An RH circulation immersion tube detection device and method

Abstract

The present invention belongs to the technical field of steelmaking equipment and automatic detection methods, and specifically discloses an RH circulation immersion tube detection device, which includes an electrical control unit. The electrical control unit detects the state of the immersion tube according to data related to the detected condition states detected under different detection conditions of the immersion tube. The electrical control unit applies a learning algorithm to a learning model containing parameters to obtain the threshold value, and judges whether the immersion tube needs to be repaired or replaced by comparing with the threshold value; the device further includes a power supply and a flow control unit respectively connected to the electrical control unit; the inlet end of the flow control unit is connected with an air source treatment device through a pipeline, and the outlet end of the flow control unit is connected with an output control joint through a pipeline. The present invention establishes models for different detection conditions, evaluates the air permeability and corrosion index of the refractory of the immersion tube, promotes the extension of the service life of the immersion tube, can improve the vacuum treatment efficiency of the RH furnace, and reduce the comprehensive cost.

Description

Technical Field

[0001] The present invention belongs to the technical field of steelmaking equipment and automatic detection methods, and particularly relates to a detection device and method for RH circulation immersion tubes. Background Art

[0002] The RH device consists of a vacuum chamber equipped with an immersion riser and an immersion downcomer and an exhaust system. The immersion tube is an important part of the RH refining furnace.

[0003] Argon is introduced into the immersion riser of the immersion tube as a driving gas to promote the circulation of the molten steel in the ladle through the vacuum chamber. The molten steel circulation flow rate is not only related to the vacuum environment of the RH, but the argon flow rate in the immersion riser directly affects the efficiency of vacuum dehydrogenation and deoxidation, powder injection desulfurization, and temperature and composition uniformity during the RH refining process.

[0004] In the RH refining furnace, the working environment of the immersion tube is the most severe. The intermittent operation brings strong thermal shock damage. The high-speed flowing molten steel generates huge mechanical erosion on the refractory material of the inner lining of the immersion tube, causing the mechanical properties and erosion resistance of the refractory material of the RH immersion tube to decline or even be damaged. The thickness of the inner lining material of the immersion tube becomes thinner, which is the most severely eroded part of the RH furnace.

[0005] Although the slag adhering to the surface of the immersion tube can protect the immersion tube from the erosion of the liquid slag, with the repeated use of the immersion tube, the adhesion of the solid molten slag on the surface of the immersion tube will continuously thicken. That is, the blown argon flow rate is affected by the erosion and slag adhesion of the immersion riser, which also affects the rhythm of the RH refining.

[0006] In the process stages such as new lining, baking, turnover preheating, gunning, and slag melting, whether the immersion tube is blocked or eroded often depends on the experience of the operators to observe and judge. Therefore, it often occurs that the immersion tube is taken offline before its service life, or the slag adhesion and blockage of the immersion tube extend the treatment time of the RH refining, and even the immersion tube is eroded excessively, resulting in the molten steel flushing into the upper part of the vacuum chamber, leading to an emergency shutdown accident.

[0007] Therefore, providing a detection device and detection method for RH circulation immersion tubes is a technical problem that those skilled in the art urgently need to solve. Summary of the Invention

[0008] The purpose of the present invention is to overcome the defect that in the prior art, it is inaccurate to judge whether the immersion tube is blocked or eroded relying on the refining furnace times, management systems, and the experience of the operators, and to provide a detection device and method for RH circulation immersion tubes.

[0009] In the first aspect of the present invention, a detection device for RH circulation immersion tubes is provided, including an electrical control unit. The electrical control unit detects the state of the immersion tube according to data related to the detected condition states detected under different detection conditions of the immersion tube. The electrical control unit applies a learning algorithm to a learning model containing parameters to obtain a threshold value, and judges whether the immersion tube needs to be repaired or replaced by comparing with the threshold value;

[0010] The device further includes a power supply and a flow control unit respectively connected to the electrical control unit; the inlet end of the flow control unit is connected with an air source treatment device through a pipeline, and the outlet end of the flow control unit is connected with an output control joint through a pipeline.

