Real-time monitoring method and device for light gas

Abstract

The invention discloses a light gas real-time monitoring method, and relates to the field of transformer oil tank fault detection and diagnosis, and the method comprises the steps: employing a sampling probe to collect light gas sample data according to a preset time interval, and transmitting the data to a data processing center; the data processing center preprocesses the light gas sample data; the data processing center analyzes the collected light gas sample data based on a fault diagnosis algorithm of a deep learning model; and in response to analysis of the light gas sample, sending out an early warning signal and providing a fault diagnosis report according to an analysis result. Collected light gas sample data are analyzed through a fault diagnosis algorithm based on a deep learning model, an early warning signal is sent out and a fault diagnosis report is provided according to an analysis result, so that light gas is monitored in real time, and early warning and accurate judgment are made according to a monitoring result.

Description

Technical Field

[0001] The invention relates to the field of transformer oil tank fault detection and diagnosis, and in particular to a real-time monitoring method and a real-time monitoring device for light gas. Background Art

[0002] If the internal fault of the transformer oil tank is relatively minor or in the early stage of the fault, the oil in the tank is decomposed and vaporized, and a small amount of gas accumulates on the top of the gas relay. When the gas volume reaches the set value, an alarm signal is issued to prompt maintenance personnel to check and prevent the development of the fault.

[0003] Traditional light gas monitoring mainly relies on gas relays, which have a relatively simple principle and can only sound an alarm when the gas accumulates to a certain amount, and cannot accurately identify subtle changes in gas composition and content. However, the time process of the accumulation of undissolved gas in the insulating liquid is precisely an important criterion for judging faults. Therefore, traditional light gas monitoring is difficult to monitor faults in real time, achieve early warning and accurate judgment.

[0004] In order to solve the above problems, the present invention provides a real-time monitoring method and device for light gas. Summary of the invention

[0005] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or critical elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0006] In order to overcome the above-mentioned defects, the present invention aims to provide a real-time monitoring method for light gas to solve the problem that traditional light gas monitoring mainly relies on gas relays, which makes it difficult to monitor faults in real time and achieve early warning and accurate judgment.

[0007] According to one aspect of the present invention, a method for real-time monitoring of light gas is provided, comprising:

[0008] Using the sampling probe to collect light gas sample data at preset time intervals to the data processing center;

[0009] The data processing center pre-processes the light gas sample data;

[0010] The data processing center analyzes light gas sample data based on the deep learning model;

[0011] In response to the analysis of the light gas sample, an early warning signal is issued based on the analysis results.

[0012] In one embodiment, the light gas sample data includes location data, diffusion velocity data, and gas composition data of the light gas sample in the transformer oil tank.

[0013] In one embodiment, the deep learning model uses a machine learning algorithm to train actual data to establish a mathematical model between changes in light gas components and internal fault types of the transformer oil tank.

[0014] In one embodiment, the data processing center analyzes the light gas sample data based on the deep learning model, including:

[0015] The deep learning model includes gas component concentration parameters, gas ratio parameters, temperature parameters, gas generation rate parameters and time parameters;

[0016] The light gas sample data is input into the deep learning model, and the corresponding internal fault type of the transformer tank and the probability of the current fault type are output.

[0017] In one embodiment, the data processing center analyzes the light gas sample data based on the deep learning model further comprising:

[0018] Adjust various parameters in the deep learning model according to actual data;

[0019] The adjusted parameters of the deep learning model are determined as the current deep learning model.

[0020] In one embodiment, in response to analyzing the light gas sample, issuing an early warning signal according to the analysis result includes:

[0021] Set the probability threshold range of the fault type to 50%;

[0022] In response to the outputted predicted probability of the current fault type being greater than or equal to 50%, an early warning signal is issued to warn of the existence of the current fault type;

[0023] When the predicted probability of the fault type output in response is less than 50%, it is considered that there is no current fault type.

[0024] In one embodiment, in response to the analysis of the light gas sample, issuing a warning signal according to the analysis result further includes:

[0025] Set warning levels according to different fault types and the probability of fault types, and send different warning types according to different warning levels.

[0026] In one embodiment, the data processing center pre-processes the light gas sample data including:

[0027] Setting a threshold value of light gas sample data to remove abnormal values ​​in the light gas sample data;

[0028] Normalize the light gas sample data;

[0029] Analyze the gas composition data in the light gas sample data.

[0030] In one embodiment, the preset time interval is dynamically adjusted according to the operating status of the transformer oil tank.

[0031] According to another aspect of the present invention, a real-time monitoring device for light gas is provided, which is characterized in that it is used to implement the real-time monitoring method for light gas as described in any of the above embodiments, and comprises a sampling probe, a data processing center and a control center.

