Turbine valve jamming early warning system and method based on force-position cooperative monitoring

Abstract

The invention belongs to the technical field of steam turbine state monitoring and fault diagnosis, and particularly relates to a steam turbine valve jamming early warning system and method based on force-position cooperative monitoring, and the system comprises a sensor array, a data acquisition and conditioning unit, an edge calculation unit, a cloud early warning platform and a client interaction unit. By synchronously collecting force parameter signals such as valve displacement, oil pressure and steam pressure, characteristic quantity is extracted through collaborative analysis, and jamming fault identification is achieved in combination with a reference model and a dynamic threshold value. According to the method, a health reference curve library is established, force-position characteristics are analyzed in real time on the edge side, motion resistance is estimated, local early warning is carried out, historical data are deeply mined at the cloud end, and fault types and trend prediction are intelligently identified. According to the method, the jamming nature is directly struck, early-stage accurate early warning is achieved, the false and missing report rate is reduced, the real-time performance and diagnosis depth are guaranteed through edge cloud cooperation, and technical support is provided for predictive maintenance of the valve.

Description

Technical Field

[0001] This invention relates to the field of turbine condition monitoring and fault diagnosis technology, and in particular to a turbine valve jamming early warning system and method based on force-position coordinated monitoring. Background Technology

[0002] As a core power equipment in thermal power generation, nuclear power, and other fields, steam turbines use key valves such as main steam valves and regulating valves as core actuators to control steam flow and regulate unit speed and load. These valves operate under harsh conditions of high temperature (up to 600℃) and high pressure (supercritical and ultra-supercritical units) for extended periods. Components such as valve stems, valve cores, and valve sleeves are prone to increased friction due to oxide scale buildup, creep deformation, and impurity intrusion, which can lead to jamming. Mild jamming can affect the sensitivity and accuracy of valve regulation, causing the unit to be unable to accurately respond to grid load dispatch commands and reducing operational economy. Severe jamming may cause the valve to delay or completely fail to close when emergency closure is required, leading to catastrophic accidents such as turbine overspeeding and seriously threatening the safety of the unit and the power grid.

[0003] Currently, monitoring the status of steam turbine valves relies heavily on the experience and judgment of operators, periodic activity tests, and simple feedback signals indicating valve position. These traditional methods have significant shortcomings: 1) Continuous, online, real-time monitoring is not possible; 2) It is difficult to quantify the degree of stickiness, the judgment is highly subjective, and misjudgment and omission are easy to occur; 3) Lack of early warning capabilities, the fault is usually only discovered when it has developed to an obvious stage (such as valve operation being significantly slow), thus missing the best maintenance opportunity.

[0004] With the advancement of sensing and data analysis technologies, it has become possible to acquire valve operation data by installing displacement and pressure sensors. However, most existing technologies analyze displacement (position) or pressure signals in isolation, failing to fully explore the rich fault information contained in the dynamic coupling relationship between force and displacement. The essence of valve jamming is an abnormal increase in motion resistance (friction, unbalanced steam force, etc.). This abnormality will inevitably be reflected in the synergistic relationship between the output force (oil pressure) of its drive system (such as hydraulic actuator) and the actual displacement response of the valve. Therefore, there is an urgent need for a system and method that can collaboratively collect, fuse, and analyze the force-position parameters during valve movement and establish an intelligent early warning model to achieve accurate identification and early warning of early jamming faults in steam turbine valves. Summary of the Invention

[0005] Based on the technical problems existing in the prior art, this invention proposes a steam turbine valve jamming early warning system and method based on force-position collaborative monitoring.

[0006] The present invention proposes a turbine valve jamming early warning system based on force-position collaborative monitoring, comprising: a sensor array, a data acquisition and conditioning unit, an edge computing unit, a cloud early warning platform, and a client interaction unit.

[0007] The sensor array is used to acquire multidimensional physical signals reflecting the status of turbine valves in real time, including at least: Displacement sensing module: used to measure the linear displacement and rate of change of valve stem or hydraulic actuator piston rod, preferably using high-precision grating ruler, draw-wire displacement sensor or laser displacement sensor.