[0011] A further solution is that the device further includes a body, an operation panel and a display device are arranged on the surface of the body, and the display device is connected with the electrical control unit;

[0012] An air inlet and an immersion tube connection port are arranged on one side of the body;

[0013] The air inlet is communicated with the air source treatment device; the immersion tube connection port is used for directly or indirectly connecting the output control joint with the immersion tube to be tested.

[0014] The display device is internally provided with an SD card physical medium memory.

[0015] A further solution is that a selection button and an indicator light are respectively connected to the electrical control unit.

[0016] A further solution is that the air source treatment device includes a water filter and a pressure regulator; the water filter and the pressure regulator are sequentially communicated with the air inlet through pipelines; a plurality of flow control units are provided, and each flow control unit is connected with a plurality of output control joints through pipelines.

[0017] A further solution is that a flow sensor and a pressure sensor are arranged in the pipeline at the outlet end of the flow control unit; respectively used for collecting the flow value and the pressure value of the branch of the device connected to the immersion tube.

[0018] A further solution is to collect the pressure and flow data of the pipeline at the detection site, and send them to the electrical control unit to construct a steady-state detection data set, and preprocess the steady-state detection data set, and train it through a long short-term memory neural network to sequentially construct a first learning model, a second learning model and a third learning model;

[0019] The first learning model is used for the detection during the new masonry, casting and welding periods of the immersion tube;

[0020] The second learning model is used for the detection during the drying and baking period of the immersion tube;

[0021] The third learning model is used for the detection during the turnover preheating period of the immersion tube on the production line, that is, during the slag melting and gunning period.

[0022] In the second aspect of the present invention, there is provided a method for detecting an RH circulation immersion tube, characterized in that the above-mentioned RH circulation immersion tube detection device is applied, including the following steps:

[0023] S1. Connect the process gas to the air inlet. The process gas passes through a water filter and a pressure regulator in sequence for impurity filtration and pressure stabilization; connect the immersion tube to be tested to the output control joint through a metal hose;

[0024] S2. Select the model for different detection conditions of the immersion tube to be tested by pressing the model selection button on the operation panel according to the immersion tube to be tested;

[0025] S3. The flow control unit gives the flow rate of the process gas, obtains the results of the flow sensor and the pressure sensor inside the device and sends them to the electrical control unit;

[0026] S4. The electrical control unit compares the obtained results with the thresholds of the corresponding models to judge whether the immersion tube is normal or needs maintenance, and displays the results on the display device through the HMI and stores the results.

[0027] A further solution is that in S2, the selected models include:

[0028] If the immersion tube to be tested is in the new masonry, casting, and welding period, select the first learning model;

[0029] If the immersion tube to be tested is in the drying and baking period, select the second learning model;

[0030] If the immersion tube to be tested is in the turnover preheating period on the production line, select the third learning model.

[0031] A further solution is that in S4, judging whether the immersion tube is normal or needs maintenance includes:

[0032] 54% of the process ventilation volume is set as the lower limit of the flow threshold, and 115% of the process ventilation volume is set as the upper limit of the flow threshold.

[0033] In the scenario of the first learning model, if the result of the flow sensor is greater than the upper limit of the flow threshold of the first learning model, it is judged that the weld leaks; if the result of the flow sensor is greater than the lower limit of the flow threshold of the first learning model and the pipeline back pressure exceeds the upper limit of the pressure threshold, it is judged that the masonry is blocked.

[0034] In the second learning model scenario, if the result of the flow sensor is less than the lower limit of the flow threshold of the second learning model and the pipeline back pressure exceeds the upper limit of the pressure threshold, it is determined that the vent hole of the dip tube is blocked; if the result of the flow sensor is greater than the upper limit of the flow threshold of the first learning model, it is determined that there is a weld leak.

