Intelligent flat valve integrated with built-in pressure detection rod

Abstract

The invention relates to the technical field of flat valves, and discloses an intelligent flat valve integrated with a built-in pressure detection rod, comprising: a valve body in which a medium channel is formed and which is provided with a valve plate for controlling the on-off of the medium channel; the pressure detection unit is used for detecting the pressure of fluid in the valve body in real time, the pressure detection unit comprises a pressure detection pipe installed on the valve body and a pressure detection rod arranged in the pipe, and the pressure detection rod movably extends into the medium channel; and the data processing unit is used for uploading the processed pressure data to the pressure test table interlocking system, so that the pressure test table interlocking system records the pressure test data and generates a detection report. The pressure in the medium channel is detected through the pressure detection rod, so that an operator can master the change of the internal pressure without dismounting a valve or additionally mounting an instrument. Data are stored in an interlocking system, it is ensured that each product has a complete and verifiable pressure test record, and later quality tracing and auditing are facilitated.

Description

Technical Field

[0001] This invention relates to the field of flat plate valve technology, and more specifically, to an intelligent flat plate valve with an integrated built-in pressure sensing rod. Background Technology

[0002] Flat plate valves, as a common fluid control device, are widely used in petrochemical, natural gas transportation, water conservancy projects, and pressure vessels. Their main function is to control the opening and closing of the medium passage through the valve plate, achieving fluid isolation and flow. In industrial production and equipment pressure testing, the reliability and sealing performance of flat plate valves are crucial, especially in situations involving high-pressure, high-temperature, or corrosive media. Strict monitoring of the internal fluid pressure of the valve body is necessary to ensure safe equipment operation and compliance with relevant standards.

[0003] In existing technologies, pressure testing primarily relies on external auxiliary equipment, such as additional pressure gauges, pressure sensors, or temporary connection of testing instruments to the pipeline. This method requires operators to shut down the machine, disassemble valve flanges, or damage the sealing structure before testing to connect the testing device. This process is not only time-consuming and labor-intensive, increasing production downtime costs, but also prone to introducing the risk of secondary leakage, especially in high-pressure or corrosive media environments, where seal failure may lead to safety accidents.

[0004] More importantly, existing integrated pressure sensing designs (such as embedding sensors directly into the valve body) often overlook the effects of fluid dynamics. When the valve opens, the high-speed fluid in the medium channel exerts a strong impact on fixed sensing elements (such as pressure probes or sensing rods), causing deformation, wear, or even breakage. Especially under high pressure differential conditions, this mechanical damage not only shortens the service life of the sensing device but may also cause cascading failures due to fluid contamination or channel blockage caused by component fragments. Furthermore, existing systems lack intelligent data processing capabilities; pressure data typically relies on manual reading and recording and cannot be automatically uploaded to the central control system; detection reports must be manually compiled, resulting in poor data traceability.

[0005] Therefore, it is necessary to propose an intelligent flat plate valve with an integrated built-in pressure sensing rod to at least partially solve the problems existing in the prior art. Summary of the Invention

[0006] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. The summary section of this invention is not intended to limit the key features and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0007] To at least partially solve the above problems, the present invention provides an intelligent flat valve with an integrated built-in pressure sensing rod, comprising: The valve body has a medium channel inside and is equipped with a valve plate to control the opening and closing of the medium channel; The pressure detection unit is used to detect the fluid pressure inside the valve body in real time. The pressure detection unit includes a pressure detection tube installed on the valve body and a pressure detection rod located inside the tube. The pressure detection rod can be movably extended into the medium channel. The data processing unit is used to upload the processed pressure data to the pressure test bench interlocking system, so that the pressure test bench interlocking system can record the pressure test data and generate a test report.

[0008] Preferably, a magnetic protection component is provided inside the pressure detection tube, the magnetic protection component comprising: A floating magnetic base is movably installed inside the pressure detection tube. The floating magnetic base is set as a ring and multiple magnetic blocks are embedded along the circumference. The pressure detection rod passes through the center of the floating magnetic base. A reset spring is connected between the bottom of the pressure detection tube and the floating magnetic base. The reference magnetic base is installed at the top of the pressure detection tube. When the reference magnetic base is energized, it generates magnetism and produces a repulsive force on the floating magnetic base. The wedge-shaped guide block, connected to the wall of the pressure detection tube and located below the floating magnetic seat, is used to guide and limit the floating magnetic seat towards the center.

[0009] Preferably, a protective groove is provided at the bottom of the pressure detection tube, and a through hole is provided on the protective groove for the pressure detection rod to pass through. The through hole and the pressure detection rod form a clearance fit. A flexible sealing ring is connected to the protective groove to block the through hole, and the flexible sealing ring is sleeved on the pressure detection rod. A flow guide ring is connected at the opening of the protective groove, and the flow guide ring contracts towards the center near the port of the medium channel.

[0010] Preferably, the floating magnetic base is connected to the pressure detection rod via a micro-deflection assembly, which includes: A retaining ring is installed on the pressure testing rod; The first deflection component, multiple first deflection components are evenly connected to the outer wall of the fixed ring, the outer side of the first deflection component is set as an arc surface, and a limit groove is provided on the arc surface; The second deflector is evenly connected to the inner wall of the floating magnetic base. The outer side of the second deflector is provided with a slot that engages with the floating magnetic base, and the inner side is provided with an arc-shaped surface that matches the outer wall of the first deflector. The second deflector is provided with a limiting slider that is slidably connected to the limiting groove.

