Gas Relay Gas Volume Identification Method Based on Image Analysis

Abstract

The present invention relates to the technical field of gas relay monitoring, and specifically relates to a method for identifying the gas volume accumulated in a gas relay based on image analysis, which includes the following steps: S1. Obtain the volume data and liquid level height of the gas injected into the gas relay and perform analysis to obtain the conversion relationship between the liquid level height and the gas volume in the gas relay; S2. Obtain the background reference image of the observation window of the gas relay; S3. Obtain the image to be detected of the observation window of the gas relay, and combine it with the background reference image to perform image processing by means of image difference to identify the liquid level height in the observation window of the current gas relay; S4. Use the conversion relationship between the liquid level and the gas volume to convert the current liquid level height into the gas volume. This method can intelligently detect the gas accumulation situation of the gas relay for a long time and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of gas relay monitoring, and particularly to a method for identifying the gas volume in a gas relay based on image analysis. Background Art

[0002] The functions of existing gas relays rely on the actions of internal mechanical structures, including light gas warning and heavy gas protection. To dynamically monitor the gas accumulation in a gas relay, on-site manual inspections are required. If further analysis of the gas components in the relay is needed, manual live operations are required, posing serious safety hazards. There is an urgent need to develop an on-line monitoring device for gas relays to analyze the gas accumulation situation in real time.

[0003] When a large amount of gas has been decomposed from the transformer oil or a short-circuit fault occurs in the transformer, not only is the transformer system threatened by serious faults, but the areas near the transformer and the gas relay are also in danger. Therefore, it is hoped to calculate the gas volume by on-line monitoring the height of the transformer oil level in the gas relay, which can not only realize the on-line monitoring of the operating state of the transformer, but also avoid the risk of personal safety for the staff on site.

[0004] The liquid level height measurement is divided into two methods: contact type and non-contact type. Contact detection has weak adaptability to different liquids, is easily corroded by liquids, and is relatively complex to install, so it is not suitable for the liquid level measurement of gas relays. Therefore, non-contact measurement is widely used in industry at present. Its advantages are remote detection of metering measurement, long service life, and relatively simple installation. However, in mainstream non-contact measurements, the price of ultrasonic detection will increase exponentially with the improvement of accuracy, making it difficult to be widely applied. Although radar detection is not affected by changes in medium density and temperature, working conditions such as noise, steam, dust, and vacuum, is not easily corroded, and can work normally in extreme environments such as high and low pressures, its price is high and its anti-electromagnetic interference ability is weak. Moreover, for gas relays, due to their metal casings, there are serious interferences for non-contact measurements and even signal interruption may occur, and most non-contact measurement methods cannot be implemented.

[0005] Therefore, how to achieve long-term and accurate intelligent detection of the gas accumulation situation in gas relays has become an urgent problem to be solved at present. Summary of the Invention

[0006] In view of the above deficiencies of the prior art, the present invention provides a method for identifying the gas volume in a gas relay based on image analysis, which can perform long-term and accurate intelligent detection of the gas accumulation situation in the gas relay.

[0007] To solve the above technical problems, the present invention adopts the following technical solutions:

[0008] A method for identifying the gas volume in a gas relay based on image analysis includes the following steps:

[0009] S1. Obtain the volume data of the gas injected into the gas relay and the liquid level height, and perform analysis to obtain the conversion relationship between the liquid level height and the gas volume in the gas relay;

[0010] S2. Obtain the background reference image of the observation window of the gas relay;

[0011] S3. Obtain the image to be detected of the observation window of the gas relay, and combine it with the background reference image to perform image processing by means of image difference to identify the liquid level height in the observation window of the current gas relay;

[0012] S4. Use the conversion relationship between the liquid level and the gas volume to convert the current liquid level height into the gas volume.