[0035] In the third learning model scenario, if the result of the flow sensor is greater than the upper limit of the flow threshold of the third learning model, it is determined that there is erosion damage; if the result of the flow sensor is less than the lower limit of the flow threshold of the third learning model and the pipeline back pressure exceeds the upper limit of the pressure threshold, it is determined that there is a blockage fault.

[0036] In the above first, second, and third learning model scenarios, if the result of the flow sensor is between the upper limit and the lower limit of the flow threshold of their respective learning models, it is determined that the dip tube inspection is qualified.

[0037] In the above first, second, and third learning model scenarios, the lower limit parameter of the pressure threshold is used for empirical analysis. For too low pipeline pressure, the operator analyzes whether there is a leak in the metal hose connecting the detection device and the dip tube. If there is no leak in the connecting hose, the process personnel need to be informed to recheck whether the model parameters match the technical parameters of the dip tube to be tested on site and the process parameters of the RH refining.

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0039] The present invention digitalizes the ventilation capacity of the dip tube, which can be used to analyze and analogize the quality indicators of the dip tube, establish models for different detection conditions, and combine with the RH processing capacity, having the consistency of indicators, which can improve the reliability of the RH refining process; the assessment of the ventilation performance and erosion index of the refractory of the dip tube promotes the extension of the life of the dip tube, which can improve the vacuum treatment efficiency of the RH furnace, reduce energy consumption, and reduce the comprehensive cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The following drawings only schematically illustrate and explain the present invention and are not used to limit the scope of the present invention, wherein:

[0041] Figure 1 : Schematic structural diagram of the present invention;

[0042] Figure 2 : Schematic diagram of the connection of internal components of the present invention;

[0043] In the figure: 1, body; 2, display device; 3, air inlet; 4, dip tube connection port; 5, model selection button; 6, power supply; 7, electrical control unit; 8, operation panel; 9, flow control unit; 10, flow sensor; 11, pressure sensor; 12, air source treatment device; 13, output control joint. Detailed implementation mode

[0044] In order to make the purpose, technical solution, design method and advantages of the present invention clearer, the present invention will be further described in detail below through specific embodiments with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] As Figure 2 shown, in the first aspect of the present invention, a detection device for RH circulation immersion tubes is provided, including an electrical control unit 7. The electrical control unit includes a PLC, a network, and terminal blocks, etc. The electrical control unit 7 detects the state (normal or abnormal) of the immersion tube according to data related to the detected condition states detected under different detection conditions of the immersion tube. The electrical control unit 7 applies a learning algorithm to a learning model containing parameters to obtain a threshold value, and judges whether the immersion tube needs to be repaired or replaced by comparing with the threshold value;

[0046] The device further includes a power supply 6 and a flow control unit 9 respectively connected to the electrical control unit 7; the flow control unit 9 is a metallurgical characteristic flow controller MFC, and the immersion tube flow measurement and control unit adopts a metallurgical characteristic mass flow controller, which ensures that the detection process conforms to the use conditions of the RH immersion tube and ensures the reliability and reproducibility of the detection data. The inlet end of the flow control unit 9 is connected with an air source treatment device 12 through a pipeline, and the outlet end of the flow control unit 9 is connected with an output control joint 13 through a pipeline.

[0047] As Figure 1 shown, in order to facilitate the handling and movement of the detection personnel, the overall structure of the present invention adopts an integrated and movable design. The device further includes a main body 1, and an operation panel 8 and a display device 2 are arranged on the surface of the main body 1. The display device 2 is connected to the electrical control unit 7; an air inlet 3 and an immersion tube connection port 4 are opened on one side of the main body 1; the air inlet 3 is communicated with the air source treatment device 12; the immersion tube connection port 4 is used to directly or indirectly connect the output control joint 13 with the immersion tube to be tested. The display device 2 is internally provided with an SD card physical medium memory for storing RH station information, test time, detection data and detection conclusions of each immersion tube.

