Intelligent valve management and control platform based on Internet of Things

Abstract

The invention provides an intelligent valve management and control platform based on the Internet of Things, and relates to the technical field of industrial automation control, and the intelligent valve management and control platform based on the Internet of Things comprises a data collection module, an edge calculation module, an Internet of Things communication module, a cloud platform management module, an intelligent diagnosis module and a self-adaptive control module. Through a mode of combining multi-source sensor data acquisition, edge calculation and cloud intelligent analysis, real-time monitoring of a valve state, accurate prediction of a fault and self-adaptive optimization of a control strategy are realized, compared with a traditional valve control system, the problems of response lag, high maintenance cost, data isolation and the like are effectively solved, and the system is suitable for large-scale popularization and application. The method can be widely applied to multiple industrial fields such as petroleum, chemical engineering and water affairs, the automation level, production efficiency and safety of industrial production are remarkably improved, the operation cost of enterprises is reduced, and the method has wide application prospects and huge economic value.

Description

Technical Field

[0001] The present invention relates to the field of industrial automation control technology, and in particular to an intelligent valve management and control platform based on the Internet of Things. Background Art

[0002] In today's era of highly automated industrial production, valves, as key components in industrial piping systems, are crucial for ensuring smooth production processes, improving production efficiency, and ensuring production safety. However, traditional valve control systems have exposed many problems that need to be solved urgently.

[0003] First, traditional valve control systems rely heavily on manual inspections to monitor valve status. Manual inspections are not only inefficient, but also often fail to detect subtle abnormal changes in valve operation due to subjective factors of inspectors and limitations of inspection cycles.

[0004] Secondly, there is a serious data island problem in the traditional system. The data generated by valve operation, such as pressure, temperature, flow, etc., cannot be effectively linked with the entire production system. This makes it impossible for production managers to comprehensively analyze the production status from a global perspective and it is difficult to optimize the control strategy of the production system based on the valve operation data.

[0005] Furthermore, the maintenance cost of traditional valve control systems remains high. The system lacks predictive maintenance capabilities and cannot accurately predict possible valve failures in advance. It can only perform passive repairs after a failure occurs. This post-failure repair method not only increases maintenance costs, but sudden failures often cause production equipment to shut down, causing additional downtime losses to the company.

[0006] In addition, traditional valve control systems generally have the problem of lack of remote control. With the expansion of enterprise scale and the decentralization of production layout, enterprises often need to centrally control valves across regions, but traditional systems cannot meet this demand. Although some valve control systems use single sensor monitoring or simple remote control technology, these systems lack multi-source data fusion, intelligent decision-making and adaptive optimization capabilities, and cannot manage valves comprehensively and accurately.

[0007] In summary, traditional valve control systems have been unable to adapt to the requirements of modern industrial production for efficiency, intelligence and safety.

[0008] Therefore, there is an urgent need for a valve intelligent management and control platform that integrates advanced technologies such as the Internet of Things, edge computing, and artificial intelligence to solve problems such as delayed response, high maintenance costs, data isolation, and insufficient remote control in traditional systems, and to improve the automation level and management efficiency of industrial production. Summary of the Invention

[0009] In order to overcome the existing problems, the embodiment of the present application provides a valve intelligent management and control platform based on the Internet of Things. Through the combination of multi-source sensor data acquisition, edge computing and cloud-based intelligent analysis, it realizes real-time monitoring of valve status, accurate prediction of faults and adaptive optimization of control strategies. Compared with traditional valve control systems, it effectively solves problems such as response lag, high maintenance cost, and data isolation. It can be widely used in many industrial fields such as petroleum, chemical industry, water supply, etc., significantly improves the automation level, production efficiency and safety of industrial production, reduces enterprise operating costs, and has broad application prospects and huge economic value.

[0010] The technical solution adopted by the embodiment of the present application to solve the technical problem is: A valve intelligent management and control platform based on the Internet of Things, comprising a data acquisition module, an edge computing module, an Internet of Things communication module, a cloud platform management module, an intelligent diagnosis module, and an adaptive control module; The data acquisition module collects valve operation data in real time through multiple types of sensors, including pressure sensors, temperature sensors, flow sensors and vibration sensors; Among them, sensor selection and deployment: Pressure sensor: Use a high-precision, high-pressure piezoresistive pressure sensor. Select a product with the appropriate range and accuracy based on the valve's operating pressure range and accuracy requirements. For valves operating between 0 and 10 MPa, choose a pressure sensor with a range of 0 to 15 MPa and an accuracy of 0.1% FS. Install the pressure sensor on the valve's inlet and outlet pipes to ensure accurate measurement of pressure changes upstream and downstream of the valve.

[0011] Temperature sensor: Use either a thermocouple or RTD temperature sensor, and select the type based on the temperature range and response speed requirements of the valve environment. In high-temperature environments, thermocouple temperature sensors are preferred; in medium- and low-temperature environments with high accuracy requirements, RTD temperature sensors are preferred.