[0013] Preferably, S1 includes:

[0014] S11. Open the exhaust hole of the gas relay without liquid inside;

[0015] S12. Send liquid into the gas relay n times from the gas release plug, and the volume of the liquid sent each time is equal; after sending the liquid each time, record the total volume V li of the liquid that has been sent, and record the corresponding liquid level height H i ;

[0016] S13. Record the total capacity of the gas relay as V, and obtain the remaining gas volume V i = V - V li ;

[0017] S14. Analyze according to the recorded total volume V li of the liquid sent, the corresponding liquid level height H i and the remaining gas volume V i to obtain the conversion relationship between the liquid level height and the gas volume in the gas relay.

[0018] Preferably, in S14, the least squares method is used for linear fitting:

[0019]

[0020]

[0021] Obtain the formula for calculating the gas volume from the liquid level height:

[0022] V 气 = k * H + b;

[0023] Among them, H is the liquid level height, V气 is the corresponding gas volume.

[0024] Preferably, in S2, the background reference image is an image with an empty liquid level in the observation window.

[0025] Preferably, S3 includes:

[0026] S31. Obtain the image to be detected of the observation window of the gas relay;

[0027] S32. Perform image difference processing on the image to be detected and the background reference image to obtain a difference image;

[0028] S33. Perform binarization and denoising processing on the difference image to obtain a denoised image;

[0029] S34. Perform hough transformation on the denoised image, detect the circular border of the observation window, and generate a circular mask with a radius smaller than the circular border according to the detection result. Use the mask to perform bitwise AND operation on the denoised image to remove the circular border of the observation window and obtain a binary image of the liquid level interface;

[0030] S35. Perform horizontal projection on the liquid level interface image to obtain a grayscale histogram in the horizontal direction; perform maximum value detection on the histogram to obtain the row number where the liquid level is located as the liquid level height in the current observation window of the gas relay.

[0031] Preferably, in S33, after obtaining the binarization threshold of the difference image by using the OTSU algorithm, perform binarization processing on the difference image to obtain an image with only black and white colors, and then perform median filtering and morphological transformation on the binary image to obtain a denoised image; wherein, the morphological transformation includes using opening operation to eliminate independent noise points at the segmentation edge and connect and thicken the liquid level information.

[0032] Preferably, in S35, when performing maximum value detection on the histogram, the measurement accuracy is 1 pixel.

[0033] Preferably, after S4, S5 is further included, to judge whether the current gas volume in the gas relay has reached a preset exhaust threshold. If so, control the automatic gas sampling system to discharge the gas in the gas relay and analyze the discharged gas.

[0034] Preferably, in S5, the automatic gas sampling system includes a three-way valve, a gas collecting box, an oil conservator, a gas detection module, and a control module; the gas pillow is connected to interface 1 of the three-way valve after passing through solenoid valve 1; the gas relay is connected to interface 2 of the three-way valve after passing through solenoid valve 2; the inlet of the gas collecting box is connected to interface 3 of the three-way valve; the exhaust port of the gas collecting box is connected to the gas detection module after passing through solenoid valve 4; a solenoid valve 3 is provided at the oil drain port of the gas collecting box; the control module is electrically connected to the gas detection module, solenoid valve 1, solenoid valve 2, solenoid valve 3, and solenoid valve 4 respectively;

[0035] The control process of the automatic gas sampling system includes: in the initial state, solenoid valve 1 is opened, and solenoid valves 2, 3, and 4 are closed; when the current gas volume in the gas relay reaches the preset exhaust threshold, the control module closes solenoid valve 1, opens solenoid valves 2 and 3 for a first preset duration; after the first preset duration, the oil level in the gas collecting box drops due to gas injection, the gas relay is refilled with oil, solenoid valve 3 is closed, solenoid valve 4 and the gas detection module are opened for a second preset duration; after the second preset duration, the gas collecting box is filled with oil and oil starts to flow out of the exhaust hole, solenoid valves 2 and 4 are closed, solenoid valve 1 is opened, and the initial state is restored.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. In the present invention, by obtaining the image to be detected of the observation window of the gas relay and combining with the background reference image, the display result of the current observation window, that is, the liquid level height in the observation window of the current gas relay, can be automatically detected by means of inter-frame image difference. Since the space inside the gas relay is entirely occupied by liquid and accumulated gas, and the internal space of the gas and electrical appliances is a fixed value, it is undoubtedly possible to know that there is a relationship conversion formula between the liquid level height in the gas relay and the gas volume. Since this conversion formula has been pre-analyzed, therefore, when the liquid level height in the observation window of the current gas relay is detected, the gas volume in the gas relay at this time can be accurately obtained through this conversion formula.