[0011] Preferably, an indicator groove is provided on the upper part of the pressure detection tube, and a magnetic indicator ring is slidably disposed in the indicator groove with a clearance fit between the two. Scale lines are provided in the indicator groove, and the cross-section of the magnetic indicator ring is circular, attracting each other with the floating magnetic base.

[0012] Preferably, the data processing unit includes: The signal processing module is used to amplify, filter, and perform analog-to-digital conversion on the raw signal output by the pressure detection unit to obtain pressure data. The upload module is used to upload pressure data to the pressure test bench interlocking system via wired or wireless means; the pressure test bench interlocking system deploys a large-scale model for pressure test data analysis; The pressure test result output module is used to output the pressure data test report obtained by the pressure test bench interlocking system through analysis of the pressure data.

[0013] Preferably, the training steps of the large-scale model for pressure test data analysis in the upload module include: Collect historical pressure test data, including the first pressure-time curve, pressure test parameters, valve specification parameters, and pressure test result labels; Historical pressure test data is preprocessed, including data cleaning, normalization, and feature extraction. The extracted features include pressure rise rate, pressure peak, pressure stability, pressure fluctuation amplitude, and leakage rate. The preprocessed data is divided into training set, validation set and test set; Construct a deep learning network model, which includes an input layer, a multi-layer convolutional neural network layer, a long short-term memory network layer, and a fully connected output layer; The deep learning network model is trained using a training set, and the model parameters are optimized using the backpropagation algorithm, enabling the model to identify qualified and abnormal stress test modes. Use the validation set to validate the trained model and fine-tune the parameters; The model performance is evaluated using a test set. When the model accuracy reaches a preset threshold, the model is deployed as a large-scale model for stress test data analysis.

[0014] Preferably, the analysis process of pressure data by the large-scale pressure test data analysis model includes: The system receives pressure data transmitted from the upload module in real time and generates a second pressure-time curve for the current pressure test. The second pressure-time curve is input into the large-scale pressure test data analysis model, and the model automatically extracts the curve features and compares them with the standard pressure test mode. Identify abnormal events during the pressure test process, including excessively rapid pressure rise, pressure peak exceeding limits, abnormal fluctuations during the pressure holding phase, and suspected leaks. Generate a pressure data test report, which includes the test pass / fail criteria, abnormal event annotations, key parameter values, and improvement suggestions.

[0015] Preferably, the large-scale model for pressure test data analysis also has a self-learning function, including: Continuously collect new pressure test data and its manually labeled actual results; Based on preset time intervals, the large-scale model for stress test data analysis is incrementally trained using new data to update the model parameters; Establish a model version management mechanism to retain historical model versions and record model update logs; The system compares the performance of the new and old models and automatically switches deployments when the new model outperforms the old model.

[0016] Preferably, the wired or wireless methods in the upload module include: It uses RS485, Modbus or Ethernet protocols and is connected to the interlocking system of the pressure test bench via physical cables; or, Wireless data transmission is achieved using LoRa, Zigbee, WiFi, or 4G / 5G protocols.

[0017] Compared to existing technologies, this invention provides an intelligent flat valve with an integrated built-in pressure sensing rod, offering at least the following advantages: It integrates intelligent detection functions onto the valve body, using the pressure sensing rod to detect pressure within the media channel. This allows operators to monitor internal pressure changes without disassembling the valve or installing additional instruments. Furthermore, the movable pressure sensing rod effectively reduces deformation and damage under fluid impact, extending its service life. Data is stored in an interlocking system, ensuring each product has complete and verifiable pressure test records, facilitating subsequent quality traceability and auditing.

[0018] The present invention discloses an intelligent flat plate valve with an integrated built-in pressure detection rod. Other advantages, objectives and features of the present invention will be apparent in part from the following description, and in part from the understanding of those skilled in the art through study and practice of the invention. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of an intelligent flat valve with an integrated built-in pressure detection rod according to the present invention. Figure 1 ; Figure 2 This is a schematic diagram of the structure of an intelligent flat valve with an integrated built-in pressure detection rod according to the present invention. Figure 2 ; Figure 3 This is a schematic diagram of the installation structure of the pressure detection unit in this invention; Figure 4 This is a schematic cross-sectional view of the pressure detection tube in this invention; Figure 5 For the present invention Figure 4 A magnified schematic diagram of the partial structure at point A in the middle; Figure 6 This is a schematic diagram of the micro-deflection component in this invention.

[0020] In the diagram: 1. Valve body; 2. Valve plate; 3. Medium channel; 4. Pressure detection tube; 5. Pressure detection rod; 11. Floating magnetic base; 12. Magnetic block; 13. Reference magnetic base; 14. Return spring; 15. Wedge-shaped guide block; 16. Protective groove; 17. Through hole; 18. Flexible sealing ring; 19. Flow guide ring; 21. Fixing ring; 22. First deflector; 23. Limiting slide groove; 24. Second deflector; 25. Bayonet; 26. Limiting slider; 27. Indicator area slide groove; 28. Magnetic indicator ring; 29. ​​Scale line. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments, so that those skilled in the art can implement it based on the description.

