluster center by minimizing intra-cluster variance; and extracting key features using principal component analysis (PCA) technology. Use support vector machines (SVMs) to distinguish between normal operating conditions and different types of insulation failure modes; store the failure modes in a failure mode database based on the distinction results; Among them, failure modes include the changing trend of specific gas concentration, temperature, humidity and maintenance measures; Among them, the key features include time series features and spectrum features.

6. The method for online monitoring of characteristic gas of insulation fault in switch cabinet according to claim 5, characterized in that: The method of using a fault prediction algorithm to calculate the switchgear fault prediction value and classify the insulation fault risk level that will occur in the future in combination with the failure mode differences includes using environmental parameters and gas concentration historical data to train a long short-term memory network (LSTM) model to capture the long-term dependencies in time series data, and using a logistic regression model to evaluate the failure probability at a future time point; based on the output of the LSTM model, further predicting the development speed and possible time point of the fault, and generating a fault prediction report based on the LSTM model and the fault development speed.

7. The method for online monitoring of characteristic gas of insulation fault in switch cabinet according to claim 6, characterized in that: The method of dynamically evaluating the health status of power equipment and formulating a maintenance plan by optimizing the maintenance strategy using a reinforcement learning mechanism based on the failure mode database includes obtaining the health status of the switchgear based on the difference between the actual value and the ideal value of the monitoring parameter of the switchgear under normal operating conditions; Based on the health status of the equipment and the relevant fault information in the failure mode database, the maintenance strategy is optimized using the reinforcement learning mechanism, a maintenance plan is tailored for the equipment, the health status and maintenance plan are updated regularly, and feedback information is collected after the actual maintenance is performed.

8. An online monitoring system for characteristic gas of insulation fault in a switch cabinet, based on the online monitoring method for characteristic gas of insulation fault in a switch cabinet according to any one of claims 1 to 7, characterized in that: include, The data acquisition module uses nanomaterial sensors to collect real-time environmental parameters and gas concentration data inside the switch cabinet; The data prediction module pre-processes the data received from the nanomaterial sensor, including denoising and outlier detection; A database building module uses a support vector machine (SVM) to distinguish between normal operating conditions and different types of insulation failure modes and stores the results in a failure mode database; The fault prediction module further predicts the development speed and possible time point of the fault based on the output of the LSTM model; The health management module tailors a management and maintenance plan for each device based on the device's health status score and classification results, combined with relevant information in the failure mode database.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for online monitoring of characteristic gas of insulation fault in a switch cabinet according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for online monitoring of characteristic gas of insulation fault in a switch cabinet according to any one of claims 1 to 7 are implemented.

Images