[0048] Technicians use the stored data curve to analyze the detection results, and can also conduct research on the masonry process, baking process, gunning system, and smelting efficiency item by item, and continuously optimize the key links during the operation process.

[0049] A model selection button 5 and an indicator light are arranged on the operation panel 8, and the model selection button 5 and the indicator light are respectively connected to the electrical control unit 7.

[0050] The gas source treatment device 12 includes a water filter and a pressure regulator; the water filter and the pressure regulator are sequentially connected to the air inlet 3 through pipelines; several flow control units 9 are provided, and each flow control unit 9 is connected with several output control joints 13 through pipelines. The detection device is used in an intermittent manner, and the pressure sampling points of the nitrogen or argon process gas source are not fixed. In order to prevent water vapor in the process gas pipeline from being brought into the detection device along with nitrogen and argon, thus affecting the detection results, the gas source treatment unit is specially designed with a water filter screen to avoid such problems.

[0051] A flow sensor 10 and a pressure sensor 11 are arranged in the pipeline at the outlet end of the flow control unit 9; they are respectively used to collect the flow value and the pressure value of the pipeline inside the device.

[0052] The method for model construction is as follows: collect the pressure and flow data of the pipeline at the detection site, and send them to the electrical control unit 7 to construct a steady-state detection data set, and preprocess the steady-state detection data set. After training by a long short-term memory neural network, construct a first learning model, a second learning model, and a third learning model in sequence;

[0053] The first learning model is used for the detection during the new masonry, casting, and welding periods of the dipping tube;

[0054] The second learning model is used for the detection during the drying and baking periods of the dipping tube;

[0055] The third learning model is used for the detection during the online turnover and preheating periods of the dipping tube, that is, the slag melting and gunning periods.

[0056] In the second aspect of the present invention, a method for detecting an RH circulation dipping tube is provided. Using the above-mentioned RH circulation dipping tube detection device, it includes the following steps:

[0057] S1. Connect the process gas to the air inlet 3, and the process gas passes through the water filter and the pressure regulator in sequence for impurity filtration and pressure stabilization; connect the dipping tube to be measured to the output control joint 13 through a metal hose;

[0058] S2. According to the different detection conditions of the dipping tube to be measured, press the model selection button 5 on the operation panel 8 to select the model;

[0059] S3. The flow control unit 9 gives the flow rate of the process gas, obtains the results of the flow sensor 10 and the pressure sensor 11 in the pipeline inside the device, and sends them to the electrical control unit 7;

[0060] S4. The electrical control unit 7 compares the obtained results with the thresholds of the corresponding models to judge whether the dipping tube to be measured needs to be repaired, and displays the results on the display device through the HMI and stores the results.

[0061] Wherein, in said S2, the selection model includes:

[0062] If the immersion pipe to be tested is in the new laying, pouring or welding stage, select the first learning model;

[0063] If the dip tube to be tested is in the drying and baking period, the second learning model is selected;

[0064] If the immersion pipe to be tested is in the preheating period of online turnover, the third learning model is selected to provide detection means for operations such as slagging and gunning.

[0065] In S3, judging whether the dip tube to be tested needs maintenance includes:

[0066] In the first learning model scenario, if the result of the flow sensor 10 is greater than the flow threshold of the first learning model, it is judged as weld leakage; if the result of the flow sensor 10 is less than the flow threshold of the first learning model, it is judged as pouring blockage;

[0067] In the second learning model scenario, if the result of the flow sensor 10 is less than the flow threshold of the second learning model, it is determined that the air vent of the immersion pipe is blocked;

[0068] In the third learning model scenario, if the result of the flow sensor 10 is greater than the flow threshold of the third learning model, it is judged as erosion damage; if the result of the flow sensor 10 is less than the flow threshold of the third learning model, it is judged as a blockage fault.