[0022] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0023] Example 1: As Figures 1 to 3 As shown, the present invention provides an intelligent flat valve with an integrated built-in pressure sensing rod, comprising: The valve body 1 has a medium channel 3 inside and is equipped with a valve plate 2 for controlling the opening and closing of the medium channel 3; The pressure detection unit is used to detect the fluid pressure inside the valve body 1 in real time. The pressure detection unit includes a pressure detection tube 4 installed on the valve body 1 and a pressure detection rod 5 located inside the tube. The pressure detection rod 5 can be movably extended into the medium channel 3. The data processing unit is used to upload the processed pressure data to the pressure test bench interlocking system, so that the pressure test bench interlocking system can record the pressure test data and generate a test report.

[0024] The working principle and beneficial effects of the above technical solution are as follows: This invention provides an intelligent flat valve with an integrated built-in pressure detection rod. The valve plate 2 is controlled by a vertical movement control channel for opening and closing. A valve plate drive structure is located on the upper part of the valve body 1 to control the vertical movement of the valve plate 2, which is operated manually or hydraulically. A valve seat is provided on the valve body 1, pressed against both sides of the valve plate 2. A pressure detection tube 4 is provided on the valve body 1, and a pressure detection rod 5 is installed inside the pressure detection tube 4 and can move vertically. The end of the pressure detection rod 5 is set as a pressure detection head. When no pressure detection is being performed, the pressure detection rod 5 is retracted into the pressure detection tube 4, offset from the main flow channel of the medium channel 3, to avoid large fluid impact forces when the valve is opened, which could cause deformation and damage to the pressure detection rod 5. When pressure detection is being performed, the pressure detection rod 5 is moved so that it extends into the medium channel 3, directly contacting the main flow channel of the medium channel 3 to detect the fluid pressure inside the valve body 1.

[0025] The detection head of pressure detection rod 5 transmits the detection signal to the data processing unit. The data processing unit amplifies, filters, and performs analog-to-digital conversion on the raw signal, and then uploads it to the pressure test bench interlocking system in real time via wired or wireless means. After receiving the data, the pressure test bench system automatically records the pressure-time curve and generates a test report containing test parameters, pass / fail criteria, etc., based on a preset template.

[0026] The above structural design provides an intelligent flat panel valve with an integrated built-in pressure sensing rod. Intelligent detection functionality is integrated into the valve body 1, using the pressure sensing rod 5 to detect the pressure within the media channel 3. This allows operators to monitor internal pressure changes without disassembling the valve or installing additional instruments. The movable pressure sensing rod 5 effectively reduces deformation and damage under fluid impact, extending its service life. Data is stored in an interlocking system, ensuring each product has complete and verifiable pressure test records, facilitating subsequent quality traceability and auditing.

[0027] Example 2: As Figures 3 to 6 As shown, based on the above embodiment 1, a magnetic protection component is provided inside the pressure detection tube 4. The magnetic protection component includes: A floating magnetic base 11 is movably set inside the pressure detection tube 4. The floating magnetic base 11 is set as a ring and multiple magnetic blocks 12 are embedded along the circumference. The pressure detection rod 5 passes through the center of the floating magnetic base 11. A reset spring 14 is connected between the bottom of the pressure detection tube 4 and the floating magnetic base 11. The reference magnetic base 13 is installed at the top of the pressure detection tube 4. When the reference magnetic base 13 is energized, it generates magnetism and generates a repulsive force on the floating magnetic base 11. The wedge-shaped guide block 15 is connected to the wall of the pressure detection tube 4 and located below the floating magnetic seat 11. It is used to guide and limit the floating magnetic seat 11 towards the center.

[0028] The working principle and beneficial effects of the above technical solution are as follows: During pressure testing, the pressure detection rod 5 is protected, as described in Chinese Patent Publication No. CN222544983U, using an openable and closable movable plate to protect the detection head. However, this method is affected by the presence of particulate matter in the medium, causing the detection process to malfunction. In particular, the presence of iron filings or magnetic particles in the medium can cause sensor data drift when these particles come into contact with the detection head of the pressure detection rod 5, affecting the accuracy and long-term stability of the pressure detection.

[0029] A magnetic protection assembly is installed inside the pressure detection tube 4. The floating magnetic seat 11 is connected to the bottom of the tube by a return spring 14, so that the floating magnetic seat 11 is located relatively far from the valve body 1 under normal conditions. After the reference magnetic seat 13 is energized, it generates a magnetic field with the same pole as the floating magnetic seat 11, forming a magnetic repulsive force that suspends the floating magnetic seat 11 and pushes it away from the bottom of the tube, increasing its length extending into the medium channel 3. A wedge-shaped guide block 15 is set on the tube wall of the pressure detection tube 4, and guides the floating magnetic seat 11 to the center of the tube through its inclined surface, ensuring the accurate initial position of the pressure detection rod 5.

[0030] As the floating magnetic base 11 approaches the pipe opening, its attraction to magnetic particles in the medium channel 3 increases. When iron filings or magnetic particles in the medium enter the pressure detection pipe 4 opening with the fluid, the magnetic block 12 on the floating magnetic base 11 attracts them near the opening. Simultaneously, as the extension length of the pressure detection rod 5 increases, the distance between it and the attracted magnetic particles increases, preventing the magnetic particles from contacting its detection end. After detection, the power is cut off, and the reset spring 14 pulls the floating magnetic base 11 back to its initial position. The attraction of the floating magnetic base 11 to iron filings and magnetic particles decreases, and the iron filings and magnetic particles fall back into the medium channel 3 under gravity and are washed away, achieving self-cleaning.

