Method for monitoring solid oxide electrolytic tank in situ based on ultrasonic flaw detection technology

Abstract

The invention belongs to the technical field of solid oxide electrolytic cell monitoring, and provides a method for in-situ monitoring of a solid oxide electrolytic cell based on an ultrasonic flaw detection technology, and the method comprises the following steps: installing an ultrasonic inspection device on the electrolytic cell; after the ultrasonic inspection device is installed, data obtained by the ultrasonic inspection device are analyzed to obtain a data analysis result, and whether the electrolytic tank is abnormal or not is judged according to the data analysis result. The method provided by the invention has the advantages of non-invasiveness, high resolution, real-time performance, multi-level information provision and the like, and is beneficial to improving the reliability, the safety and the maintenance efficiency of equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of solid oxide electrolyzer monitoring, and more particularly, to a method for in-situ monitoring of solid oxide electrolyzers based on ultrasonic flaw detection technology. Background Art

[0002] Ultrasonic flaw detection technology is a method widely used in the field of non-destructive testing for detecting internal defects, cracks, deformations or other problems in materials. In the early 20th century, with the development of acoustics and electronic technologies, ultrasonic flaw detection technology has been continuously improved and applied in various fields, including engineering, medicine, materials science, and aerospace. Ultrasonic waves are high-frequency sound waves with frequencies typically ranging from 20 kHz (kilohertz) to 100 MHz (megahertz). When ultrasonic waves propagate through a material, they are affected by the internal structure and defects of the material, such as reflection, scattering, and absorption. By measuring the propagation time and intensity of ultrasonic waves, the internal properties and problems of the material can be inferred. Ultrasonic flaw detection equipment generally includes ultrasonic sensors (also known as probes or transducers), pulse generators, receivers, and data processing units. The sensors are used to emit ultrasonic waves and receive echoes, while the pulse generator provides the excitation signal. The receiver captures and records the echo signals, and the data processing unit is used to analyze and interpret the data.

[0003] Proton ceramic electrolyzers use proton-conducting ceramics as electrolytes and decompose water into hydrogen and oxygen through an electrochemical reaction. Different from traditional water electrolysis technology, proton ceramic electrolyzers operate at high temperatures, usually between 500°C and 800°C. The high temperature is one of the key characteristics of proton ceramic electrolyzers. The relatively high operating temperature increases the rate of the electrolysis reaction, thereby increasing the production efficiency of hydrogen and oxygen. However, high-temperature operation also brings challenges such as thermal management problems and energy consumption. In addition, corrosion and degradation of materials at high temperatures can also reduce the lifespan of the battery and increase the maintenance cost. Therefore, ultrasonic flaw detection technology can be used to monitor problems inside the electrolyzer, such as cracks or other defects in the proton electrolyte layer. It provides a non-invasive method that can help improve the reliability and performance of the battery, and diagnose and solve potential problems at an early stage, thus promoting the development of sustainable energy technologies.

[0004] Although ultrasonic flaw detection technology has a wide range of applications in monitoring various materials and devices, there are still some technical drawbacks and challenges when monitoring proton ceramic electrolyzers, including:

[0005] High-temperature operating environment: Proton ceramic electrolyzers usually operate in a high-temperature environment, with temperatures typically between 500°C and 800°C. This high-temperature environment may have an adverse impact on ultrasonic sensors and equipment, including thermal expansion of the sensors, deformation of ceramic materials, and changes in the speed of sound. This may require special sensor designs and thermal management measures.

[0006] Sensor Adaptability: The ultrasonic sensor needs to adapt to high-temperature environments and possess sufficient stability and heat resistance. The material selection and design of the sensor need to consider the impact of high-temperature environments on the sensor's performance.

[0007] Material Characteristics: The ceramic material in the proton ceramic electrolytic cell may interfere with the propagation of ultrasonic waves, such as reflection, scattering, or absorption. This may lead to distortion or attenuation of the ultrasonic signal, making detection and analysis more difficult.

[0008] Material Complexity: The internal structure of the proton ceramic electrolytic cell is complex, including electrolytes, electrodes, and other components. The different acoustic characteristics and geometric shapes of these components may require more complex ultrasonic imaging and analysis methods.

[0009] Monitoring Depth: The depth of ultrasonic wave propagation in materials is limited, so it may not be able to fully detect problems inside the electrolytic cell, especially in the case of complex structures or thicker components.