[0032] The sampling probe is used to collect light gas sample data.

[0033] The data processing center processes and analyzes the light gas sample data.

[0034] The control center issues early warning signals and provides fault diagnosis reports based on the analysis results of the data processing center.

[0035] The present invention provides a real-time monitoring method and device for light gas. The method analyzes the collected light gas sample data through a fault diagnosis algorithm based on a deep learning model, and issues a warning signal and provides a fault diagnosis report according to the analysis result, thereby achieving real-time monitoring of light gas, and achieving rapid and accurate judgment of the fault type and corresponding alarm in the early stage of the fault, effectively avoiding the maintenance personnel from checking and determining the corresponding fault type when it is found that the light gas exceeds the threshold range, thereby effectively saving human resource costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments disclosed in the present invention in conjunction with the following drawings.

[0037] Figure 1 is a flow chart of a method for real-time monitoring of light gas in an embodiment according to one aspect of the present invention;

[0038] Figure 2 It is a partial flow diagram of a method for real-time monitoring of light gas in an embodiment according to one aspect of the present invention;

[0039] Figure 3 It is a partial flow diagram of a method for real-time monitoring of light gas in an embodiment according to one aspect of the present invention;

[0040] Figure 4 It is a partial flow diagram of a method for real-time monitoring of light gas in an embodiment according to one aspect of the present invention;

[0041] Figure 5 It is a schematic block diagram of modules of a real-time monitoring device for light gas in an embodiment according to another aspect of the present invention. DETAILED DESCRIPTION

[0042] The following description is given to enable those skilled in the art to implement and use the invention and incorporate it into a specific application context. Various modifications, as well as various uses in different applications will be readily apparent to those skilled in the art, and the general principles defined herein are applicable to a wide range of embodiments. Thus, the present invention is not limited to the embodiments given herein, but should be granted the broadest scope consistent with the principles and novel features disclosed herein.

[0043] In the following detailed description, many specific details are set forth to provide a more thorough understanding of the present invention. However, it will be apparent to those skilled in the art that the practice of the present invention need not be limited to these specific details. In other words, known structures and devices are shown in block diagram form without detailed display to avoid obscuring the present invention.

[0044] Unless otherwise expressly stated, all features disclosed in this specification (including any accompanying claims, abstracts and drawings) may be replaced by alternative features that achieve the same, equivalent or similar purpose. Therefore, unless expressly stated otherwise, each feature disclosed is only an example of a group of equivalent or similar features.

[0045] Note that, where used, the symbols left, right, front, back, top, bottom, forward, reverse, clockwise and counterclockwise are used only for convenience and do not imply any specific fixed direction. In fact, they are used to reflect the relative position and / or direction between the various parts of the object. In addition, the terms "first" and "second" are used only for descriptive purposes and should not be understood as indicating or implying relative importance.

[0046] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "connected", "connected", and "cross-connected" should be understood in a broad sense, for example, it can be directly connected, or indirectly connected through an intermediate medium, or it can be the internal connection of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] Note that, in the case of use, further, preferably, further and more preferably are simple beginnings for explaining another embodiment based on the previous embodiment, and the content of further, preferably, further or more preferably followed by the above embodiment is combined with the previous embodiment as a complete composition of another embodiment. Several further, preferably, further or more preferably settings followed by the same embodiment can be arbitrarily combined to form another embodiment.

[0048] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Note that the aspects described below in conjunction with the accompanying drawings and specific embodiments are only exemplary and should not be construed as limiting the scope of protection of the present invention in any way.

[0049] According to one aspect of the present invention, a real-time monitoring method for light gas is provided, which analyzes the collected light gas sample data by a fault diagnosis algorithm based on a deep learning model, and issues a warning signal and provides a fault diagnosis report based on the analysis result, thereby monitoring the light gas in real time, and making early warnings and accurate judgments based on the monitoring results.

[0050] Figure 1 FIG. 1 shows a flow chart of a method for real-time monitoring of light gas in an embodiment, such as Figure 1 As shown, the real-time monitoring method of light gas may include steps S1 to S4.

[0051] Wherein, step S1 is: using a sampling probe to collect light gas sample data to a data processing center at a preset time interval.

[0052] The sampling probe refers to the collection of gas samples from a specific location of the equipment for subsequent composition analysis and fault diagnosis. Preferably, an embedded sampling probe can be used, which is directly embedded in a key location inside the equipment, so that light gas samples that reflect the operating status of the equipment can be collected more accurately, thereby improving the accuracy of fault diagnosis.