[0031] Through the above structural design, the retracted and extended states of the pressure detection rod 5 are controlled by the combination of magnetic force and spring force. The non-contact setting of the reference magnetic base 13 and the floating magnetic base 11 reduces coupling in the transmission and avoids situations where the pressure detection rod 5 cannot move flexibly due to jamming at any position. At the same time, the magnetic force is used to attract iron filings and magnetic particles in the medium channel 3, preventing them from adhering to or impacting the pressure detection rod 5 and interfering with its detection, thereby improving the pressure detection accuracy and ensuring the reliability and stability of the valve in harsh industrial environments.

[0032] Example 3: As Figures 3 to 6As shown, based on the above embodiment 2, a protective groove 16 is provided at the bottom of the pressure detection tube 4, and a through hole 17 is provided on the protective groove 16 for the pressure detection rod 5 to pass through. The through hole 17 and the pressure detection rod 5 form a clearance fit. A flexible sealing ring 18 is connected to the protective groove 16 to block the through hole 17. The flexible sealing ring 18 is sleeved on the pressure detection rod 5. A guide ring 19 is connected at the groove opening of the protective groove 16, and the guide ring 19 shrinks towards the center near the port of the medium channel 3.

[0033] The working principle and beneficial effects of the above technical solution are as follows: A protective groove 16 is provided at the bottom of the pressure detection tube 4. When the valve is not open or in the initial stage of opening, the pressure detection rod 5 is completely located within the protective groove 16. The through hole 17 forms a clearance fit with the pressure detection rod 5, ensuring that the pressure detection rod 5 can move freely. A flexible sealing ring 18 is provided on the protective groove 16 to prevent the medium from entering the interior of the pressure detection tube 4. A guide ring 19 is provided at the opening of the protective groove 16. When the medium flows at high speed, the guide ring 19 guides the fluid to the center, reducing the direct impact on the end of the pressure detection rod 5. After being guided, part of the fluid enters the protective groove 16 and maintains a low flow rate, preventing excessive deformation and damage to the pressure detection rod 5 and extending its service life.

[0034] Example 4: Figures 3 to 6 As shown, based on the above embodiment 2, the floating magnetic base 11 is connected to the pressure detection rod 5 through a micro-deflection assembly, which includes: The retaining ring 21 is installed on the pressure detection rod 5; First deflector 22, multiple first deflector 22 are evenly connected to the outer wall of the fixed ring 21, the outer side of the first deflector 22 is set as an arc surface, and a limit groove 23 is provided on the arc surface; The second deflector 24, a plurality of second deflectors 24 are evenly connected to the inner wall of the floating magnetic base 11. The outer side of the second deflector 24 is provided with a slot 25 that engages with the floating magnetic base 11, and the inner side is provided with an arc-shaped surface that matches the outer wall of the first deflector 22. The second deflector 24 is provided with a limiting slider 26 that is slidably connected to the limiting groove 23.

[0035] The working principle and beneficial effects of the above technical solution are as follows: Under fluid pressure, the pressure detection rod 5 will slightly deflect. The traditional rigid connection of the pressure detection rod 5 is prone to transmission jamming, stress concentration or even breakage, affecting the continuity and safety of detection. A micro-deflection component is set on the floating magnetic base 11. When the pressure detection rod 5 deflects slightly by ≤5° due to fluid pressure, the first deflection component 22 is driven to deflect through the fixed ring 21. The first deflection component 22 slides relative to the second deflection component 24 between the arc surfaces, and the limiting slider 26 slides in the limiting groove 23. The deflection angle is limited by the length of the limiting groove 23.

[0036] Through the above structural design, by combining rigid force transmission with flexible adaptation, the pressure detection rod 5 is allowed to be freely fine-tuned within a preset angle range under the action of fluid pressure, avoiding jamming and stress concentration. While ensuring that the axial movement of the pressure detection rod 5 is not affected, micro-angle deflection compensation of the floating magnetic base 11 relative to the pressure detection rod 5 is achieved, improving the self-adaptive capability of the pressure detection process and enhancing its reliability.

[0037] Example 5: Figures 3 to 6 As shown, based on the above embodiment 4, an indicator area groove 27 is provided on the upper part of the pressure detection tube 4. A magnetic indicator ring 28 is slidably disposed in the indicator area groove 27 and the two are in clearance fit. A scale line 29 is provided in the indicator area groove 27. The cross-section of the magnetic indicator ring 28 is circular and attracts each other with the floating magnetic seat 11.

[0038] The working principle and beneficial effects of the above technical solution are as follows: Pressure detection data is transmitted only through an electronic system, and on-site operators cannot directly see the real-time pressure value, relying on a display screen or remote terminal, which is inconvenient for quick judgment and emergency handling. By setting an indicator component on the pressure detection tube 4, when the floating magnetic seat 11 moves, the magnetic attraction causes the magnetic indicator ring 28 to move accordingly. Its circular cross-section design reduces sliding resistance, ensuring that it does not jam in vibration or tilting environments. Operators can directly read the current pressure value by observing the position of the indicator ring relative to the scale line. When the angle deflection of the pressure detection rod 5 reaches the preset angle limit value, the wedge-shaped guide block 15 initially limits the floating magnetic seat 11 and the pressure detection rod 5. Since the guide surface of the wedge-shaped guide block 15 is set as an inclined surface, when the pressure detection rod 5 is subjected to excessive fluid pressure, it will cause the floating magnetic seat 11 to deflect relative to the wedge-shaped guide block 15. Under the action of magnetic force, the magnetic indicator ring 28 deflects accordingly, causing the corresponding scale line 29 on both sides to be in different positions. Operators can determine the deflection state of the pressure detection rod 5 by observing the numerical difference on both sides of the indicator ring 28.