[0012] Flow sensor: Select an appropriate flow sensor, such as an electromagnetic flowmeter, vortex flowmeter, or orifice flowmeter, based on the properties of the fluid in the pipeline (e.g., liquid, gas), flow range, and installation space. For conductive liquids, an electromagnetic flowmeter is recommended; for gases or low-viscosity liquids, a vortex flowmeter is a more suitable option. Install the flow sensor on a straight pipe section downstream of the valve to ensure accurate measurement.

[0013] Vibration Sensor: A piezoelectric vibration sensor is used to sensitively detect vibration signals during valve operation. Select a vibration sensor with appropriate sensitivity and frequency response range based on the valve size and vibration frequency range. Install the vibration sensor at key locations on the valve, such as the valve body and stem, to obtain key information such as valve vibration amplitude and frequency.

[0014] Data collection process: Each type of sensor collects valve operation data in real time according to the set sampling frequency. The sampling frequency can be adjusted according to actual needs. Generally, the sampling frequency of pressure and temperature sensors is set to 1-10Hz, the sampling frequency of flow sensors is 0.1-1Hz, and the sampling frequency of vibration sensors is higher and can be set to 100-1000Hz.

[0015] The collected data is preliminarily processed by the sensor's built-in signal conditioning circuit, such as amplification and filtering, to remove noise interference and improve signal quality.

[0016] The conditioned analog signal is converted into a digital signal through an analog-to-digital conversion circuit (ADC) for subsequent processing and transmission.

[0017] The digital signal is transmitted to the edge computing module via the data bus for further processing; The edge computing module is deployed at the valve terminal to implement data preprocessing, anomaly detection and localized control. The edge computing module performs preprocessing operations such as denoising and filling missing values ​​on the raw data obtained by the data acquisition module. Among them, the hardware platform is built: Choose a high-performance, low-power embedded processor as the core of the edge computing module, such as an ARM architecture-based processor, which has powerful computing capabilities and rich interface resources to meet data processing and communication needs.

[0018] Equipped with an appropriate amount of memory (such as 2GB-4GB of DDR4 memory) and storage devices (such as 16GB-64GB of eMMC flash memory) to store program code, intermediate data, configuration files, etc.

[0019] It integrates multiple communication interfaces, including SPI, I2C, UART, Ethernet interface, etc., to facilitate data interaction with various sensors, IoT communication modules and other devices.

[0020] Data preprocessing: Data cleaning: Check the validity of the collected sensor data and remove data points with obvious errors or anomalies.

[0021] Data normalization: Normalize data with different dimensions and value ranges collected by different types of sensors to make them comparable.

[0022] Feature extraction: Based on the characteristics of valve operating data, key features that reflect the valve's operating status are extracted. For example, for vibration data, features such as vibration amplitude, frequency, and phase are extracted; for pressure data, features such as pressure change rate and pressure fluctuation amplitude are extracted. These features serve as an important basis for subsequent anomaly detection and localized control.

[0023] Anomaly Detection: Threshold-based detection: Set the normal threshold range for various sensor data based on the valve's historical operating data and design parameters.

[0024] Machine learning-based detection: Machine learning algorithms, such as support vector machines (SVMs) and Gaussian mixture models (GMMs), are used to model and train data after normalization and feature extraction. The trained model learns the distribution characteristics of data under normal operating conditions. When new data significantly differs from the normal distribution predicted by the model, it is identified as an anomaly. For example, when using GMM to model valve vibration data, if a new vibration data point falls outside the probability density function predicted by the model, the valve vibration is considered abnormal.

[0025] Localization Control: When an abnormal situation is detected, the edge computing module takes corresponding control measures locally according to the preset control strategy.

[0026] For some common operating status adjustment requirements, such as adjusting the valve opening according to the flow set value, the edge computing module can calculate the appropriate valve opening locally in real time based on a pre-set control algorithm (such as the PID control algorithm) and send control instructions to the valve actuator to achieve automatic adjustment of the valve opening; The IoT communication module uses LoRa, NB-IoT or 5G protocols to achieve remote data transmission and command issuance; Among them, communication module integration and configuration: Based on the selected communication protocol, select the corresponding communication module, such as LoRa module (such as SX1278), NB-IoT module (such as BC95-B8), or 5G communication module (such as Quectel RM500Q-GN8).

[0027] Configure the communication module, set network parameters (such as LoRa frequency band, spreading factor, NB-IoT access network parameters, 5G APN, etc.), device identification (IMEI, etc.), and data transmission format.

[0028] The IoT communication module monitors the control instructions issued by the cloud platform in real time. When receiving the instructions, it parses them and forwards them to the edge computing module. The edge computing module performs corresponding operations based on the instructions, such as adjusting the valve opening and querying the valve status. The cloud platform management module includes data storage, analysis engine, visualization interface and API interface, supporting multi-user access. The data storage module of the cloud platform management module stores the collected valve operation data for a long time, and the analysis engine uses big data analysis and machine learning algorithms to conduct in-depth data mining. Among them, the cloud platform architecture is built: Choose cloud computing resources provided by mature cloud service providers (such as Alibaba Cloud and Tencent Cloud) to build the cloud platform infrastructure. This includes computing resources (such as virtual machine instances), storage resources (such as object storage (OSS) and cloud databases), and network resources (such as virtual private clouds (VPCs).