[0010] Real-time Monitoring Challenges: Proton ceramic electrolytic cells usually need to operate at high temperatures continuously, so real-time monitoring may be more challenging. The sensor needs to operate stably at high temperatures for a long time for continuous monitoring.

[0011] Despite these challenges, ultrasonic flaw detection technology remains a promising method for monitoring the internal conditions of proton ceramic electrolytic cells. With the continuous progress of material and sensor technologies, as well as more research on ultrasonic flaw detection technology in high-temperature environments, it is expected to overcome these drawbacks and improve the accuracy and feasibility of monitoring. Summary of the Invention

[0012] The purpose of the present invention is to provide a method for in-situ monitoring of solid oxide electrolytic cells based on ultrasonic flaw detection technology to improve the above problems.

[0013] To achieve the above purpose, the embodiments of the present application provide the following technical solutions:

[0014] On the one hand, the embodiments of the present application provide a method for in-situ monitoring of solid oxide electrolytic cells based on ultrasonic flaw detection technology, the method comprising:

[0015] Install an ultrasonic inspection device on the electrolytic cell;

[0016] After the ultrasonic inspection device is installed, analyze the data obtained according to the ultrasonic inspection device to obtain a data analysis result, and judge whether the electrolytic cell is abnormal according to the data analysis result.

[0017] Optionally, the electrolytic cell sequentially includes an anode material, an electrolyte, and a cathode material from top to bottom. The ultrasonic inspection device has heat resistance and thermal stability. The ultrasonic inspection device includes an ultrasonic generator, an ultrasonic transmitting probe, an ultrasonic receiving probe, and an ultrasonic receiver. The ultrasonic generator is connected to the ultrasonic transmitting probe. The ultrasonic transmitting probe is installed on the surface of the cathode material. The ultrasonic receiver is connected to the ultrasonic receiving probe. The ultrasonic receiving probe is installed on the surface of the anode material.

[0018] Optionally, the ultrasonic inspection device further includes an oscilloscope, and the ultrasonic receiver is connected to the oscilloscope.

[0019] Optionally, analyze the data obtained according to the ultrasonic inspection device to obtain a data analysis result, and judge whether the electrolytic cell is abnormal according to the data analysis result, including:

[0020] Analyze the data obtained according to the ultrasonic inspection device. The data is an image of the electrolytic cell. The image of the electrolytic cell includes a sound velocity profile diagram, a reflectivity diagram, or an acoustic image.

[0021] Perform image enhancement processing on the image of the electrolytic cell to obtain an enhanced image, and judge whether the electrolytic cell is abnormal according to the enhanced image.

[0022] Optionally, performing image enhancement processing on the image of the electrolytic cell to obtain an enhanced image includes:

[0023] Perform image enhancement processing on the image of the electrolytic cell using an image enhancement algorithm to obtain an enhanced image.

[0024] Optionally, performing image enhancement processing on the image of the electrolytic cell using an image enhancement algorithm to obtain an enhanced image includes:

[0025] Decompose the image of the electrolytic cell using the Gaussian-Laplacian pyramid decomposition method to obtain a Gaussian pyramid and a Laplacian pyramid. The Gaussian pyramid has multiple layers of Gaussian sub-images, and the Laplacian pyramid has multiple layers of Laplacian sub-images. The number of layers of the Gaussian pyramid and the Laplacian pyramid is the same. Calculate the neighborhood standard deviation corresponding to each layer of the Gaussian sub-image, and combine all the neighborhood standard deviations. After combination, perform normalization processing to obtain a normalized pyramid. And obtain an enhanced image according to the normalized pyramid.

[0026] Optionally, obtaining an enhanced image according to the normalized pyramid includes:

[0027] Multiply each layer of the normalized pyramid with the corresponding layer of the Laplacian pyramid to obtain a multiplied pyramid, and then use image reconstruction technology to reconstruct the multiplied pyramid to obtain an enhanced image.

[0028] Optionally, judging whether the electrolytic cell is abnormal according to the enhanced image includes:

[0029] Sending the enhanced image to a staff member, and the staff member judges whether the electrolytic cell is abnormal; or

[0030] Sending the enhanced image to a defect recognition model, and using the defect recognition model to judge whether the electrolytic cell is abnormal.

[0031] Optionally, the method for constructing a defect recognition model includes

[0032] Obtaining historical images of the electrolytic cell, and performing defect annotation on the historical images to obtain annotated images;

[0033] Using the annotated images to train a deep learning model to obtain the defect recognition model.