[0039] Example 6: Based on Example 1 above, the data processing unit includes: The signal processing module is used to amplify, filter, and perform analog-to-digital conversion on the raw signal output by the pressure detection unit to obtain pressure data.

[0040] The upload module is used to upload pressure data to the pressure test bench interlocking system via wired or wireless means; the pressure test bench interlocking system deploys a large-scale model for pressure test data analysis.

[0041] In this embodiment, the pressure testing bench interlocking system is a host computer management system deployed on the production site or a remote server. It is responsible for receiving pressure testing data uploaded from multiple valves, and for centralized storage, analysis, monitoring, and report generation. The pressure testing data analysis model is an artificial intelligence algorithm model deployed within the pressure testing bench interlocking system. Built based on deep learning technology, it learns from a large amount of historical pressure testing data to automatically identify the pass / fail status of the pressure testing process, diagnose abnormalities, and provide suggestions.

[0042] The pressure test result output module is used to output the pressure data test report obtained by the pressure test bench interlocking system through analysis of the pressure data.

[0043] In this embodiment, the pressure data detection report is a structured document generated by the pressure test bench interlocking system based on large model analysis. It includes the determination of whether the pressure test is qualified, the detected abnormalities, the values ​​of key parameters, and improvement suggestions.

[0044] The working principle and beneficial effects of the above technical solution are as follows: This invention uploads processed pressure data to a test bench interlocking system that deploys a large-scale test data analysis model. The test bench interlocking system receives test data uploaded from multiple valves and performs centralized storage, analysis, monitoring, and report generation, outputting an official record and traceable certificate of test quality. This is more convenient and intelligent because it does not require damaging the original sealing structure of the flat valve or manually compiling test reports.

[0045] Example 7: Based on Example 6 above, the training steps of the large-scale model for pressure test data analysis in the upload module include: Collect historical pressure test data, including the first pressure-time curve, pressure test parameters, valve specification parameters, and pressure test result labels.

[0046] In this embodiment, historical pressure test data consists of real pressure test records accumulated from past production practices, including information such as the first pressure-time curve, pressure test parameters, valve specification parameters, and pressure test result labels. The first pressure-time curve is a graph plotted with time on the horizontal axis and pressure on the vertical axis, which fully records the trajectory of pressure change over time during historical pressure tests, including the pressurization stage, pressure holding stage, and pressure release stage, reflecting the dynamic characteristics of the pressure test process. The pressure test parameters are the parameter values ​​set in the pressure test process, such as the target pressure test pressure, pressure holding time, pressurization rate, type of test medium, and ambient temperature. The valve specification parameters are the design and manufacturing parameters of the valves tested in the history of pressure testing, including nominal diameter, pressure rating, material, and sealing type. The pressure test result labels are the classification of pressure test results manually judged and marked by quality inspectors, including: qualified, unqualified, and the specific reasons for unqualified (such as leakage, overpressure, etc.).

[0047] Historical pressure test data is preprocessed, including data cleaning, normalization, and feature extraction. The extracted features include pressure rise rate, pressure peak, pressure stability, pressure fluctuation amplitude, and leakage rate.

[0048] In this embodiment, data cleaning refers to identifying and processing errors, missing values, duplicates, and outliers in the data, such as deleting erroneous records with negative pressure values, filling in missing points in curves, and removing obviously unreasonable data. Normalization refers to mapping data with different dimensions and numerical ranges to the same standard range, for example, normalizing pressure values ​​from 0-50 MPa to 0-1. Feature extraction refers to calculating or extracting key indicators or feature values ​​that reflect the essential characteristics of the data from the original data (such as the first pressure-time curve), such as: pressure rise rate, pressure peak value, pressure stability, pressure fluctuation amplitude, and leakage rate; pressure rise rate refers to the increase in pressure per unit time, pressure peak value refers to the maximum pressure value that occurs during the pressure test, pressure stability refers to the degree of pressure fluctuation during the pressure holding phase, expressed as standard deviation, the smaller the value, the more stable the pressure and the better the sealing performance; pressure fluctuation amplitude refers to the difference between the maximum and minimum pressure values ​​during the pressure holding phase; leakage rate refers to the rate of pressure decrease per unit time during the pressure holding phase.

[0049] The preprocessed data is divided into training set, validation set and test set.

[0050] In this embodiment, the data partitioning ratio is preset manually, for example: 70% for the training set, 15% for the validation set, and 15% for the test set.

[0051] A deep learning network model is constructed, which includes an input layer, a multi-layer convolutional neural network layer, a long short-term memory network layer, and a fully connected output layer.

[0052] In this embodiment, the input layer receives external input data and passes it to subsequent network layers for processing. The multi-layer convolutional neural network is a network structure composed of multiple stacked convolutional layers. By sliding convolutional kernels across the data, it automatically identifies key shapes, abrupt changes, slope variations, and other features in the curve. The long short-term memory network layer learns and remembers long-term dependencies in the pressure-time curve. The fully connected output layer maps the high-dimensional features extracted from the preceding layers to the final output result (such as the probability of "qualified" or "unqualified"). Each neuron is connected to all neurons in the previous layer, achieving a comprehensive feature-based judgment.

[0053] The deep learning network model is trained using a training set, and the model parameters are optimized using the backpropagation algorithm, enabling the model to identify qualified and abnormal stress test modes.