[0038] Compared with contact detection, the present application does not need to contact the liquid in the gas relay during the detection process, and the situation of corrosion of the detection end (such as the sensor probe) can be avoided. Compared with conventional non-contact detection, since the present application does not need to penetrate the external shell of the gas relay, there is no serious interference and signal interruption caused by the external shell in the invention.

[0039] In summary, the present invention can intelligently detect the accumulated gas situation of the gas relay for a long time and accurately.

[0040] 2. When analyzing the conversion relationship between the liquid level height and the gas volume in the gas relay, the present invention adopts the method of sending n volumes of liquid with equal volume into the gas relay from the gas release plug, and sequentially records the total volume V of the liquid that has been sent in li , and the corresponding liquid level height H i . Then, combined with the default condition that the space inside the gas relay is completely occupied by liquid and accumulated gas, and the internal space of the gas and the electrical appliance is a fixed value, linear fitting is performed by the least square method, and a calculation formula that can accurately reflect the conversion relationship between the liquid level height and the gas volume in the gas relay can be obtained. Thus, the accuracy of this method is ensured.

[0041] 3. When analyzing the current liquid level height, since the result of the differential operation may have interference noise due to reasons such as reflection, binarization and noise reduction are required for subsequent processing. The OTSU algorithm is used to obtain the binarization threshold of the image, the differential image is binarized to obtain an image with only black and white colors, and then the binary image is subjected to median filtering and morphological transformation to remove interference factors.

[0042] However, at this time, there is still interference introduced by the circular observation window border in the image, so further processing is required. The present invention selects to perform hough transformation on the denoised image to detect the circular border. According to the detection result, a circular mask with a radius smaller than the detected circle is generated, and the denoised image is subjected to bitwise AND operation using the mask, so that the observation window border can be removed to obtain a binary image of the liquid level interface. Then, horizontal projection is performed on the liquid level interface image to obtain a grayscale histogram in the horizontal direction. Since the projection value of the liquid level position will be significantly larger than that of other positions, the maximum value of the histogram is detected to obtain the row number where the liquid level position is located as the liquid level height H, and the measurement accuracy is 1 pixel, which is used for subsequent judgment of the accumulated gas volume.

[0043] 4. Through the settings of S5 and the automatic gas sampling system, the present invention introduces control devices such as solenoid valves to transform the gas circuit and oil circuit without changing the internal structure of the original device and without affecting its functions, so as to realize remote monitoring, automatic gas sampling and real-time measurement.

[0044] The gas collection box is connected to the oil conservator and the gas relay through a three-way valve. In the daily working state, the gas collection box is filled with transformer oil, solenoid valve 1 is open, and the other three solenoid valves are closed. The function of connecting the oil conservator is to exclude the gas in the gas collection box by injecting oil when measuring the gas composition to prevent the interference of the stored gas.

[0045] When there is accumulated gas in the gas relay and the detection result shows that the accumulated gas volume reaches the exhaust threshold, the control unit will trigger the gas sampling and measurement operation. Specifically, solenoid valve 1 is closed, and solenoid valves 2 and 3 are opened for the first preset duration. While draining the oil, the accumulated gas in the gas relay flows into the gas collection box under its own pressure.

[0046] After the first preset time period, the oil level in the gas collecting box will drop due to gas injection, and the gas relay will be refilled with oil. At this time, the solenoid valve 3 is closed to stop oil drainage, the solenoid valve 4 is opened to allow the accumulated gas to flow to the gas sensor, and at the same time, the data acquisition and analysis process of the gas sensor is started to obtain gas composition data for analyzing the operating state of the transformer.