[0008] Force-related sensing module: used to measure parameters related to valve motion resistance, including: oil pressure in the high-pressure and low-pressure chambers of the hydraulic actuator, control oil pressure driving the servo valve, and steam pressure upstream and downstream of the valve (used to estimate steam force).

[0009] The data acquisition and conditioning unit is connected to the sensor array and is used to filter, amplify, and convert the analog signals output by the sensors to digital, and to synchronously acquire and package them according to a unified timestamp. Its technical requirements include: a sampling frequency of not less than 100Hz, a number of channels of not less than 12, and an Ethernet communication interface.

[0010] The edge computing unit, deployed at the unit site, communicates with the data acquisition and conditioning unit and has a built-in force-position collaborative analysis and early diagnosis model. Its core functions include: Real-time collaborative analysis: Receives synchronized force-position data streams and calculates the force-position characteristic curves of valve movement online (such as the oil pressure-displacement curve during valve opening / closing).

[0011] Feature extraction: Based on the force-position characteristic curve, extract the feature quantities used to characterize the jamming, such as: the starting friction force threshold, the average / peak friction force during the motion process, the motion hysteresis, the oil pressure value corresponding to a specific displacement point, etc.

[0012] Model calculation: Run the built-in valve motion resistance observation model and hydraulic system pressure-displacement transfer function model, and combine real-time operating data (such as main steam temperature and pressure) to dynamically estimate the theoretical motion resistance of the valve and compare it with the actual measured value.

[0013] Local early warning: Based on the results of feature extraction and model calculation, a preliminary judgment on the degree of blockage is made. When the feature quantity exceeds the set dynamic threshold, an early warning signal is generated and alerted through the local display (such as an OLED screen) or sound and light alarm of the edge computing unit.

[0014] The cloud-based early warning platform connects to the edge computing unit via a factory network or security isolation device, receiving raw data, feature data, and early warning events uploaded by the unit. The platform's core functions include: Big data storage and management: Long-term storage of all valves' complete historical data to establish digital valve archives.

[0015] Advanced diagnostics and health assessment: Run more sophisticated valve jamming fault models and machine learning algorithms to perform deep data analysis, fault mode identification, and health status trend prediction. For example, train support vector machines (SVM) or deep learning models based on historical data to intelligently classify jamming types (such as oxide scale jamming and deformation jamming).

[0016] Early warning management and traceability: Manage early warning rules, generate early warning reports that include fault location, severity, possible causes and handling suggestions, and provide functions such as historical data playback, Bode plot analysis, and multi-valve performance comparison for fault traceability and status assessment.

[0017] The client interaction unit is a web application or mobile app, through which authorized users can access the cloud-based early warning platform to view monitoring data, early warning information, historical curves and diagnostic reports in real time, and remotely configure system parameters.

[0018] On the other hand, the present invention provides a turbine valve jamming early warning method based on force-position collaborative monitoring applied to the above-mentioned system, comprising the following steps: S1: Install sensor arrays on key valves of the steam turbine (high / medium pressure main steam valves, regulating valves), and complete system integration and commissioning; S2: When the system is powered on, the sensor array synchronously and continuously acquires the displacement signals and related force parameter signals of the valve during static, dynamic tests and normal adjustment processes; S3: The data acquisition and conditioning unit synchronously acquires and preprocesses the raw signal to form a time-aligned force-position coordination data packet; S4: The edge computing unit receives data packets in real time, plots the real-time force-position characteristic curve of the current action process, and compares it with the reference force-position characteristic curve of the valve in a healthy state stored locally. S5: The edge computing unit extracts multiple preset feature quantities from the real-time characteristic curve and calls the built-in valve motion resistance observation model. Combined with the current unit load and steam parameters, it calculates the difference between the theoretical resistance and the actual resistance of the current action. S6: Based on the change amplitude of characteristic quantities and the difference between theoretical and actual resistance, combined with preset multi-level dynamic thresholds (early warning, alarm, danger), a comprehensive judgment of the degree of jamming is made; S7: If the identification result is normal, only the data is stored; if the warning level is triggered, the edge computing unit immediately generates a local warning and compresses and uploads the key data of this event (time stamp, feature quantity, curve segment) to the cloud warning platform. S8: After receiving event data, the cloud-based early warning platform performs in-depth analysis based on the valve's long-term health record data, verifies the early warning event, updates the valve's health index, and generates a detailed diagnostic report which is then pushed to the client interaction unit of the relevant maintenance personnel. S9: Based on the early warning information and diagnostic reports, maintenance personnel develop targeted inspection or maintenance plans, such as performing valve activity tests with specific parameters or arranging shutdown for maintenance.