[0054] In this embodiment, the backpropagation algorithm refers to calculating the error between the model output and the true label, propagating the error layer by layer from the output layer to the input layer, and updating the weight parameters of each layer according to the error gradient, so that the model output gradually approaches the true value. Model parameters refer to the weight and bias values ​​of each connection in the neural network, which are continuously adjusted and optimized through the training process. Qualified pressure test patterns refer to typical characteristic patterns summarized from historical qualified pressure test data, such as stable pressure rise, stable pressure holding phase, low leakage rate, and no abnormal fluctuations; the model can identify these patterns after learning. Abnormal pressure test patterns refer to typical characteristic patterns summarized from historical unqualified pressure test data, such as sudden pressure changes, rapid pressure drop during holding, drastic fluctuations, and peak values ​​exceeding limits; the model can identify these and issue an alarm after learning.

[0055] Use the validation set to validate the trained model and fine-tune its parameters.

[0056] In this embodiment, validation and parameter tuning involve adjusting the model's hyperparameters (such as learning rate, batch size, number of network layers, number of neurons, etc.) based on the performance of the validation set to find the optimal configuration and obtain the best model performance.

[0057] The model performance is evaluated using a test set. When the model accuracy reaches a preset threshold, the model is deployed as a large-scale model for stress test data analysis.

[0058] In this embodiment, the preset threshold is set manually, for example, 95%.

[0059] The working principle and beneficial effects of the above technical solution are as follows: Traditional pressure test quality assessment relies on quality inspectors observing pressure gauge readings or graphs and judging whether a test is qualified based on experience. In batch pressure test scenarios, this consumes a lot of manpower and is inefficient.

[0060] This invention collects pressure testing data from multiple sources, including historical production records, automatic recording systems on pressure testing benches, and quality inspection databases. Each data record contains a complete pressure-time curve, pressure testing parameters, valve specifications, and pressure testing result labels marked by professional quality inspectors. This multi-dimensional data provides rich learning materials for the model.

[0061] Next, the historical pressure test data is preprocessed. First, data cleaning is performed to remove obviously erroneous records, and interpolation methods are used to fill in missing points in the curves, ensuring data integrity and accuracy. Then, normalization is performed to eliminate the influence of dimensions. Next, feature extraction is performed to calculate key characteristic values ​​reflecting the pressure test quality from high-dimensional time-series data such as pressure-time curves: pressure rise rate (reflecting the smoothness of the pressurization process), peak pressure (determining whether overpressure has occurred), pressure holding stability (quantifying the degree of fluctuation during the pressure holding phase using standard deviation), pressure fluctuation amplitude (the difference between the maximum and minimum pressure values ​​during the pressure holding phase), and leakage rate (the rate of pressure drop during the pressure holding phase, a key indicator of sealing performance). These features transform the original curves into quantifiable numerical vectors.

[0062] After processing, a deep learning network model containing an input layer, multiple convolutional neural network layers, a long short-term memory (LSTM) network layer, and a fully connected output layer is trained. The input layer receives the normalized first pressure-time curve and feature vector. The multiple convolutional neural network layers automatically extract local pattern features, such as pressure abrupt changes, curve slope changes, and waveform features, by sliding convolutional kernels across the curve, eliminating the need for manually designed feature extraction rules. The LSTM network layer specifically processes time-series data, memorizing the correlation information between different stages of the pressure test process. For example, the rate of pressurization affects the stability of the holding pressure stage; LSTM can capture this long-term dependency. The fully connected output layer fuses the high-dimensional features extracted by the CNN and LSTM, mapping them to the final output space through weighted summation and nonlinear transformation of multiple neurons. The Softmax activation function is used to output probability values ​​for the pass and fail categories.

[0063] The training process is an iterative optimization process. After a batch of data is input into the model, the model performs forward computation based on the current parameters to obtain the prediction result. Then, it calculates the error between the prediction result and the true label. Next, it uses the backpropagation algorithm to calculate the gradient of the loss function with respect to each parameter (i.e., the direction and magnitude of the influence of parameter changes on the error). Finally, it updates the parameters based on the gradient (adjusting the weights along the direction that reduces the error). This process is repeated dozens to hundreds of times, and the model parameters are gradually optimized, making the model output closer and closer to the true label. During the training process, the model gradually learns to recognize the characteristics of qualified pressure testing patterns (smooth pressure rise, stable pressure holding, low leakage rate, no abnormal fluctuations) and the characteristics of abnormal pressure testing patterns (violent pressure fluctuations, significant pressure drop, pressure exceeding limits, etc.).

[0064] After each training epoch, the model performance is evaluated using a validation set, calculating metrics such as accuracy, precision, recall, and F1 score. If the validation set performance is poor or overfitting occurs, the model hyperparameters are adjusted, such as reducing the learning rate, decreasing the number of network layers, or increasing regularization, and then retraining is performed. This process is repeated to find the optimal model configuration.

[0065] After training, a test set is used for final evaluation. The test set data is completely independent and has not participated in any training or tuning process. When the model accuracy reaches a preset threshold, the model is deployed as a large-scale model for stress test data analysis.

[0066] This invention trains a large-scale model for pressure test data analysis using historical pressure test data, thereby achieving automated quality judgment and significantly improving the efficiency of pressure test data analysis. The model can identify complex patterns, improve the accuracy of judgment, and also greatly reduce labor costs.