[0047] After the second preset time period, when the gas collecting box is filled with oil and oil comes out of the exhaust port, the solenoid valve 2 and the solenoid valve 4 are closed, the solenoid valve 1 is opened, and the initial state is restored to complete this measurement, gas sampling, and gas analysis work. Description of the Drawings

[0048] In order to make the objectives, technical solutions, and advantages of the invention clearer, the present invention will be further described in detail below with reference to the drawings, where:

[0049] Figure 1 is the flowchart in Embodiment 1;

[0050] Figure 2 is the diagram for explaining the liquid level height measurement in Embodiment 1;

[0051] Figure 3 is the grayscale histogram in the horizontal direction in Embodiment 1;

[0052] Figure 4 is the schematic diagram of the automatic gas sampling system in Embodiment 2. Detailed Embodiments

[0053] The following will be further described in detail through specific embodiments:

[0054] Embodiment 1

[0055] As Figure 1 shown, in this embodiment, a method for identifying the volume of accumulated gas in a gas relay based on image analysis is disclosed, including the following steps:

[0056] S1. Obtain the volume data of the gas injected into the gas relay and the liquid level height, and analyze according to the data to obtain the conversion relationship between the liquid level height and the gas volume in the gas relay. Specifically, in implementation, S1 includes:

[0057] S11. Open the exhaust hole of the gas relay without liquid inside;

[0058] S12. Send liquid into the gas relay from the gas release plug n times, and the volume of the liquid sent each time is equal; after each time the liquid is sent, record the total volume V li of the liquid that has been sent, and record the corresponding liquid level height H i. During specific implementation, after unscrewing the air release plug, connect the injection needle of a medical syringe, and inject liquid each time using the syringe. To ensure the effectiveness of the analysis results, transformer oil is used as the liquid.

[0059] S13. Record the total capacity of the gas relay as V, and obtain the remaining gas volume V after each injection of liquid i = V - V li ;

[0060] S14. Based on the recorded total volume V of the liquid fed in li , the corresponding liquid level height H i and the remaining gas volume V i for analysis to obtain the conversion relationship formula between the liquid level height and the gas volume in the gas relay.

[0061] Among them, in S14, the least squares method is used for linear fitting:

[0062]

[0063]

[0064] Obtain the formula for calculating the gas volume from the liquid level height:

[0065] V 气 = k * H + b;

[0066] Among them, H is the liquid level height, and V 气 is the corresponding gas volume.

[0067] When analyzing the conversion relationship formula between the liquid level height and the gas volume in the gas relay, the present invention feeds n times of liquid with equal volumes into the gas relay from the air release plug, and successively records the total volume V of the liquid that has been fed in li , and the corresponding liquid level height H i . Then, combining the default condition that the space inside the gas relay is entirely occupied by liquid and accumulated gas, and the internal space of the gas and electrical appliance is a fixed value, through linear fitting using the least squares method, a calculation formula that can accurately reflect the conversion relationship between the liquid level height and the gas volume in the gas relay can be obtained. Thus, the accuracy of this method is ensured.

[0068] S2. Obtain the background reference image of the observation window of the gas relay; during specific implementation, the background reference image is an image with an empty liquid level in the observation window. As Figure 2 (a) shown.

[0069] S3. Obtain the image to be detected of the observation window of the gas relay, and combine it with the background reference image to perform image processing by means of image difference to identify the liquid level height in the observation window of the current gas relay;

[0070] In specific implementation, S3 includes:

[0071] S31. Obtain the image to be detected of the observation window of the gas relay, as shown in Figure 2 (b).

[0072] S32. Perform image difference processing on the image to be detected and the background reference image to obtain a difference image, as shown in Figure 2 (c);

[0073] S33. Perform binarization and noise reduction processing on the difference image to obtain a noise-reduced image; specifically, after obtaining the binarization threshold of the difference image using the OTSU algorithm, perform binarization processing on the difference image to obtain an image with only black and white colors, and then perform median filtering and morphological transformation on the binary image to obtain a noise-reduced image; among them, the morphological transformation includes using opening operation to eliminate independent noise points at the segmentation edge and connect and thicken the liquid level information.