[0019] Compared with the prior art, the present invention provides a turbine valve jamming early warning system and method based on force-position coordinated monitoring, which has the following beneficial effects: 1. By conducting collaborative monitoring and fusion analysis of force and position parameters, the essential characteristics of valve jamming faults (abnormal motion resistance) are grasped. Compared with single parameter monitoring, the information dimensions are richer and the diagnostic basis is more sufficient.

[0020] 2. By establishing a benchmark model and dynamic threshold, it is possible to sensitively capture subtle deterioration trends in valve motion characteristics, enabling early warning. The comprehensive criterion combining model calculation and feature analysis effectively reduces the false alarm rate and the missed alarm rate.

[0021] 3. Adopting an edge + cloud collaborative computing architecture, the edge side enables rapid response and preliminary diagnosis to ensure real-time performance; the cloud side performs big data mining and intelligent learning to continuously improve the accuracy of the diagnostic model and support health trend prediction.

[0022] 4. The system hardware design takes into account the industrial environment and meets the requirements of miniaturization, robustness, durability, dust and water resistance. The software has comprehensive functions, providing not only real-time early warning, but also powerful historical data analysis and fault tracing capabilities, providing complete technical support for predictive maintenance of valves. Attached Figure Description

[0023] Figure 1 This is a diagram illustrating the overall architecture of a steam turbine valve jamming early warning system based on force-position collaborative monitoring, as proposed in this invention. Figure 2 This is a flowchart of the real-time diagnostic process for the edge computing unit of this invention. Figure 3 This is a schematic diagram of the installation of the hydraulic actuator sensor for the high-pressure regulating valve of the present invention.

[0024] In the picture: 10. Sensor array; 11. Displacement sensing module; 12. Hydraulic pressure sensing submodule; 13. Control hydraulic pressure sensing submodule; 14. Steam pressure sensing submodule; 20. Data acquisition and conditioning unit; 21. Multi-channel data acquisition card; 22. Signal conditioning module; 23. Main control and communication module; 30. Edge computing unit; 31. Data interface module; 32. Benchmark model storage module; 33. Real-time analysis engine; 34. Local diagnosis and early warning module; 40. Cloud-based early warning platform; 41. Data access and storage services; 42. Valve health management services; 43. Advanced diagnostic model services; 44. Early warning and reporting services; 50. Client-side interaction unit; 70. Hydraulic actuator; 71. Piston rod; 72. High-pressure chamber (oil inlet chamber) of hydraulic actuator; 73. Low-pressure chamber (oil return chamber) of hydraulic actuator; 74. Servo valve. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not 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.

[0027] Example 1, Reference Figure 1 This embodiment provides a turbine valve jamming early warning system based on force-position collaborative monitoring, which specifically includes the following components: Sensor array 10: This is the data source of the system; the high-pressure regulating valve (GV) of a supercritical 660MW steam turbine will be used as an example for explanation. Figure 3 .

[0028] Displacement sensing module 11: It adopts a high-precision magnetostrictive linear displacement sensor. The sensor body is fixed on the cylinder of the hydraulic motor 70 by a bracket. Its measuring rod is rigidly connected to the piston rod 71 of the hydraulic motor 70 through a special connector. The sensor has a range of 0-300mm, a linear accuracy of ±0.05% FS, and outputs a 4-20mA analog signal for real-time measurement of the valve's precise opening degree (0-100%) and movement speed. The sensor has an IP67 protection rating, which meets the requirements of high temperature and oily environment on site.