[0034] Optionally, after performing defect annotation on the historical images to obtain annotated images, it further includes:

[0035] Obtaining a target style image set, where the target style image set includes at least three target style images;

[0036] Inputting the target style image set and each of the annotated images into an image style transfer model to obtain multiple newly generated first images, where the annotation information between the first images and the annotated images is the same, combining the first images and the annotated images to generate a training set, and using the training set to train a deep learning model to obtain the defect recognition model.

[0037] Optionally, the image style transfer model is an AdaIN model.

[0038] Optionally, the deep learning model includes any one of a recurrent neural network model, a convolutional neural network model, and a recursive neural network model.

[0039] Optionally, using the annotated images to train a deep learning model to obtain the defect recognition model includes:

[0040] The deep learning model is trained using the MobileNet optimization technique and the labeled images. During training, for each of the labeled images, when the first preset convolutional kernel of the deep learning model performs a convolution operation on the input feature parameters, a width factor is introduced for optimization. For each channel of the output feature parameters, a second preset convolutional kernel is used for convolution operation, and a split factor is used for optimization. The results of the two optimizations are concatenated and used as the output of the deep learning model during training.

[0041] Optionally, the first preset convolutional kernel is a 1×1 convolutional kernel, and the second preset convolutional kernel is a 3×3 convolutional kernel.

[0042] Optionally, after the ultrasonic inspection device is installed on the electrolytic cell, it further includes calibrating the ultrasonic inspection device.

[0043] Optionally, after determining whether the electrolytic cell is abnormal according to the data analysis result, it further includes:

[0044] Continuously analyze the data obtained by the ultrasonic inspection device at different times to obtain the electrolytic cell abnormality results corresponding to each time;

[0045] Establish an alarm mechanism, and the early warning mechanism alarms according to the electrolytic cell abnormality results corresponding to each time.

[0046] Optionally, establishing an alarm mechanism, and the early warning mechanism alarms according to the electrolytic cell abnormality results corresponding to each time, includes:

[0047] Count the number of times the electrolytic cell is abnormal within a preset time length; compare and analyze the number with a preset number threshold. Among them, if the number exceeds the preset number threshold, an alarm is issued, otherwise no alarm is issued.

[0048] Optionally, after determining whether the electrolytic cell is abnormal according to the data analysis result, it further includes:

[0049] Analyze the reasons for each abnormality to form multiple text messages;

[0050] Extract the features of each text message to obtain feature information, and use a clustering algorithm to cluster all the feature information to obtain an abnormality statistical result, which is used to help the staff make an electrolytic cell maintenance strategy.

[0051] Optionally, extracting the features of each text message to obtain feature information includes:

[0052] Input each text information into a recurrent neural network model to obtain the hidden layer output data of the recurrent neural network model, and use the hidden layer output data as the feature information.

[0053] Optionally, use a clustering algorithm to cluster all the feature information to obtain an anomaly statistical result, including:

[0054] Use the DBSCAN algorithm, K-means clustering algorithm or hierarchical clustering algorithm to cluster all the feature information to obtain an anomaly statistical result.

[0055] Optionally, before analyzing the data obtained according to the ultrasonic inspection device, it further includes:

[0056] Encrypt the data obtained according to the ultrasonic inspection device, transmit it after encryption, and decrypt it after the transmission ends and then perform data analysis.

[0057] The beneficial effects of the present invention are:

[0058] Non-invasive: Ultrasonic flaw detection is a non-invasive technique that does not require introducing external sensors or probes inside the electrolytic cell. This means that monitoring can be carried out without interrupting the operation of the electrolytic cell, reducing downtime and production costs.

[0059] High resolution: Ultrasonic waves can provide high-resolution images and data, and can detect small defects, cracks or problems. This helps to detect and solve potential internal problems of the electrolytic cell at an early stage, avoiding more serious damage or failures.

[0060] Real-time monitoring: Ultrasonic technology can achieve real-time monitoring to continuously track the state changes of the electrolytic cell. This allows operators to take immediate action when problems occur to reduce the risk of production interruption and equipment damage.

[0061] Multi-level information: Ultrasonic technology provides multi-level information about the internal structure, density, sound velocity and acoustic wave attenuation of materials. This can help analyze the physical and chemical changes inside the electrolytic cell and contribute to a deeper understanding of the health status of the electrolytic cell.