[0053] Utilizing high-precision sampling probes, light gas sample data is accurately collected from key locations of the transformer oil tank, such as the oil pillow, gas relay connecting pipes, etc., strictly according to pre-set time intervals. The collected sample data is then completely and securely transmitted to a fully functional data processing center through a high-speed and stable data transmission line, facilitating subsequent in-depth data analysis and processing, thereby providing strong data support for accurately judging the operating status of the transformer and the early warning and diagnosis of potential faults.

[0054] In a normally operating transformer tank, the preset time interval can be set to collect light gas samples every 3 hours. This can ensure effective monitoring of the daily operating status of the equipment without placing too much burden on the monitoring work.

[0055] Preferably, the preset time interval can be set to be dynamic, such as setting the collection of light gas samples every 1 hour during the peak period of transformer oil tank use, and setting the collection of light gas samples every 3 hours during the low period of transformer oil tank use.

[0056] Step S2 is: the data processing center pre-processes the light gas sample data.

[0057] Among them, the light gas sample includes position data, diffusion velocity data and gas composition data in the transformer oil tank.

[0058] In a specific embodiment, if Figure 2 As shown, step S2 may include steps S21 to S23.

[0059] Step S21 is: setting a threshold value of the light gas sample data to remove abnormal values ​​in the light gas sample data.

[0060] According to the measured data, a reasonable threshold range is set for the data, and data points that are obviously deviated from the normal data fluctuation range and seriously inconsistent with the established threshold range standards are accurately identified, and outliers that exceed the threshold range are removed from the original light gas sample data. Outliers will interfere with subsequent data processing and analysis, and may lead to erroneous conclusions.

[0061] Step S22 is: normalizing the light gas sample data.

[0062] Use the minimum-maximum normalization method to normalize the light gas sample data to a specific range to avoid excessively high or low weights of certain variables during model training due to large differences in numerical values.

[0063] Step S23 is: analyzing the gas composition data in the light gas sample data.

[0064] Different gas components in the light gas sample correspond to different fault types. The alkane gas component corresponds to overheating fault or discharge fault. When the gas component is hydrogen, it corresponds to partial discharge fault or moisture fault. When the gas component is carbon monoxide and carbon dioxide, it corresponds to solid insulation overheating and aging fault.

[0065] Step S3 is: the data processing center analyzes the light gas sample data based on the deep learning model.

[0066] Among them, the deep learning model uses machine learning algorithms to learn and train actual data to establish a mathematical model between the changes in light gas components and the internal fault types of the transformer tank.

[0067] In a specific embodiment, if Figure 3 As shown, step S3 may include steps S31 to S32.

[0068] Among them, step S31 is: the deep learning model includes gas component concentration parameters, gas ratio parameters, temperature parameters, gas generation rate parameters and time parameters.

[0069] Preferably, the deep learning model uses a long short-term memory network to analyze the collected light gas sample data. The processed light gas sample data is divided into a training set, a validation set, and a test set.

[0070] Randomly initialize the weights and biases of the deep learning model. These initial parameters will be continuously adjusted and optimized in the subsequent training process. Organize the training set data into an input format suitable for the deep learning model, select a suitable loss function based on the nature of the analysis task, and select a suitable optimization algorithm to update the network parameters to minimize the loss function.

[0071] The training set data is sent to the deep learning model in sequence. At each time step, based on the current input and the hidden state at the previous moment, calculations are performed according to the workflow of the long short-term memory network: forgetting stage, selective memory stage, and output stage, and the data is continuously propagated forward to obtain the output results.

[0072] During the training process, regularly use the validation set data to input the deep learning model in training, calculate the corresponding evaluation indicators, and observe the performance of the deep learning model on the validation set to prevent overfitting. If it is found that the validation set indicators no longer improve or even decrease, the training can be stopped in advance and the parameters in the deep learning model can be adjusted.

[0073] Final test evaluation: After training is completed, the test set data is used to conduct a final performance evaluation of the model to obtain the generalization ability index of the deep learning model on actual unseen data and determine whether the model meets the requirements of practical applications. Finally, the gas component concentration parameters, gas ratio parameters, temperature parameters, gas generation rate parameters and time parameters in the deep learning model are obtained.

[0074] Step S32 is: inputting the light gas sample data into the deep learning model, and outputting the corresponding transformer oil tank internal fault type and the probability of the current fault type.

[0075] In a specific embodiment, step S3 also includes adjusting various parameters in the deep learning model according to actual data, where the actual data includes light gas data and corresponding fault types.

[0076] The parameters of the adjusted deep learning model are determined as the current deep learning model, making the model more accurate and practical.

[0077] Step S4 is: in response to the analysis of the light gas sample, issuing an early warning signal according to the analysis result.

[0078] In a specific embodiment, if Figure 4 As shown, step S4 may include steps S41 to S43.