[0029] Force-related sensing module: Hydraulic pressure sensing submodule 12: includes two high-performance piezoresistive pressure transmitters, one installed at the pressure measuring connector of the high-pressure chamber (oil inlet chamber, 72) of the hydraulic actuator, and the other installed at the pressure measuring connector of the low-pressure chamber (oil return chamber, 73) of the hydraulic actuator. The measuring ranges are 0-20MPa and 0-5MPa, respectively, with an accuracy of 0.1 class and an output of 4-20mA signal, which is used to directly obtain the core force source driving the valve - the oil pressure difference.

[0030] Control oil pressure sensing submodule 13: A small pressure sensor with a range of 0-10MPa is installed at the control port of the servo valve 74 driving the hydraulic motor 70 to monitor the control command pressure.

[0031] Steam pressure sensing submodule 14: Utilizes the existing steam pressure measuring points of the unit to read the pressure values ​​upstream (main steam) and downstream (after the regulating stage) of the valve through an analog acquisition card. This is used to calculate the force exerted by the steam on the valve core (this force is an important component of the valve's movement resistance). If there are no existing measuring points, pressure transmitters can be installed on the pipelines before and after the valve.

[0032] Data Acquisition and Conditioning Unit 20: Utilizes an integrated industrial data acquisition box with dimensions of 380mm × 580mm × 280mm and a total weight of approximately 12kg, meeting portability and field installation requirements. The box integrates: Multi-channel data acquisition card 21: It adopts a 16-channel synchronous acquisition card, supports analog voltage / current input, provides independent AD conversion channels for all sensor signals, ensures strict synchronization of all signal sampling, and sets the sampling frequency to 200Hz, which is much higher than the valve action frequency (usually less than 1Hz), satisfies the Nyquist sampling theorem and leaves sufficient margin.

[0033] Signal conditioning module 22: Filters (low-pass filter, cutoff frequency 50Hz) and isolates the input 4-20mA current signal to eliminate common electromagnetic interference in industrial settings.

[0034] Main control and communication module 23: It adopts an industrial-grade embedded processor, which is responsible for adding high-precision time stamps to the collected data and packaging the data into data frames of a custom protocol. This module is equipped with dual network ports, one for connecting sensors and edge computing units (through a switch), and the other for connecting to the factory information network. The power module supports dual input of AC 220V and DC 24V to ensure power supply reliability.

[0035] Edge computing unit 30: It adopts a high-performance industrial edge computing gateway, which is installed in the control cabinet of the steam turbine platform. Its hardware configuration includes: a multi-core ARM processor, 192KB RAM for real-time data caching and algorithm operation, 1MB Flash for storing diagnostic algorithm library and benchmark model parameters, 4GB eMMC for storing short-term historical data. The gateway is equipped with an OLED display for local status display and has a digital output interface for connecting to an audible and visual alarm. Its main software function modules include.

[0036] Data interface module 31: Receives synchronization data packets from the data acquisition box at 10ms intervals via Ethernet UDP protocol.

[0037] Reference model storage module 32: Stores a set of oil pressure-displacement reference curves (covering different oil temperatures and different steam parameters) obtained from multiple activity tests of the valve under initial health conditions, as well as the calculated reference values ​​of characteristic quantities (such as starting oil pressure and average friction oil pressure) and their normal fluctuation range.

[0038] Real-time analytics engine 33: This is the core of edge computing, and its workflow is as follows: Figure 2 As shown: Whenever a valve is detected to start moving (displacement change rate is greater than the threshold), the engine starts an analysis cycle. It first plots the real-time oil pressure-displacement curve (PD curve) for this movement, and then extracts the feature quantities F1 (oil pressure difference at the starting point ΔP_start), F2 (average oil pressure during the movement segment P_avg), and F3 (hysteresis area Area_hys of the opening and closing curves). At the same time, it calls the built-in "valve motion resistance observation model", which is a simplified physical model: R_estimated = f(P_high, P_low, A_piston, P_steam_up, P_steam_down, Valve_Position), where A_piston is the effective area of ​​the piston. This function estimates the total resistance (including friction, steam imbalance force, inertial force, etc.) on the valve stem at the current moment based on the real-time collected oil pressure, steam pressure, and valve position.