[0062] Reduce maintenance costs: By regularly using ultrasonic flaw detection technology to monitor the electrolytic cell, problems can be detected in advance and preventive maintenance measures can be taken, thereby reducing the costs of maintenance and repair.

[0063] Visualization and reporting: Ultrasonic technology generates visual images and reports, enabling operators to clearly see the problems inside the electrolytic cell. This helps to better manage and maintain the equipment.

[0064] In summary, the application of ultrasonic flaw detection monitoring in proton ceramic electrolytic cells and other industrial equipment has many advantages, including non-invasiveness, high resolution, real-time performance, and the provision of multi-level information, which helps to improve the reliability, safety, and maintenance efficiency of equipment.

[0065] Other features and advantages of the present invention will be described in the subsequent specification, and in part will be obvious from the specification, or can be understood by implementing the embodiments of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the written specification, claims, and drawings. Brief Description of the Drawings

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0067] Figure 1 It is a schematic flow diagram of the method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology described in the embodiments of the present invention. Detailed Embodiments

[0068] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. Usually, the components of the embodiments of the present invention described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but only represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0069] It should be noted that similar reference numerals or letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present invention, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0070] Embodiment 1

[0071] As Figure 1As shown, this embodiment provides a method for in-situ monitoring of a solid oxide electrolyzer based on ultrasonic flaw detection technology, and this method includes step S1 and step S2.

[0072] Step S1: Install an ultrasonic inspection device on the electrolyzer.

[0073] A solid oxide electrolyzer can efficiently convert electrical energy into hydrogen, providing a feasible approach for clean and green hydrogen energy production. Hydrogen, as an efficient energy medium, can be used for fuel cell power generation, fuel storage, and industrial applications, helping to reduce carbon emissions and the use of fossil fuels. However, a solid oxide electrolyzer operates in a high-temperature and extreme environment, and its performance may be affected by various factors, such as material instability, gas leakage, and changes in electrode catalytic activity. The materials used in the electrolyzer need to maintain stability in an environment of high temperature, oxygen, and current. In-situ monitoring can not only evaluate the performance of the electrolyzer in real time but also be used to study the properties such as material corrosion and oxygen ion migration, and evaluate the durability and lifespan of the materials, discover potential problems, and take timely measures. Ultrasonic waves are mechanical waves that can propagate in substances and interact with different internal structures. When ultrasonic waves encounter defects, cracks, foreign objects, etc. inside the material, reflection, refraction, and scattering will occur, thus generating echo signals. By analyzing the characteristics of the echo signals, problems inside the material can be determined. Therefore, ultrasonic waves can be used to detect problems such as internal defects, foreign objects, and cracks in the material, that is, ultrasonic flaw detection. The present invention uses ultrasonic waves to monitor the performance of the electrolyzer in real time, such as the integrity of the electrolyte, the state of the electrodes, and the interface changes inside the electrolyzer.

[0074] In this step, the electrolyzer includes an anode material, an electrolyte, and a cathode material in sequence from top to bottom. The ultrasonic inspection device has heat resistance and thermal stability. The ultrasonic inspection device includes an ultrasonic generator, an ultrasonic transmitting probe, an ultrasonic receiving probe, and an ultrasonic receiver. The ultrasonic generator is connected to the ultrasonic transmitting probe. The ultrasonic transmitting probe is installed on the surface of the cathode material. The ultrasonic receiver is connected to the ultrasonic receiving probe. The ultrasonic receiving probe is installed on the surface of the anode material. In addition, the ultrasonic inspection device further includes an oscilloscope, and the ultrasonic receiver is connected to the oscilloscope.

[0075] Meanwhile, in this step, the ultrasonic inspection device can be understood as an ultrasonic sensor. When selecting an ultrasonic sensor, an ultrasonic sensor or probe suitable for high-temperature environments can be chosen. These sensors need to have good thermal stability and heat resistance to work under the high-temperature operating conditions of the proton ceramic electrolytic cell. Also, the frequency range of the sensor needs to be considered to adapt to the required detection depth and resolution. Place the ultrasonic transmitting probe on the surface of the cathode material to introduce ultrasonic signals into the materials of the proton ceramic electrolytic cell. Place the ultrasonic receiving probe on the surface of the anode material and record the ultrasonic signals received by the sensor, including the echo time and intensity.