[0069] In order to facilitate handling and movement during detection, the overall structure of the present invention adopts an integrated movable design. The immersion pipe detection device of the present invention can also be integrated with the existing RH circulation lifting gas control valve station, and the second learning model and the third learning model of the present invention are added. The RH main control room host computer stores the detection model, and the remote control of the RH circulation lifting gas control valve station is used to realize the performance detection of the immersion pipe, and all the detection data are stored in the RH main control room host computer. To achieve this purpose, a solenoid valve can be added to the N immersion pipe control branch to realize the rotation detection of N branches. This method has high detection efficiency and does not require on-site personnel to frequently switch manual valves to select the immersion pipe to be detected. This technical point is also within the protection scope of the present invention.

[0070] The second learning model is mainly applicable to some occasions where both RH riser and settling pipes are purchased as finished products. It provides a means of testing the immersion pipes of the purchased RH riser and gives the test results. If there is an RH immersion pipe station on site and new masonry, pouring and welding are being carried out, the test of the second learning model can be omitted after the test of the first learning model is completed, thereby improving the commissioning efficiency of the RH immersion pipe.

[0071] This application uses quantitative and qualitative analysis to evaluate the ventilation performance of the entire process of using the dipping tube, establishing a first learning model for the new masonry, casting, and welding of the dipping tube, a second learning model for drying and baking, and a third learning model for turnover and preheating. The models are mainly reflected in the process pressure and flow curves. According to the technical requirements of the RH for the dipping tube and in combination with the refining rhythm and the assessment requirements for the service life of the RH dipping tube, the allowable flow range for detection is proposed.

[0072] The performance indicators to be evaluated for the dipping tube are different in the three process stages of new masonry and casting, drying and baking, and turnover and preheating (slag melting, gunning). For process requirements, 3 models are set up, each model corresponding to a set of data, for a total of 3 sets of data. It is suitable for performance detection throughout the entire cycle of using the dipping tube and provides effective index quantification for the process links.

[0073] Taking 200t RH as an example, the total treatment flow rate is 90 NM 3 / h, the total gunning flow rate is 180 NM 3 / h, the minimum process flow rate is NM 3 / h. With 12 dipping tubes arranged, the ventilation capacity of a single dipping tube is 28 NM 3 / h, and the assessment index is 15 NM 3 / h.

[0074] If the ventilation volume is less than 15 NM 3 / h (corresponding to 54% of the process ventilation volume), and the back pressure exceeds the upper limit value, a blockage fault is reported;

[0075] If the ventilation volume is 15 NM 3 / h - 22 NM 3 / h (corresponding to 54% - 80% of the process ventilation volume), and the back pressure is higher than the normal range, a qualified report is given;

[0076] If the ventilation volume is 22 NM 3 / h - 28 NM 3 / h (corresponding to 80% - 100% of the process ventilation volume), and the back pressure is higher than the normal range, a qualified report is given;

[0077] If the ventilation volume is 28 NM 3 / h (corresponding to 100% ventilation volume), a qualified report is given;

[0078] If the ventilation volume > 32 NM 3 / h (corresponding to 115% of the process ventilation volume), and the flow rate exceeds the normal range, an erosion and breakage report is given;

[0079] The operating conditions in the three stages are different, and the data sets are different. The corresponding data are used to compare the detection data respectively, with high accuracy.

[0080] The detection device is arranged for the RH site, and can respectively record the test data of the dip pipes of multi-station RH, and store them respectively in the physical medium memory of the HMI and the host computer in the RH main control room. These test data can be used by process personnel for on-line analysis or off-line analysis.