[0039] Local diagnostic and early warning module 34: Compares the extracted real-time feature quantities (F1, F2, F3) with their corresponding dynamic thresholds (with minor corrections based on the baseline value and current oil temperature). Simultaneously, it calculates ΔR = R_estimated - R_smooth, where R_smooth is the sliding average of resistance under recent healthy conditions. It formulates comprehensive early warning rules: If F1 exceeds the limit or ΔR continues to increase positively beyond the threshold T1, a "caution" level warning (yellow) is triggered; if F1 and F2 exceed the limit simultaneously, or ΔR exceeds a larger threshold T2, an "abnormal" level warning (orange) is triggered; if F1 and F2 severely exceed the limit and F3 significantly increases, a "danger" level warning (red) is triggered, and an on-site audible and visual alarm is immediately driven through digital output. All early warning events and corresponding data segments are recorded and prepared for uploading.

[0040] Cloud-based early warning platform 40: Deployed in the power plant's private cloud or group-level data center, it adopts a microservice architecture, and its main services include: Data access and storage services 41: Receive data uploaded from multiple units and edge gateways via the MQTT protocol, use time-series databases (such as InfluxDB) to store high-frequency raw and feature data, and use relational databases to store structured data such as events, reports, and models.

[0041] Valve Health Management Service 42: Establish a digital twin file for each valve. The service uses long-term stored data to calculate the valve's "health index" regularly (e.g., daily). This index is a value from 0 to 100, derived from multiple dimensions: recent warning frequency, characteristic trend slope, deviation from the average performance of similar valves, etc. The health index is visualized to intuitively reflect the trend of valve condition degradation.

[0042] Advanced Diagnostic Model Service 43: Run a machine learning-based fault classification model. For example, use force-position curve data (including normal, oxide rust, slight deformation, etc.) accumulated over the past three years to train a one-dimensional convolutional neural network (1D-CNN). When an early warning event is uploaded from the edge, in addition to rule verification, the platform will also call the CNN model to intelligently analyze the uploaded curve segment and output the probability distribution of the fault type, providing maintenance personnel with more accurate fault cause inferences.

[0043] Early Warning and Reporting Service 44: Manages plant-wide early warning events, automatically generates diagnostic reports, and pushes messages to clients via RESTful API. It also supports email and SMS (via integrated SMS gateway) alerts.

[0044] Client-side interaction unit 50: Develop a B / S architecture web application that allows maintenance engineers and inspectors to log in via a browser from anywhere with internet access. The main interface includes: Real-time monitoring dashboard: Graphically displays the real-time opening degree, health index, and latest warning status of all monitored valves, organized by unit.

[0045] In-depth analysis interface: Allows users to query historical time periods and view multiple curves such as valve displacement, oil pressure, and calculated resistance simultaneously for overlay and comparative analysis. It also provides a "Bohr plot" drawing function to analyze the amplitude-frequency and phase-frequency characteristics of the valve under different command frequencies, which is used to evaluate the dynamic performance of the valve servo system.

[0046] Report Center: View historical warning events and their corresponding detailed diagnostic reports. The report content includes: event time, valve information, triggering characteristic quantity and threshold, fault type prediction by intelligent model analysis, statistics of similar historical events, and handling suggestions (such as "It is recommended to check the valve stem oxide scale condition during the next shutdown").

[0047] Example 2, combining a turbine valve jamming early warning system based on force-position collaborative monitoring, details the implementation steps of the early warning method, referring to... Figure 1 and Figure 2 : S1: Installation and calibration, during planned downtime maintenance, according to... Figure 3 The proposed scheme involves installing a displacement sensor and an oil pressure sensor on the selected high-pressure regulating valve hydraulic actuator. After installation, multiple full-stroke mobility tests are conducted on the valve before unit startup, with the steam pressure at zero. The system automatically records the oil pressure-displacement curves from these tests. After confirmation by engineers, the average value is set as the initial cold-state reference curve and characteristic quantity reference value for the valve. After the unit is connected to the grid and under load, mobility tests are performed again at multiple stable load points (such as 50%, 75%, and 100% rated load). The system records and establishes a cluster of hot-state reference curves under different steam conditions. This step is the basis for subsequent accurate diagnosis.