[0074] Median filtering has the characteristics of suppressing noise and maintaining details, that is, m numbers are successively taken from the input sequence, and the median among them is taken as the filtering output, and the expression formula is as shown in formula (1),

[0075]

[0076] Opening operation is a morphological transformation method, which can be used to eliminate small objects, separate objects at thin places, and can also be used to smooth the boundaries of large objects, as shown in formula (2):

[0077]

[0078] Where is erosion operation, is dilation operation, and the obtained noise-reduced image is as shown in Figure 2 (d).

[0079] S34. Perform hough transformation on the noise-reduced image, detect the circular border of the observation window, and generate a circular mask with a radius smaller than the circular border according to the detection result. Use the mask to perform bitwise AND operation on the noise-reduced image to remove the circular border of the observation window and obtain a binary image of the liquid level interface, as shown in Figure 2 (e);

[0080] S35. Perform horizontal projection on the liquid level interface image to obtain a grayscale histogram in the horizontal direction, as shown in Figure 3; it can be seen that the projection value of the liquid level position will be significantly greater than the values of other positions. Perform maximum value detection on the histogram to obtain the row number where the liquid level position is located as the liquid level height inside the observation window of the current gas relay. Among them, when performing maximum value detection on the histogram, the measurement accuracy is 1 pixel.

[0081] When analyzing the current liquid level height, since the result of the differential operation may have interference noise due to reasons such as reflection, binarization and noise elimination are required for subsequent processing. The OTSU algorithm is used to obtain the binarization threshold of the image, and the differential image is binarized to obtain an image with only black and white colors. Then, median filtering and morphological transformation are performed on the binary image to remove interference factors. However, at this time, there is still interference introduced by the circular observation window border in the image, so further processing is required. The present invention selects to perform a hough transform on the noise-eliminated image to detect the circular border. According to the detection result, a circular mask with a radius smaller than the detected circle is generated, and the noise-eliminated image is subjected to a bitwise AND operation using the mask, so that the observation window border can be removed to obtain a binary image of the liquid level interface. Then, horizontal projection is performed on the liquid level interface image to obtain a gray histogram in the horizontal direction. Since the projection value of the liquid level position will be significantly greater than the values of other positions, the maximum value of the histogram is detected to obtain the row number where the liquid level position is located as the liquid level height H, and the measurement accuracy is 1 pixel, which is used for subsequent determination of the gas accumulation volume.

[0082] S4. Use the conversion relationship between the liquid level and the gas volume to convert the current liquid level height into the gas volume.

[0083] In the present invention, by obtaining the image to be detected of the observation window of the gas relay and combining with the background reference image, the display result of the current observation window can be automatically detected through the inter-frame image difference method, that is, the liquid level height in the observation window of the current gas relay. Since the space inside the gas relay is entirely occupied by liquid and accumulated gas, and the internal space of the gas and the electrical appliance is a fixed value, it is undoubtedly known that there is a relationship conversion formula between the liquid level height and the gas volume in the gas relay. Since this conversion formula has been pre-analyzed, when the liquid level height in the observation window of the current gas relay is detected, the gas volume in the gas relay at this time can be accurately obtained through this conversion formula.

[0084] Compared with contact detection, the present application does not need to contact the liquid inside the gas relay during the detection process, and can avoid the situation that the detection end (such as the sensor probe) is corroded. Compared with conventional non-contact detection, since the present application does not need to penetrate the external shell of the gas relay, there is no serious interference and signal interruption caused by the external shell.

[0085] The present invention can intelligently detect the gas accumulation situation of the gas relay for a long time and accurately. When it is determined that the accumulated gas volume is abnormal, automatic processing or automatic alarm can be further realized, so that the situation of the transformer can be automatically detected without shutting down the machine, the operation state of the transformer can be grasped in real time, the fault analysis and risk warning of the transformer can be realized, and while ensuring the safe operation of the transformer, the personal safety risk of on-site inspection by maintenance personnel is greatly reduced.