[0081] In order to better illustrate the use process of the device of the present application, some examples are listed below:

[0082] Example 1

[0083] A set of 120t RH vacuum refining equipment in a steel plant, with 12 dip pipes. Four flow controllers provide the circulating process flow. Each flow controller outputs one into three, that is, each flow controller is connected to 3 dip pipe pipelines. Each branch has a manual control valve and a flow switch. The flow switch provides a low-flow alarm for the dip pipe during RH vacuum refining. The branch hand valve cooperates with the established turnover preheating (slag melting, gunning) model 3, and uses the flow controller shared by 3 branches to provide detection results during turnover preheating (slag melting, gunning), and digitalizes the ventilation capacity data of the dip pipe. The specific parameters are shown in the following table:

[0084] It can be seen from this that press the model selection button 5 on the operation panel 8 to select the third learning model, the given flow rate is 30NM 3 / h, the leak detection flow rate is 30NM 3 / h, the pressure value range of the third learning model is 1.2Bar - 6Bar, and the flow rate value range is 15NM 3 / h - 24NM 3 / h; in this example, the ventilation performance of the dip pipe is mainly judged by the flow rate value and supplemented by the pressure value. In this example, if the parameter of the flow sensor is less than 15NM 3 / h, it is considered that the dip pipe is blocked and needs slag melting repair. If the parameter is greater than 24NM 3 / h, it is considered that the dip pipe is damaged or eroded and needs gunning repair.

[0085] Example 2

[0086] A set of: 210t RH vacuum refining equipment in a steel plant, with 12 dip pipes. Six flow controllers provide the circulating process flow. Each flow controller outputs one into two, that is, it is divided into 2 branches and connected to the dip pipe pipeline. Each branch has a manual control valve and a flow switch. The flow switch provides a low-flow alarm for the dip pipe during RH vacuum refining. The branch hand valve cooperates with the established turnover preheating (slag melting, gunning) model 3, and uses the flow controller shared by 2 branches to provide detection results during turnover preheating (slag melting, gunning), and digitalizes the ventilation capacity data of the dip pipe. The specific parameters are shown in the following table:

[0087]

[0088] It can be seen from this that by pressing the model selection button 5 on the operation panel 8 to select the third learning model, the given plugging detection flow rate is 40 NM 3 / h, the leak detection flow rate is 30 NM 3 / h, the pressure value range of the third learning model is 1.2 Bar - 6 Bar, and the flow rate value range is 15 NM 3 / h - 24 NM 3 / h; in this example, the ventilation performance of the dip tube is mainly judged by the flow rate value and supplemented by the pressure value. In this example, if the parameter of the flow sensor is less than 15 NM 3 / h, it is considered that the dip tube is blocked and slag melting maintenance is required. If the parameter is greater than 24 NM 3 / h, it is considered that the dip tube is damaged or eroded and gunning maintenance is required.

[0089] Example 3

[0090] A set of 60t RH vacuum refining equipment in a steel plant, with the 1# tank car and 2# tank car used alternately. There are 11 dip tubes at each station, and a total of 1 set of RH circulating gas control valve stations is used. The circulating process flow rate is provided by 11 flow controllers, with each flow controller controlling 1 branch, and the dip tubes at the stations are switched by solenoid valves. The branch solenoid valves cooperate with the established turnover preheating (slag melting, gunning) model 3, use the shared flow controller, and provide detection results during turnover preheating (slag melting, gunning) to digitalize the ventilation capacity data of the dip tubes. See the following table for details:

[0091]

[0092] It can be seen from this that by pressing the model selection button 5 on the operation panel 8 to select the third learning model, the given flow rate is 12 NM 3 / h, the leak detection flow rate is 12 NM 3 / h, the pressure value range of the third learning model is 1.2 Bar - 6 Bar, and the flow rate value range is 1 NM 3 / h - 11 NM 3 / h; in this example, the ventilation performance of the dip tube is mainly judged by the flow rate value and supplemented by the pressure value. In this example, if the parameter of the flow sensor is less than 1 NM 3 / h, it is considered that the dip tube is blocked and slag melting maintenance is required. If the parameter is greater than 11 NM 3 / h, it is considered that the dip tube is damaged or eroded and gunning maintenance is required.