[0067] Example 8: Based on Example 7 above, the analysis process of pressure data by the large-scale pressure test data analysis model includes: The system receives pressure data transmitted from the upload module in real time and generates a second pressure-time curve for the current pressure test.

[0068] In this embodiment, the current pressure test refers to the ongoing pressure test process. The second pressure-time curve is the pressure-time curve generated by the current pressure test.

[0069] The second pressure-time curve is input into the pressure test data analysis model, and the model automatically extracts the curve features and compares them with the standard pressure test mode.

[0070] In this embodiment, curve features include shape, slope, peak value, and fluctuations. Standard pressure test modes include qualified pressure test modes and abnormal pressure test modes. Comparison refers to calculating the feature similarity between the curve features of the current pressure test curve and the curve features of the standard pressure test mode to determine which mode the second pressure-time curve conforms to.

[0071] Identify abnormal events during the pressure test process, including excessively rapid pressure rise, pressure peak exceeding limits, abnormal fluctuations during the pressure holding phase, and suspected leaks.

[0072] In this embodiment, an abnormal event refers to the event corresponding to an abnormal pressure test pattern that is successfully matched in the comparison results.

[0073] Generate a pressure data test report, which includes the test pass / fail criteria, abnormal event annotations, key parameter values, and improvement suggestions.

[0074] In this embodiment, the pressure test pass / fail determination refers to the conclusion explicitly marked in the report, including both pass and fail. Abnormal event labeling refers to all abnormal situations detected during the pressure test, including information such as abnormality type, occurrence time, and severity. Key parameter values ​​include the maximum pressure, average holding pressure, pressure rise rate, leakage rate, and test duration during the pressure test. Improvement suggestions are cause analyses and improvement measures automatically generated by the model based on identified abnormal pressure test patterns and historical experience knowledge, such as "suggest checking the valve seat sealing surface" and "suggest reducing the pressurization rate."

[0075] The working principle and beneficial effects of the above technical solution are as follows: This invention converts pressure data into a second pressure-time curve in real time, and inputs the second pressure-time curve into a trained pressure test data analysis model for real-time intelligent analysis. It then outputs a pressure data detection report in a timely manner, helping operators to quickly locate problems and take corrective measures. This achieves precise closed-loop control and improves production quality.

[0076] Example 9: Based on Example 8 above, the large-scale model for pressure test data analysis also has a self-learning function, including: We continuously collect new pressure test data and their manually labeled results.

[0077] Based on preset time intervals, the large-scale model for stress test data analysis is incrementally trained using new data to update the model parameters.

[0078] In this embodiment, the preset time interval is manually set, for example, weekly. Incremental training refers to updating and fine-tuning local parameters using new data based on existing model parameters. Updating model parameters refers to adjusting the weights and biases in the neural network through incremental training.

[0079] Establish a model version management mechanism to retain historical model versions and record model update logs.

[0080] In this embodiment, the model version management mechanism is a set of rules for managing different versions of the model. Historical model versions are model files that have been trained and deployed in the past. The model update log is a document that records detailed information about each model update, including update time, amount of data used, training parameters, changes in performance metrics, and key improvements.

[0081] The system compares the performance of the new and old models and automatically switches deployments when the new model outperforms the old model.

[0082] In this embodiment, the model performance is confirmed through A / B testing. Once the test confirms that the new model performs better than the old model, the new model is automatically set as the default model without manual intervention.

[0083] The working principle and beneficial effects of the above technical solution are as follows: This invention introduces a self-learning mechanism into the application of large-scale stress test data analysis models. It utilizes new stress test data containing manually labeled real results for incremental training and local optimization, iterating the model iteratively. Simultaneously, it retains historical model versions and automatically switches between old and new models based on test results, improving the model's ability to identify edge cases and new scenarios, thus expanding its applicability.

[0084] Example 10: Based on Example 6 above, the wired or wireless methods in the upload module include: It uses RS485, Modbus or Ethernet protocols and is connected to the interlocking system of the pressure test bench via physical cables; or, Wireless data transmission is achieved using LoRa, Zigbee, WiFi, or 4G / 5G protocols.

[0085] The working principle and beneficial effects of the above technical solution are as follows: This invention introduces both wired and wireless methods to achieve communication between the pressure detection unit and the pressure testing bench interlocking system, thereby improving communication stability.

[0086] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0087] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0088] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. Other modifications can be easily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A smart flat valve with an integrated built-in pressure detection rod, characterized in that, include: The valve body (1) has a medium channel (3) inside and is equipped with a valve plate (2) for controlling the opening and closing of the medium channel (3). The pressure detection unit is used to detect the fluid pressure inside the valve body (1) in real time. The pressure detection unit includes a pressure detection tube (4) installed on the valve body (1) and a pressure detection rod (5) located inside the tube. The pressure detection rod (5) can be movably extended into the medium channel (3). The data processing unit is used to upload the processed pressure data to the pressure test bench interlocking system, so that the pressure test bench interlocking system can record the pressure test data and generate a test report.

2. The intelligent flat valve with integrated built-in pressure detection rod according to claim 1, characterized in that, A magnetic protection component is installed inside the pressure detection tube (4). The magnetic protection component includes: A floating magnetic seat (11) is movably set inside the pressure detection tube (4). The floating magnetic seat (11) is set as a ring and multiple magnetic blocks (12) are embedded along the circumference. The pressure detection rod (5) passes through the center of the floating magnetic seat (11). A reset spring (14) is connected between the bottom of the pressure detection tube (4) and the floating magnetic seat (11). The reference magnetic base (13) is installed at the top of the pressure detection tube (4). When the reference magnetic base (13) is energized, it generates magnetism and generates a repulsive force on the floating magnetic base (11). The wedge-shaped guide block (15) is connected to the wall of the pressure detection tube (4) and located below the floating magnetic seat (11) to guide and limit the floating magnetic seat (11) towards the center.