[0076] After installation, it also includes calibrating the ultrasonic inspection device. Measure the acoustic parameters of the ceramic materials involved in the proton ceramic electrolytic cell, including sound velocity, density, acoustic impedance, etc. These parameters are crucial for the propagation and analysis of ultrasonic waves. Calibrate the sensor to ensure its accuracy and reliability.

[0077] Step S2: After the ultrasonic inspection device is installed, analyze the data obtained according to the ultrasonic inspection device to obtain a data analysis result, and judge whether the electrolytic cell is abnormal according to the data analysis result.

[0078] In this step, the data generated when monitoring PCECs may contain confidential information, so measures need to be taken to ensure the security and confidentiality of the data. This may include security measures for encrypting data transmission, storage, and access. That is to say, after obtaining the data, it also includes: encrypting the data obtained according to the ultrasonic inspection device, transmitting the encrypted data, decrypting it after the transmission ends, and then performing data analysis. Meanwhile, in this step, the specific data analysis steps include Step S21 and Step S22;

[0079] Step S21: Analyze the data obtained according to the ultrasonic inspection device. The data is an image of the electrolytic cell, and the image of the electrolytic cell includes a sound velocity profile diagram, a reflectivity diagram, or an acoustic image;

[0080] Use data analysis software to process the ultrasonic signals recorded by the sensor. This can include converting the data into a sound velocity profile, a reflectivity diagram, or an acoustic image. Among them, analyze the acoustic image to detect problems inside the material, such as cracks, bubbles, or other defects.

[0081] Step S22: Perform image enhancement processing on the image of the electrolytic cell to obtain an enhanced image, and judge whether the electrolytic cell is abnormal according to the enhanced image.

[0082] In this step, to improve the accuracy of detection, the image is subjected to image enhancement processing in this step, and the quality of the image can be improved through this method; in this step, the specific implementation steps of performing image enhancement processing on the image of the electrolytic cell to obtain the enhanced image include step S221:

[0083] Step S221: Use an image enhancement algorithm to perform image enhancement processing on the image of the electrolytic cell to obtain the enhanced image.

[0084] In this step, there are many image enhancement algorithms, and the staff can select the enhancement according to needs. The specific implementation steps of this step include step S2211;

[0085] Step S2211: Decompose the image of the electrolytic cell using the Gaussian-Laplacian pyramid decomposition method to obtain a Gaussian pyramid and a Laplacian pyramid. The Gaussian pyramid has multiple layers of Gaussian sub-images, and the Laplacian pyramid has multiple layers of Laplacian sub-images. The number of layers of the Gaussian pyramid and the Laplacian pyramid is the same; calculate the neighborhood standard deviation corresponding to each layer of the Gaussian sub-image, and combine all the neighborhood standard deviations. After combination, perform normalization processing to obtain the normalized pyramid; and obtain the enhanced image according to the normalized pyramid.

[0086] In this step, after decomposing the image of the electrolytic cell using the Gaussian-Laplacian pyramid decomposition method, a Gaussian pyramid and a Laplacian pyramid are obtained. Among them, both the Gaussian pyramid and the Laplacian pyramid have three layers; and the number of layers of the normalized pyramid is also three. At the same time, in this step, the specific implementation steps of obtaining the enhanced image according to the normalized pyramid include step S22111;

[0087] Step S22111: Multiply each layer of the normalized pyramid with each layer of the Laplacian pyramid to obtain the multiplied pyramid. After multiplication, use image reconstruction technology to reconstruct the multiplied pyramid to obtain the enhanced image.

[0088] This step can be understood as: multiplying the first layer of the normalized pyramid with the first layer of the Laplacian pyramid, multiplying the second layer of the normalized pyramid with the second layer of the Laplacian pyramid, and multiplying the third layer of the normalized pyramid with the third layer of the Laplacian pyramid;

[0089] In step S22, the specific implementation steps of determining whether the electrolytic cell is abnormal according to the enhanced image include:

[0090] Step S23: Send the enhanced image to the staff, and the staff determines whether the electrolytic cell is abnormal; or

[0091] Step S24: Send the enhanced image to the defect recognition model, and use the defect recognition model to determine whether the electrolytic cell is abnormal.