[0086] To verify the actual operation effect of the present invention, the OpenMV embedded image processing module was selected as the image processing device. The image acquisition device was installed on the gas relay. After purchasing the corresponding items according to the overall connection structure diagram of the device and assembling them, physical tests were carried out.

[0087] Since the camera of the OpenMV has been fixed on the cover plate box during actual operation, and the buckle of the cover plate box is firmly fixed to the gas relay, the position of the gas relay in the image will not move. The circular detection by the Hough transform can be replaced by the established circular parameters under the experiment, and the mask can be made and saved in advance to save memory and shorten the startup time of the program. The code was optimized, and the image resolution used was 640×480, so that the above process could be realized on the OpenMV.

[0088] After the image test equipment was prepared, a data fitting experiment was carried out for calculating the volume of the accumulated gas. 50 ml of liquid was injected each time using a syringe. After 13 injections, the liquid level was visually observed to reach the 400 mL scale, and the total capacity could be obtained as 400 + 50 * 13 = 1050 mL. The experimental results were recorded as shown in Table 1:

[0089]

[0090]

[0091] Table 1 Result Record

[0092] It was analyzed that the formula for calculating the gas volume from the liquid level height was:

[0093] V 气 = k * H + b = 1.516 * H + 186.187;

[0094] The verification experimental results are shown in Table 2:

[0095]

[0096] Table 2 Test Data Record and Calculation Results

[0097] The results show that the maximum gas measurement error is 2.5%, with relatively high accuracy.

[0098] Combined with Figure 2 the specific example of Figure 2 in the case shown in 气 (b), the liquid level pixels measured by this device are located at the 300th row of the image, that is, the liquid level height H is about 300. At this time, according to V

[0099] = k * H + b = 1.516 * H + 186.187, the volume of the accumulated gas was calculated to be approximately 641 ml.For the test of the gas extraction device, first fill the gas relay with transformer oil, then open the oil drainage device, inject gas through the gas release plug, drain the oil under pressure to lower the oil level, set the threshold of the accumulated gas volume for starting the gas extraction and measurement operations. When the oil level drops to a certain extent and the calculated accumulated gas volume reaches the threshold, the device realizes the automatic gas extraction and real-time measurement process, and the test results meet the design requirements of the present invention.

[0100] Embodiment 2

[0101] Different from Embodiment 1, in this embodiment, after S4, S5 is further included, which is to judge whether the current gas volume in the gas relay reaches the preset exhaust threshold. If it reaches, control the automatic gas extraction system to discharge the gas in the gas relay and analyze the discharged gas.

[0102] During specific implementation, as Figure 4 shown, the automatic gas extraction system includes a three-way valve, a gas collection box, an oil conservator, a gas detection module and a control module; the oil conservator is connected to interface 1 of the three-way valve through solenoid valve 1; the gas relay is connected to interface 2 of the three-way valve through solenoid valve 2; the inlet of the gas collection box is connected to interface 3 of the three-way valve; the exhaust port of the gas collection box is connected to the gas detection module through solenoid valve 4; the oil drain port of the gas collection box is provided with solenoid valve 3; the control module is electrically connected to the gas detection module, solenoid valve 1, solenoid valve 2, solenoid valve 3 and solenoid valve 4 respectively. During specific implementation, the gas detection module can adopt a gas sensor, and the control module can adopt a microprocessor, such as a single-chip microcomputer.

[0103] The control process of the automatic gas extraction system includes: in the initial state, solenoid valve 1 is opened, and solenoid valves 2, 3 and 4 are closed; when the current gas volume in the gas relay reaches the preset exhaust threshold, the control module closes solenoid valve 1, opens solenoid valves 2 and 3 for a first preset duration; after the first preset duration, the oil level in the gas collection box drops due to gas injection, the gas relay is refilled with oil, solenoid valve 3 is closed, solenoid valve 4 and the gas detection module are opened for a second preset duration; after the second preset duration, the gas collection box is filled with oil and oil starts to flow out of the exhaust hole, solenoid valves 2 and 4 are closed, solenoid valve 1 is opened, and the initial state is restored. The specific values of the first preset duration and the second preset time can be specifically set by those skilled in the art according to the specific models of the gas solenoid valves and related equipment, as long as the above effects can be achieved, which will not be elaborated here.