[0093] Example 4

[0094] A set of 180t RH vacuum refining equipment in a steel plant, with 16 immersion tubes. The circulation process flow is provided by 16 flow controllers, and each flow controller controls 1 immersion tube branch. The first learning model for new lining, casting, and welding of the immersion tubes. When tamping the self-leveling material filled between the refractory bricks and the steel structure during new lining, there is a possibility of blocking the ventilation holes of the immersion tubes; during casting, the casting material is extremely likely to block the ventilation holes of the immersion tubes. This model mainly avoids the blocking of the ventilation holes of the immersion tubes during the new lining, casting, and welding of the immersion tubes, as well as the generation of leakage seams during welding.

[0095] The designed process flow of this project is 240 NM3 / h, with 16 immersion tubes arranged. The given flow value of the flow controller is 40 NM 3 / h. The data collected is as follows:

[0096]

[0097]

[0098] It can be seen from this that by pressing the model selection button 5 on the operation panel 8 to select the first learning model, the given flow is 40 NM 3 / h. The measured flow values and pressure values are as shown in the above table. The pressure value range of the first learning model is 0.8 Bar - 2.5 Bar, and the flow value range is 15 NM 3 / h - 32 NM 3 / h; in this example, the ventilation performance of the immersion tubes is mainly judged based on the flow value and supplemented by the pressure value. In this example, if the parameter of the flow sensor is less than 15 NM 3 / h, it is considered that the immersion tube is blocked, such as the casting material blocking the ventilation holes of the immersion tube; if the parameter is greater than 32 NM 3 / h, it is considered that the immersion tube leaks air, such as air leakage at the weld seam, etc.

[0099] Example 5

[0100] The second learning model for drying and baking. Drying and baking is carried out 48 hours after masonry is completed. This model is a pre - online detection, providing a reliable guarantee for the RH circulation ventilation ability. This model mainly avoids the blocking of the ventilation holes of the immersion tubes during the drying and baking process, and also provides detection for the ventilation ability of the immersion tubes of the RH riser tubes purchased externally.

[0101] The data collected by the second learning model is as follows:

[0102]

[0103]

[0104] The second learning model is for the detection before the dipping tube is put on line. Press the model selection button 5 on the operation panel 8 to select the second learning model. Set the given pressure to 4 Bar and the given flow rate to 40 NM3 / h. The test pressure value and flow rate value are as shown in the above table. The pressure value range of the second learning model is 0.8 Bar - 3 Bar, and the flow rate value range is 15 NM 3 / h - 32 NM 3 / h; In this example, if the parameter of the flow sensor is less than 15 NM 3 / h or greater than 32 NM 3 / h, the detection device prompts that the dipping tube needs to be maintained.

[0105] The embodiments of the present invention have been described above. The above description is exemplary, the above data are limited and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art of this technology without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, the practical application or the improvement of the technology in the market, or to enable other ordinary skill in the art of this technology to understand the disclosed embodiments.

Claims

1. An RH circulation dipping tube detection device, comprising an electrical control unit (7), which detects the state of the dipping tube according to data related to the detected condition state detected under different detection conditions of the dipping tube, and is characterized in that, The electrical control unit (7) applies a learning algorithm to a learning model containing parameters to obtain a threshold value, and makes a comparison through the threshold value to determine whether the dipping tube needs to be repaired or replaced; The device further includes a power supply (6) and a flow control unit (9) respectively connected to the electrical control unit (7); the inlet end of the flow control unit (9) is connected to an air source treatment device (12) through a pipeline, and the outlet end of the flow control unit (9) is connected to an output control joint (13) through a pipeline. The flow control unit (9) is used to control the fluid flow rate into the dipping tube; The device further includes a body (1), and an air inlet (3) and a dipping tube connection port (4) are provided on one side of the body (1); The air inlet (3) is communicated with the air source treatment device (12); the dipping tube connection port (4) is used to directly or indirectly connect the output control joint (13) with the dipping tube to be tested; The air source treatment device (12) includes a water filter and a voltage regulator; the water filter and the voltage regulator are sequentially communicated with the air inlet (3) through pipelines; several flow control units (9) are provided, and each flow control unit (9) is connected with several output control joints (13) through pipelines; A flow sensor (10) and a pressure sensor (11) are arranged in the pipeline at the outlet end of the flow control unit (9), and are respectively used for collecting the flow rate value and the pressure value of the branch of the device connected to the dipping tube; The flow sensor (10) and the pressure sensor (11) collect the pressure and flow rate data of the on-site pipeline, and send them to the electrical control unit (7) to construct a steady-state detection data set, and preprocess the steady-state detection data set. After being trained by a long short-term memory neural network, a learning model is constructed.