[0092] In this step, the enhanced image can be sent to the staff, for example, to technicians. The technicians can determine whether the electrolytic cell is abnormal based on their experience. In addition to this manual method, this step also provides a model recognition method, which can improve the efficiency and accuracy of recognition. In this step, the construction method of the defect recognition model includes Step S241 and Step S242;

[0093] Step S241: Obtain the historical images of the electrolytic cell, perform defect annotation on the historical images to obtain the annotated images;

[0094] Step S242: Use the annotated images to train the deep learning model to obtain the defect recognition model.

[0095] In this step, the deep learning model includes any one of a recurrent neural network model, a convolutional neural network model, and a recursive neural network model. At the same time, the specific implementation steps of using the annotated images to train the deep learning model to obtain the defect recognition model include Step S2421;

[0096] Step S2421: Use the MobileNet optimization technique and the annotated images to train the deep learning model. Among them, during training, for each of the annotated images, when performing a convolution operation on the input feature parameters with the first preset convolution kernel in the deep learning model, a width factor is introduced for optimization. For each channel of the output feature parameters, a second preset convolution kernel is used for convolution operation, and a splitting factor is used for optimization. The results of the two optimizations are concatenated, and the concatenated result is used as the output of the deep learning model during training.

[0097] In this step, the first preset convolution kernel is a 1×1 convolution kernel, and the second preset convolution kernel is a 3×3 convolution kernel. At the same time, the width factor and the splitting factor can be custom-set according to the needs of the staff. For example, the width factor can be 0.5, and the splitting factor can be 0.8;

[0098] In addition to the above method for constructing a defect recognition model, the following steps can also be adopted for construction, specifically including step S243 and step S244. In the following steps, the labeled images are expanded to enrich the training samples, and the accuracy of model training can be improved through this method;

[0099] Step S243: Obtain a target style image set, where the target style image set includes at least three target style images;

[0100] In this step, the target style image set can be some images obtained by the staff, and the style of the images is not restricted. To ensure the richness of the samples, at least three target style images are included in this step;

[0101] Step S244: Input the target style image set and each of the labeled images into an image style transfer model to obtain multiple newly generated first images. The annotation information between the first images and the labeled images is the same. Combine the first images and the labeled images to generate a training set, and use the training set to train a deep learning model to obtain the defect recognition model.

[0102] In this step, the image style transfer model is the AdaIN model. Suppose there are three target style images. After inputting the target style image set and each of the labeled images into the image style transfer model, each labeled image will correspondingly generate three new images. The annotation information of these three new images is consistent with the annotation information of the corresponding labeled image. Through this method, the richness of the samples can be improved;

[0103] In addition to the above steps, after step S2, the state of the proton ceramic electrolytic cell can also be monitored in real time, and a real-time data acquisition and analysis system can be established to continuously monitor the change of ultrasonic signals and detect problems in a timely manner. The real-time monitoring can be integrated with an alarm system to issue an alarm when a problem is found. The specific implementation steps include step S3 and step S4;

[0104] Step S3: Continuously analyze the data obtained by the ultrasonic inspection device at different times to obtain the electrolytic cell abnormality results corresponding to each time;

[0105] Step S4: Establish an alarm mechanism, and the early warning mechanism alarms according to the electrolytic cell abnormality results corresponding to each time. The specific implementation steps of this step include step S41;

[0106] Step S41: Count the number of times the electrolytic cell has an abnormality within a preset time length; compare and analyze the number with a preset number threshold. Among them, if the number exceeds the preset number threshold, an alarm will be issued, otherwise no alarm will be issued.

[0107] In this step, the preset number threshold can be customized according to the needs of the staff, and the preset time length can also be customized according to the needs of the staff. For example, in the past day, past half day, past hour, etc.;

[0108] In addition to the above steps, it is also possible to analyze the reasons for the statistical anomalies to facilitate the staff to make an electrolytic cell maintenance strategy. That is, after step S2, it further includes:

[0109] Step S5: Analyze the reasons for each anomaly to form multiple text messages;

[0110] Step S6: Extract the features of each text message to obtain feature information, and use a clustering algorithm to cluster all the feature information to obtain an anomaly statistical result, where the anomaly statistical result is used to help the staff make an electrolytic cell maintenance strategy.

[0111] In this step, use the DBSCAN algorithm, K-means clustering algorithm or hierarchical clustering algorithm to cluster all the feature information to obtain an anomaly statistical result. At the same time, the specific implementation steps of extracting the features of each text message to obtain feature information include step S61;

[0112] Step S61: Input each text message into a recurrent neural network model, obtain the hidden layer output data of the recurrent neural network model, and use the hidden layer output data as the feature information.