[0048] S2: Continuous online monitoring. During normal unit operation, the system continuously collects data. For regulating valves, they are frequently adjusted slightly with load changes; for main steam valves, they are usually fully open, but they will activate during unit start-up, shutdown, or activity tests. The system captures and analyzes all these action events.

[0049] S3-S6: Real-time edge-side diagnostics (core process). Taking the normal opening action of a high-pressure regulating valve as an example, the edge gateway detects an increase in displacement, initiates analysis, and compares the real-time plotted PD curve with the hot-state reference curve stored in the gateway under similar steam parameters. Assuming that the extracted starting oil pressure F1 is 8% higher than the reference value, while the threshold T1_note is 10% and T1_alert is 20%, and at the same time, the real-time resistance R_estimated calculated by the resistance observation model is 12% higher than its recent sliding average R_smooth, and the threshold ΔR_note is 15%, according to the rules, F1 and ΔR are not exceeded, but they are close to the attention threshold. At this time, the edge gateway does not trigger an alarm, but records the event of a slight increase in the characteristic quantity in the local log and uploads the brief characteristic data of this action (time, load, F1, F2, ΔR) to the cloud.

[0050] S7-S8: Cloud-based in-depth analysis and trend warning. The cloud platform received characteristic data of the valve's multiple operations over the past week. Valve health management service 42 analysis revealed that the 7-day moving average of the valve's "starting oil pressure F1" showed a slow but continuous upward trend (e.g., from 100% to 108% of the baseline). At the same time, advanced diagnostic model service 43 analyzed the recently uploaded curve segments and gave an inference of "probability of oxide scale accumulation: 65%". Although a single event did not trigger an edge warning, the cloud automatically generated an "early deterioration trend warning" based on the comprehensive judgment of trend analysis and intelligent model, lowering the health index from 95 to 82, and pushed this information to the client of the relevant engineer.

[0051] S9: Operation and Maintenance Decision and Verification. After seeing the warning on the client side, the engineer reviewed the detailed historical curves and diagnostic reports of the valve. The report suggested "when convenient, perform a complete activity test from the current opening to full closure and then full opening to obtain more comprehensive evaluation data." The engineer arranged to perform the test during a peak shaving and load reduction process. The system captured the data from this complete test, and the analysis showed that the jamming characteristics during the closing process were more obvious. This further verified the accuracy of the cloud warning. Based on this, the engineer added the valve to the key inspection list in the subsequent planned shutdown. Finally, it was found that there was local oxide scale buildup on the valve stem. After cleaning, the valve was restored. After cleaning, the system recalibrated the benchmark, and the health index rebounded.

[0052] Through the detailed description of the two embodiments above, the system constructs a closed loop from data perception to intelligent decision-making through hardware collaboration, software intelligence, and edge-cloud integration, truly realizing predictive maintenance of turbine valve jamming faults.

[0053] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A turbine valve jamming early warning system and method based on force-position co-monitoring, characterized in that, include: A sensor array is used to synchronously acquire displacement signals of turbine valves and at least one force parameter signal directly related to their motion resistance. The data acquisition and conditioning unit, connected to the sensor array, is used to synchronously sample, condition, and convert the acquired analog signals into digital signals. The edge computing unit is communicatively connected to the data acquisition and conditioning unit, and is used to receive synchronous data in real time, perform force-position collaborative analysis, extract fault feature quantities, run valve resistance observation models, and perform localized primary diagnosis and early warning of jamming faults. The cloud-based early warning platform is network-connected to the edge computing unit and is used to receive and store data, run advanced diagnostic and health management models, and perform in-depth fault analysis, trend prediction, and global early warning management. The client interaction unit is used to authorize users to access the cloud-based early warning platform and realize human-computer interaction.