2. The RH circulation dipping tube detection device according to claim 1, wherein An operation panel (8) and a display device (2) are arranged on the surface of the body (1), and the display device (2) is connected to the electrical control unit (7); The display device (2) is internally provided with an SD card physical medium memory.

3. The RH circulation immersion tube detection device according to claim 2, characterized in that, At least a model selection button (5) and an indicator light are arranged on the operation panel (8), and the model selection button (5) and the indicator light are respectively connected to the electrical control unit (7).

4. The RH circulation immersion tube detection device according to claim 3, characterized in that, Construct a first learning model, a second learning model and a third learning model in sequence; The first learning model is used for the detection during the period of new construction, casting and welding of the dipping tube; The second learning model is used for the detection during the drying and baking period of the dipping tube; The third learning model is used for the detection during the preheating period of the dipping tube during on-line turnover.

5. A method for detecting an RH circulation immersion tube, characterized in that, Applying the RH circulation dipping tube detection device according to any one of claims 1-4, including the following steps: S1. Connect the process gas to the air inlet (3), and the process gas passes through the water filter and the voltage regulator in sequence for impurity filtration and voltage stabilization; connect the dipping tube to be tested to the output control joint (13) through a metal hose; S2. Select the model of different detection conditions of the dipping tube to be tested on the operation panel (8) according to the working conditions of the dipping tube to be tested; S3. The flow control unit (9) sets the flow rate of the process gas, obtains the results of the pipeline flow sensor (10) and the pressure sensor (11) inside the device, and sends them to the electrical control unit (7). S4. The electrical control unit (7) compares the obtained results with the thresholds of the corresponding models to determine whether the impregnation tube to be tested needs to be repaired, displays the results on the display device through the HMI, and stores the results.

6. The RH circulation immersion tube detection method according to claim 5, wherein, In S2, the selected models include: If the impregnation tube to be tested is in the period of new masonry, casting, or welding, select the first learning model. If the impregnation tube to be tested is in the period of dry baking, select the second learning model. If the impregnation tube to be tested is in the period of preheating during on-line turnover, select the third learning model.

7. The RH circulation dipping tube detection method according to claim 6, wherein In S4, determining whether the impregnation tube to be tested needs to be repaired includes: In the scenario of the first learning model, if the result of the flow sensor (10) is greater than the upper limit of the flow threshold of the first learning model, it is judged that the weld leaks; if the result of the flow sensor (10) is greater than the lower limit of the flow threshold of the first learning model and the pipeline back pressure exceeds the upper limit of the pressure threshold, it is judged that the masonry is blocked. In the scenario of the second learning model, if the result of the flow sensor (10) is less than the lower limit of the flow threshold of the second learning model and the pipeline back pressure exceeds the upper limit of the pressure threshold, it is judged that the air holes of the impregnation tube are blocked; if the result of the flow sensor (10) is greater than the upper limit of the flow threshold of the first learning model, it is judged that the weld leaks. In the scenario of the third learning model, if the result of the flow sensor (10) is greater than the upper limit of the flow threshold of the third learning model, it is judged that there is erosion and damage; if the result of the flow sensor (10) is less than the lower limit of the flow threshold of the third learning model and the pipeline back pressure exceeds the upper limit of the pressure threshold, it is judged that there is a blockage fault.

Images