[0113] Through the method in this embodiment, this embodiment has the following advantages:

[0114] Non-invasive: Ultrasonic flaw detection is a non-invasive technique that does not require the introduction of external sensors or probes inside the electrolytic cell. This means that monitoring can be carried out without interrupting the operation of the electrolytic cell, reducing downtime and production costs.

[0115] High resolution: Ultrasonic waves can provide high-resolution images and data, and can detect small defects, cracks or problems. This helps to detect and solve potential internal problems of the electrolytic cell at an early stage, avoiding more serious damage or failures.

[0116] Real-time monitoring: Ultrasonic technology can achieve real-time monitoring to continuously track the state changes of the electrolytic cell. This allows operators to take immediate action when problems occur to reduce the risk of production interruption and equipment damage.

[0117] Multi-level information: Ultrasonic technology provides multi-level information about the internal structure, density, sound velocity and acoustic wave attenuation of materials. This can help analyze the physical and chemical changes inside the electrolytic cell and contribute to a deeper understanding of the health status of the electrolytic cell.

[0118] Reducing maintenance costs: By regularly using ultrasonic flaw detection technology to monitor the electrolytic cell, problems can be detected in advance and preventive maintenance measures can be taken, thus reducing the costs of maintenance and repair.

[0119] Visualization and reporting: Ultrasonic technology generates visual images and reports, enabling operators to clearly see the problems inside the electrolytic cell. This helps in better managing and maintaining the equipment.

[0120] In summary, the application of ultrasonic flaw detection monitoring in proton ceramic electrolytic cells and other industrial equipment has many advantages, including non-invasiveness, high resolution, real-time nature, and provision of multi-level information, which helps improve the reliability, safety, and maintenance efficiency of the equipment.

[0121] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. Method for in-situ monitoring of solid oxide electrolytic cell based on ultrasonic flaw detection technology, characterized in that, Including: Install an ultrasonic inspection device on the electrolytic cell; After the ultrasonic inspection device is installed, analyze the data obtained according to the ultrasonic inspection device to obtain a data analysis result, and judge whether the electrolytic cell is abnormal according to the data analysis result.

2. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, wherein The electrolytic cell sequentially includes an anode material, an electrolyte, and a cathode material from top to bottom. The ultrasonic inspection device has heat resistance and thermal stability. The ultrasonic inspection device includes an ultrasonic generator, an ultrasonic transmitting probe, an ultrasonic receiving probe, and an ultrasonic receiver. The ultrasonic generator is connected to the ultrasonic transmitting probe. The ultrasonic transmitting probe is installed on the surface of the cathode material. The ultrasonic receiver is connected to the ultrasonic receiving probe. The ultrasonic receiving probe is installed on the surface of the anode material.

3. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 2, wherein The ultrasonic inspection device further includes an oscilloscope, and the ultrasonic receiver is connected to the oscilloscope.

4. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, wherein Analyze the data obtained according to the ultrasonic inspection device to obtain a data analysis result, and judge whether the electrolytic cell is abnormal according to the data analysis result, including: Analyze the data obtained according to the ultrasonic inspection device. The data is an image of the electrolytic cell. The image of the electrolytic cell includes a sound velocity profile, a reflectivity map, or an acoustic image; Perform image enhancement processing on the image of the electrolytic cell to obtain an enhanced image, and judge whether the electrolytic cell is abnormal according to the enhanced image.

5. The method for in-situ monitoring of a solid oxide electrolysis cell based on ultrasonic flaw detection technology according to claim 4, characterized in that, Performing image enhancement processing on the image of the electrolytic cell to obtain an enhanced image, including: Use an image enhancement algorithm to perform image enhancement processing on the image of the electrolytic cell to obtain an enhanced image.

6. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 5, wherein, Using an image enhancement algorithm to perform image enhancement processing on the image of the electrolytic cell to obtain an enhanced image, including: Adopt the Gaussian-Laplacian pyramid decomposition method to decompose the image of the electrolytic cell to obtain a Gaussian pyramid and a Laplacian pyramid. The Gaussian pyramid has multiple layers of Gaussian sub-images. The Laplacian pyramid has multiple layers of Laplacian sub-images. The number of layers of the Gaussian pyramid and the Laplacian pyramid is the same; calculate the neighborhood standard deviation corresponding to each layer of the Gaussian sub-image, and combine all the neighborhood standard deviations, and perform normalization processing after combination to obtain a normalized pyramid; and obtain an enhanced image according to the normalized pyramid.

7. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 6, wherein Obtaining an enhanced image according to the normalized pyramid, including: Multiply each layer of the normalized pyramid by each layer of the Laplacian pyramid to obtain a multiplied pyramid, and use image reconstruction technology to reconstruct the multiplied pyramid after multiplication to obtain an enhanced image.

8. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 4, wherein Judging whether the electrolytic cell is abnormal according to the enhanced image, including: Send the enhanced image to the staff, and the staff judge whether the electrolytic cell is abnormal; or Send the enhanced image to a defect recognition model, and use the defect recognition model to judge whether the electrolytic cell is abnormal.

9. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 8, wherein, The construction method of the defect recognition model includes Obtain the historical image of the electrolytic cell, perform defect annotation on the historical image to obtain the annotated image; Use the annotated image to train a deep learning model to obtain the defect recognition model.

10. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 9, characterized in that, After performing defect annotation on the historical image to obtain the annotated image, it further includes: Obtain a target style image set, where the target style image set includes at least three target style images; Input the target style image set and each annotated image into an image style transfer model to obtain multiple newly generated first images. The annotation information between the first images and the annotated images is the same. Combine the first images and the annotated images to generate a training set, and use the training set to train a deep learning model to obtain the defect recognition model.

11. The method for in-situ monitoring of a solid oxide electrolyzer based on ultrasonic flaw detection technology according to claim 10, wherein, The image style transfer model is an AdaIN model.

12. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 9, wherein The deep learning model includes any one of a recurrent neural network model, a convolutional neural network model, and a recursive neural network model.

13. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 9, wherein Using the annotated image to train a deep learning model to obtain the defect recognition model includes: Use the MobileNet optimization technique and the annotated image to train a deep learning model. During training, for each annotated image, when a first preset convolutional kernel in the deep learning model performs a convolution operation on the input feature parameters, a width factor is introduced for optimization. For each channel of the output feature parameters, a second preset convolutional kernel is used for convolution operation, and a splitting factor is used for optimization. The results of the two optimizations are spliced together, and the spliced result is used as the output of the deep learning model during training.

14. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 13, wherein, The first preset convolutional kernel is a 1×1 convolutional kernel, and the second preset convolutional kernel is a 3×3 convolutional kernel.

15. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, characterized in that, After installing the ultrasonic inspection device on the electrolytic cell, it further includes calibrating the ultrasonic inspection device.

16. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, wherein, After judging whether the electrolytic cell is abnormal according to the data analysis result, it further includes: Continuously analyze the data obtained by the ultrasonic inspection device at different times to obtain the electrolytic cell abnormality result corresponding to each time; Establish an alarm mechanism, and the early warning mechanism alarms according to the electrolytic cell abnormality result corresponding to each time.

17. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, characterized in that, Establish an alarm mechanism, and the early warning mechanism alarms according to the electrolytic cell abnormality result corresponding to each time, including: Count the number of times the electrolytic cell is abnormal within a preset time length; compare and analyze the number with a preset number threshold. If the number exceeds the preset number threshold, an alarm is issued, otherwise no alarm is issued.

18. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, characterized in that After judging whether the electrolytic cell is abnormal according to the data analysis result, it further includes: Analyze the cause of each abnormality to form multiple text messages; Extract the features of each text message to obtain feature information, and use a clustering algorithm to cluster all the feature information to obtain an abnormality statistics result, which is used to help the staff make an electrolytic cell maintenance strategy.

19. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 18, characterized in that, Extracting the features of each text message to obtain feature information includes: Input each text information into a recurrent neural network model to obtain the hidden layer output data of the recurrent neural network model, and use the hidden layer output data as the feature information.

20. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 18, wherein Use a clustering algorithm to cluster all the feature information to obtain an anomaly statistical result, including: Use the DBSCAN algorithm, K-means clustering algorithm or hierarchical clustering algorithm to cluster all the feature information to obtain an anomaly statistical result.

21. The method for in-situ monitoring of a solid oxide electrolytic cell based on ultrasonic flaw detection technology according to claim 1, wherein Before analyzing the data obtained according to the ultrasonic inspection device, it further includes: Perform encryption processing on the data obtained according to the ultrasonic inspection device, transmit the data after encryption processing, and decrypt the data after the transmission ends before performing data analysis.

Images