2. The turbine valve jamming early warning system based on force-position coordinated monitoring according to claim 1, characterized in that, The sensor array includes: Displacement sensing module, used to measure the linear displacement of valve stem or hydraulic actuator piston rod; The force-related sensing module includes at least one of the following: an oil pressure sensor for measuring the oil pressure in the working chamber of a hydraulic actuator, a control oil pressure sensor for measuring the control oil pressure of a servo valve, and a steam pressure sensing submodule for measuring or acquiring the steam pressure before and after the valve.

3. The turbine valve jamming early warning system based on force-position coordinated monitoring according to claim 1 or 2, characterized in that, The data acquisition and conditioning unit has a sampling frequency of no less than 100Hz, has no fewer than 12 synchronous acquisition channels, and supports Ethernet communication.

4. The turbine valve jamming early warning system based on force-position coordinated monitoring according to claim 1, characterized in that, The edge computing unit has the following built-in features: The baseline model storage module is used to store the baseline force-position characteristic curve and characteristic quantities of the monitored valve in a healthy state. The real-time analysis engine generates a real-time force-position characteristic curve for each valve action, extracts multiple preset feature quantities, and calls the valve motion resistance observation model to estimate the current motion resistance. The local early warning module is used to compare the extracted real-time feature quantities with dynamic thresholds and combine them with the estimated change in resistance to generate local early warning signals of different levels based on preset comprehensive criteria.

5. The turbine valve jamming early warning system based on force-position coordinated monitoring according to claim 4, characterized in that, The extracted preset features include at least one of the following: the driving pressure difference at the moment of valve activation, the average driving pressure during the movement process, and the hysteresis area of ​​the force-position curve during the opening and closing processes.

6. The turbine valve jamming early warning system based on force-position coordinated monitoring according to claim 1, characterized in that, The cloud-based early warning platform includes: Data storage service is used for long-term storage of valve historical operating data, characteristic data, and event records; Health management services are used to calculate and track valve health indices based on historical data and assess trends of condition degradation. Advanced diagnostic model service, used to run machine learning-based fault classification models to intelligently identify fault types and infer causes of early warning events.

7. A method for early warning of valve jamming in steam turbines based on force-position co-monitoring, applied to the system described in any one of claims 1-6, characterized in that, Includes the following steps: S1: The displacement signal and related force parameter signal during valve operation are synchronously acquired through a sensor array; S2: Perform synchronous preprocessing on the acquired signals to form a time-aligned force-potential collaborative dataset; S3: On the edge side, for a single action of the valve, plot the real-time force-position characteristic curve, extract the characteristic quantities, and run the resistance observation model to calculate the real-time resistance; S4: Compare the extracted feature quantities and / or calculated real-time resistance changes with preset benchmarks and dynamic thresholds to comprehensively determine whether the valve currently has a jamming fault and its severity level. S5: If a fault risk is detected, a warning message of the corresponding level is generated and a local notification is displayed on the edge side, while the relevant data is uploaded to the cloud. S6: In the cloud, the system performs in-depth analysis and verification of early warning events by combining the valve's historical health records, updates the valve's health status, and generates diagnostic reports that are pushed to the client.

8. The turbine valve jamming early warning method based on force-position co-monitoring according to claim 7, characterized in that, Before step S1, there is also a system initialization step S0: under the valve's healthy state, force-position data under different working conditions are collected through multiple activity tests to establish the valve's reference force-position characteristic curve library and characteristic quantity reference value range.

9. The turbine valve jamming early warning method based on force-position coordinated monitoring according to claim 7, characterized in that, The comprehensive judgment mentioned in step S4 specifically refers to: A multi-level threshold criterion is adopted, including attention level, abnormal level and danger level; The judgment logic integrates the deviation of real-time feature quantities from the benchmark, the deviation of estimated resistance from recent historical averages, and the combined relationships between multiple feature quantities.

10. The turbine valve jamming early warning method based on force-position coordinated monitoring according to claim 7, characterized in that, The in-depth analysis and verification described in step S6 includes: Fault mode identification based on machine learning model is performed on the uploaded force-potential curve segments; Analyze the time series trends of characteristic quantities in all recent action events of this valve; The performance of this valve is compared horizontally with the average performance of similar valves in the same type of unit.

Images