[0104] Through the setting of S5 and the automatic gas extraction system, the present invention introduces control devices such as solenoid valves to transform the gas circuit and oil circuit without changing the internal structure of the original device and without affecting its functions, so as to realize remote monitoring, automatic gas extraction and real-time measurement. The specific process and principle are as follows:

[0105] The gas collecting box is connected to the conservator and the gas relay through a three-way valve. In the daily working state, the gas collecting box is filled with transformer oil, solenoid valve 1 is open, and the other three solenoid valves are closed. The function of connecting the conservator is to exclude the gas in the gas collecting box by injecting oil when measuring the gas composition, so as to prevent the interference of the stored gas.

[0106] When there is gas accumulation in the gas relay and the detection result shows that the volume of the accumulated gas reaches the exhaust threshold, the control unit will trigger the gas sampling and measurement operation. Specifically, solenoid valve 1 is closed, and solenoid valves 2 and 3 are opened for a first preset duration. While draining the oil, the accumulated gas in the gas relay flows into the gas collecting box under its own pressure.

[0107] After the first preset duration, the oil level in the gas collecting box will drop due to the injection of gas, while the gas relay is refilled with oil; at this time, solenoid valve 3 is closed to stop draining the oil, solenoid valve 4 is opened to let the accumulated gas flow to the gas sensor, and at the same time, the data acquisition and analysis process of the gas sensor is started to obtain the gas composition data for analyzing the operation state of the transformer.

[0108] After the second preset duration, when the gas collecting box is filled with oil and oil comes out of the exhaust port, solenoid valves 2 and 4 are closed, solenoid valve 1 is opened, and the initial state is restored, completing the current measurement, gas sampling, and gas analysis work.

[0109] Based on the image liquid level detection, the present invention can automatically collect and detect the gas generated in the gas relay without power-off by designing relevant peripheral intelligent detection devices, so as to monitor the operation state of the transformer and give early warning of faults online. With the rapid development of Internet of Things technology and embedded devices, the above intelligent detection device has high reliability, low production cost, and low development difficulty. On the other hand, the rapid development of the performance of microprocessors enables image processing to be realized in an embedded environment, making the device portable, with low cost, and easy to measure the accumulated gas volume and composition of the gas relay in real time. The present invention establishes a complete technical chain including liquid level detection based on image analysis, identification of the accumulated gas volume in the gas relay, automatic control of the gas sampling device, gas composition measurement, and linkage with the oil chromatograph. The whole process is automatically completed without human intervention, which not only improves the accuracy and speed of detection, but also ensures the safety of relevant staff, and has very important application value.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Those of ordinary skill in the art should understand that any modifications or equivalent replacements to the technical solutions of the present invention without departing from the purpose and scope of the present technical solution shall be covered by the scope of the claims of the present invention.

Claims

1. A method for identifying the gas volume in a gas relay based on image analysis, characterized in that, It includes the following steps: S1. Obtain the volume data and liquid level height of the gas injected into the gas relay, and analyze them to obtain the conversion relationship between the liquid level height and the gas volume in the gas relay; S2. Obtain the background reference image of the observation window of the gas relay; S3. Obtain the image to be detected of the observation window of the gas relay, and combine it with the background reference image to perform image processing by means of image difference to identify the liquid level height in the observation window of the current gas relay; S4. Use the conversion relationship between the liquid level and the gas volume to convert the current liquid level height into the gas volume; Among them, S3 includes: S31. Obtain the image to be detected of the observation window of the gas relay; S32. Perform image difference processing on the image to be detected and the background reference image to obtain a difference image; S33. Perform binarization and noise reduction processing on the difference image to obtain a noise reduction image; S34. Perform Hough transform on the noise reduction image to detect the circular border of the observation window, and generate a circular mask with a radius smaller than the circular border according to the detection result. Use the mask to perform bitwise AND operation on the noise reduction image to remove the circular border of the observation window and obtain a binary image of the liquid level interface; S35. Perform horizontal projection on the liquid level interface image to obtain a grayscale histogram in the horizontal direction; perform maximum value detection on the histogram to obtain the row number where the liquid level is located as the liquid level height in the observation window of the current gas relay.