3. The intelligent flat valve with integrated built-in pressure detection rod according to claim 2, characterized in that, The pressure detection tube (4) has a protective groove (16) at the bottom. The protective groove (16) has a through hole (17) through which the pressure detection rod (5) passes. The through hole (17) and the pressure detection rod (5) form a clearance fit. A flexible sealing ring (18) that blocks the through hole (17) is connected to the protective groove (16). The flexible sealing ring (18) is sleeved on the pressure detection rod (5). A guide ring (19) is connected to the groove opening of the protective groove (16), and the guide ring (19) shrinks towards the center near the port of the medium channel (3).

4. The intelligent flat valve with integrated built-in pressure detection rod according to claim 2, characterized in that, The floating magnetic base (11) is connected to the pressure detection rod (5) via a micro-deflection assembly, which includes: A retaining ring (21) is installed on the pressure detection rod (5); First deflection element (22), multiple first deflection elements (22) are evenly connected to the outer wall of the fixed ring (21), the outer side of the first deflection element (22) is set as an arc surface, and a limit groove (23) is provided on the arc surface. The second deflector (24) has multiple second deflectors (24) evenly connected to the inner wall of the floating magnetic base (11). The outer side of the second deflector (24) is provided with a slot (25) that engages with the floating magnetic base (11), and the inner side is provided with an arc-shaped surface that matches the outer wall of the first deflector (22). The second deflector (24) is provided with a limiting slider (26) that is slidably connected to the limiting groove (23).

5. The intelligent flat valve with integrated built-in pressure detection rod according to claim 2, characterized in that, The pressure detection tube (4) is provided with an indicator area groove (27) at the top. A magnetic indicator ring (28) is slidably arranged in the indicator area groove (27) and the two are in clearance fit. A scale line (29) is provided in the indicator area groove (27). The cross section of the magnetic indicator ring (28) is set as a circle and attracts each other with the floating magnetic seat (11).

6. The intelligent flat valve with integrated built-in pressure detection rod according to claim 1, characterized in that, The data processing unit includes: The signal processing module is used to amplify, filter, and perform analog-to-digital conversion on the raw signal output by the pressure detection unit to obtain pressure data. The upload module is used to upload pressure data to the pressure test bench interlocking system via wired or wireless means; the pressure test bench interlocking system deploys a large-scale model for pressure test data analysis; The pressure test result output module is used to output the pressure data test report obtained by the pressure test bench interlocking system through analysis of the pressure data.

7. The intelligent flat valve with integrated built-in pressure detection rod according to claim 6, characterized in that, The training steps for the large-scale model for stress test data analysis in the upload module include: Collect historical pressure test data, including the first pressure-time curve, pressure test parameters, valve specification parameters, and pressure test result labels; Historical pressure test data is preprocessed, including data cleaning, normalization, and feature extraction. The extracted features include pressure rise rate, pressure peak, pressure stability, pressure fluctuation amplitude, and leakage rate. The preprocessed data is divided into training set, validation set and test set; Construct a deep learning network model, which includes an input layer, a multi-layer convolutional neural network layer, a long short-term memory network layer, and a fully connected output layer; The deep learning network model is trained using a training set, and the model parameters are optimized using the backpropagation algorithm, enabling the model to identify qualified and abnormal stress test modes. Use the validation set to validate the trained model and fine-tune the parameters; The model performance is evaluated using a test set. When the model accuracy reaches a preset threshold, the model is deployed as a large-scale model for stress test data analysis.

8. The intelligent flat valve with integrated built-in pressure detection rod according to claim 7, characterized in that, The analysis process of pressure data by the large-scale pressure test data analysis model includes: The system receives pressure data transmitted from the upload module in real time and generates a second pressure-time curve for the current pressure test. The second pressure-time curve is input into the large-scale pressure test data analysis model, and the model automatically extracts the curve features and compares them with the standard pressure test mode. Identify abnormal events during the pressure test process, including excessively rapid pressure rise, pressure peak exceeding limits, abnormal fluctuations during the pressure holding phase, and suspected leaks. Generate a pressure data test report, which includes the test pass / fail criteria, abnormal event annotations, key parameter values, and improvement suggestions.

9. The intelligent flat valve with integrated built-in pressure detection rod according to claim 8, characterized in that, The large-scale model for stress test data analysis also has self-learning capabilities, including: Continuously collect new pressure test data and its manually labeled actual results; Based on preset time intervals, the large-scale model for stress test data analysis is incrementally trained using new data to update the model parameters; Establish a model version management mechanism to retain historical model versions and record model update logs; The system compares the performance of the new and old models and automatically switches deployments when the new model outperforms the old model.

10. The intelligent flat valve with integrated built-in pressure detection rod according to claim 6, characterized in that, The wired or wireless methods in the upload module include: It uses RS485, Modbus or Ethernet protocols and is connected to the interlocking system of the pressure test bench via physical cables; or, Wireless data transmission is achieved using LoRa, Zigbee, WiFi, or 4G / 5G protocols.

Images