[0079] Wherein, step S41 is: setting the probability threshold range of the fault type to 50%.

[0080] In the process of the above deep learning model judging the data of the collected light gas samples, it is easy to misjudge the normal state as a fault state or the fault state as a normal state, both of which need to be avoided. Setting the probability threshold to 50% helps to strike a balance between the two. In addition, when there is noise in the data, the sample size is limited, or the fault type is complex and difficult to distinguish, the deep learning model may not be very accurate in outputting the probability of the fault type. Therefore, when the probability output by the model is higher than 50%, we can think that the model is relatively more confident in judging the fault type, and there is enough evidence to show that there may be a fault.

[0081] Step S42 is: in response to the output that the predicted probability of the current fault type is greater than or equal to 50%, issuing a warning signal to warn of the existence of the current fault type.

[0082] Step S43 is: when the predicted probability of the output fault type is less than 50%, it is considered that the current fault type does not exist.

[0083] It can be understood that in response to the analysis of light gas samples, warning signals can be issued and fault diagnosis reports can be provided based on the analysis results. Warning levels can be set according to different fault types and probabilities of fault types, and different warning types can be sent according to different warning levels.

[0084] According to another aspect of the present invention, there is also provided a real-time monitoring device for light gas, which is used to implement the real-time monitoring method for light gas as described in any of the above embodiments, and includes a sampling probe, a data processing center and a control center.

[0085] The sampling probe is used to collect light gas sample data. The data processing center processes and analyzes the light gas sample data. The control center issues an early warning signal and provides a fault diagnosis report based on the analysis results of the data processing center.

[0086] Preferably, the sampling probe may be an embedded sampling probe, which monitors the generation of light gas inside the transformer oil tank in real time, promptly discovers potential fault hazards, and ensures the safe and stable operation of the transformer oil tank.

[0087] The foregoing description is provided to enable any person skilled in the art to practice the various aspects described herein. However, it should be understood that the scope of protection of the present invention shall be subject to the appended claims and shall not be limited to the specific structures and components of the above-described embodiments. Those skilled in the art may make various changes and modifications to the embodiments within the spirit and scope of the present invention, and these changes and modifications also fall within the scope of protection of the present invention.

Claims

1. A real-time monitoring method for light gas, characterized in that: include: Using the sampling probe to collect light gas sample data at preset time intervals to the data processing center; The data processing center pre-processes the light gas sample data; The data processing center analyzes the light gas sample data based on a deep learning model; In response to the analysis of the light gas sample, an early warning signal is issued based on the analysis results.

2. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: The light gas sample data includes position data, diffusion velocity data and gas composition data of the light gas sample in the transformer oil tank.

3. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: The deep learning model uses a machine learning algorithm to train actual data to establish a mathematical model between changes in light gas components and internal fault types of the transformer oil tank.

4. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: The data processing center analyzes the light gas sample data based on the deep learning model, including: The deep learning model includes gas component concentration parameters, gas ratio parameters, temperature parameters, gas generation rate parameters and time parameters; The light gas sample data is input into the deep learning model, and the corresponding transformer oil tank internal fault type and the probability of the current fault type are output.

5. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: The data processing center analyzes the light gas sample data based on the deep learning model and further includes: Adjust various parameters in the deep learning model according to actual data; The adjusted parameters of the deep learning model are determined as the current deep learning model.

6. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: In response to the analysis of the light gas sample, issuing an early warning signal according to the analysis result includes: Set the probability threshold range of the fault type to 50%; In response to the outputted predicted probability of the current fault type being greater than or equal to 50%, an early warning signal is issued to warn of the existence of the current fault type; When the predicted probability of the fault type output in response is less than 50%, it is considered that there is no current fault type.

7. The real-time monitoring method for light gas as claimed in claim 6, characterized in that: In response to the analysis of the light gas sample, issuing an early warning signal according to the analysis result also includes: Set warning levels according to different fault types and the probability of fault types, and send different warning types according to different warning levels.

8. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: The data processing center preprocesses the light gas sample data including: Setting a threshold value of light gas sample data to remove abnormal values ​​in the light gas sample data; Normalizing the light gas sample data; The gas composition data in the light gas sample data is analyzed.

9. The real-time monitoring method for light gas as claimed in claim 1, characterized in that: The preset time interval is dynamically adjusted according to the operating status of the transformer oil tank.

10. A real-time monitoring device for light gas, characterized in that: A method for real-time monitoring of light gas as claimed in any one of claims 1 to 9, comprising a sampling probe, a data processing center and a control center, The sampling probe is used to collect light gas sample data. The data processing center processes and analyzes the light gas sample data. The control center sends out an early warning signal and provides a fault diagnosis report according to the analysis result of the data processing center.

Images