2. The method for identifying the gas accumulation volume of a gas relay based on image analysis according to claim 1, wherein: S1 includes: S11. Open the exhaust hole of the gas relay without liquid inside; S12. Send liquid into the gas relay from the gas release plug for n times, and the volume of the liquid sent each time is equal; after each time of sending liquid, record the total volume V of the liquid that has been sent li , and record the corresponding liquid level height H i ; S13. Record the total capacity of the gas relay as V, and obtain the remaining gas volume V after each injection of liquid i = V - V li ; S14. According to the total volume V of the fed liquid recorded li , the corresponding liquid level height H i and the remaining gas volume V i for analysis to obtain the conversion relationship between the liquid level height and the gas volume in the gas relay.

3. The method for identifying the gas accumulation volume of a gas relay based on image analysis according to claim 2, wherein: In S14, least squares method is used for linear fitting: Obtain the formula for calculating the gas volume from the liquid level height: V 气 = k * H + b; where H is the liquid level height and V 气 is the corresponding gas volume.

4. The method for identifying the volume of gas accumulated in a gas relay based on image analysis according to claim 1, wherein: In S2, the background reference image is an image with an empty liquid level in the observation window.

5. The method for identifying the gas accumulation volume of a gas relay based on image analysis according to claim 1, wherein: In S33, after obtaining the binarization threshold of the difference image by using the OTSU algorithm, perform binarization processing on the difference image to obtain an image with only black and white colors, and then perform median filtering and morphological transformation on the binary image to obtain a noise reduction image; among them, the morphological transformation includes using opening operation to eliminate independent noise points at the segmentation edge and connect and thicken the liquid level information.

6. The method for identifying the gas accumulation volume of a gas relay based on image analysis according to claim 1, wherein: In S35, when performing maximum value detection on the histogram, the measurement accuracy is 1 pixel.

7. The method for identifying the gas accumulation volume of a gas relay based on image analysis according to claim 1, wherein: After S4, it also includes S5. Judge whether the current gas volume in the gas relay reaches the preset exhaust threshold. If it reaches, control the automatic gas sampling system to exhaust the gas in the gas relay and analyze the exhausted gas.

8. The method for identifying the gas accumulation volume of a gas relay based on image analysis according to claim 7, wherein: In S5, the automatic gas sampling system includes a three-way valve, a gas collection box, an oil conservator, a gas detection module and a control module; the gas conservator is connected to interface 1 of the three-way valve through solenoid valve 1; the gas relay is connected to interface 2 of the three-way valve through solenoid valve 2; the inlet of the gas collection box is connected to interface 3 of the three-way valve; the exhaust port of the gas collection box is connected to the gas detection module through solenoid valve 4; the oil drain port of the gas collection box is provided with solenoid valve 3; the control module is electrically connected to the gas detection module, solenoid valve 1, solenoid valve 2, solenoid valve 3 and solenoid valve 4 respectively; The control process of the automatic gas extraction system includes: in the initial state, solenoid valve 1 is opened, and solenoid valves 2, 3, and 4 are closed; when the current gas volume in the gas relay reaches the preset exhaust threshold, the control module closes solenoid valve 1, opens solenoid valves 2 and 3 for a first preset duration; after the first preset duration, the oil level in the gas collection box drops due to gas injection, the gas relay is refilled with oil, solenoid valve 3 is closed, solenoid valve 4 and the gas detection module are opened for a second preset duration; after the second preset duration, the gas collection box is filled with oil and oil starts to flow out of the exhaust hole, solenoid valves 2 and 4 are closed, solenoid valve 1 is opened, and the initial state is restored.

Images