[0029] Visual interface: Interface design and development: We utilize modern front-end development technologies (e.g., HTML5, CSS3, and JavaScript) combined with visualization libraries (e.g., Echarts and D3.js) to develop an intuitive and user-friendly visualization interface. This interface design adheres to the principles of simplicity and clarity, using charts and graphs to display valve operating status, health assessment results, fault warning information, and relevant production system indicators.

[0030] Real-time data display: The key operating parameters of the valve (such as pressure, temperature, flow, valve opening, etc.) are displayed in real time on the visual interface. By dynamically updating the data, users can understand the working status of the valve in real time. At the same time, different colors and icons are used to intuitively display whether the valve operating status is normal, such as green for normal and red for abnormal. The intelligent diagnosis module analyzes historical data based on machine learning models to predict valve failures and remaining life. The machine learning models used by the intelligent diagnosis module include long short-term memory networks (LSTMs) and random forest models. Among them, data preparation: The historical operation data of the valve is extracted from the database of the cloud platform, including sensor data such as pressure, temperature, flow, vibration, and valve operation records (such as opening and closing time, opening adjustment records, etc.).

[0031] The preprocessed data is divided into training set, validation set and test set according to the time series.

[0032] Model selection and training: LSTM Model: Build an LSTM-based time series prediction model for predicting valve failures and remaining life. The LSTM model consists of an input layer, multiple LSTM unit layers, and an output layer. The input layer encodes preprocessed time series data. The LSTM unit layer uses memory cells and a gating mechanism to learn long-term dependencies in the time series. The output layer makes predictions based on the output of the LSTM unit layer. The LSTM model is trained using training data, and the backpropagation algorithm adjusts the model's weights and biases to minimize the loss function (such as the mean squared error) between the predicted and actual values.

[0033] Model evaluation and optimization: The trained LSTM and random forest models are evaluated using test data, using metrics such as accuracy, recall, and root mean square error (RMSE) to assess model performance. For example, for fault prediction tasks, accuracy represents the proportion of correctly predicted fault samples to the total number of fault samples; recall represents the proportion of correctly predicted fault samples to the total number of actual fault samples; and RMSE measures the degree of error between the predicted and actual values.

[0034] Optimize the model based on the evaluation results; The adaptive control module combines process parameters and environmental data to dynamically adjust valve opening to optimize energy efficiency; Among them, data acquisition and fusion: The operating data of the valve body (such as pressure, temperature, flow, etc.) is obtained from the data acquisition module, the environmental data (such as temperature, humidity, air pressure, etc.) is obtained from the environmental sensor, and the process parameters (such as pipeline pressure setting value, flow setting value, production task requirements, etc.) are obtained from the production system.

[0035] The acquired multi-source data is fused and processed, and different types of data are integrated into a comprehensive data set through operations such as data cleaning, feature extraction and data association, providing comprehensive information support for subsequent control decisions.

[0036] Control strategy development: Rule-based control: Develop a series of rule-based control strategies based on the valve's working principle, production process requirements, and historical operating experience.

[0037] Model-based control: Establish a mathematical model of the valve, combine the production process and environmental factors, and calculate the optimal valve opening through methods such as model predictive control (MPC).

[0038] Control execution and feedback adjustment: The calculated valve opening control instruction is sent to the edge computing module through the Internet of Things communication module, and the edge computing module then forwards the instruction to the valve actuator to adjust the valve opening.

[0039] Real-time monitoring of operating data after valve opening adjustment and feedback from the production system, such as changes in pipeline pressure and flow, allows for adjustment and optimization of control strategies based on this feedback.

[0040] It also includes multi-source data fusion function, integrating valve body data, environmental data and production system data to generate multi-dimensional health assessment indicators.

[0041] Preferably, the IoT-based valve intelligent management and control platform also includes a digital twin modeling function to build a real-time mapping between the valve physical entity and the virtual model, and support simulation testing and optimization strategy verification.

[0042] Preferably, the adaptive control module dynamically adjusts the valve opening according to the following formula: =

[0043] in, is the valve opening, for The process pressure parameters at the moment, for The process temperature parameters at the moment, for Traffic demand parameters at the moment, for Environmental data parameters at the moment, is a nonlinear mapping function trained based on historical data.

[0044] Preferably, the nonlinear mapping function f is obtained by training through the following formula:

[0045] in, For the The actual valve opening in the group history data, For the The process pressure parameters, process temperature parameters, flow demand parameters and environmental data parameters in the group historical data, is the number of historical data samples.

[0046] Preferably, the intelligent diagnostic module predicts the remaining life of the valve The formula is: =

[0047] in, For valves The operating state characteristic vector at the moment, including pressure, temperature, and vibration parameters; is the characteristic matrix of historical fault data; is the prediction function obtained based on LSTM model training.

[0048] The advantages of the embodiments of the present application are: The present invention combines multi-source sensor data acquisition, edge computing and cloud-based intelligent analysis to achieve real-time monitoring of valve status, accurate prediction of faults and adaptive optimization of control strategies. Compared with traditional valve control systems, it effectively solves problems such as response lag, high maintenance cost and data isolation. It can be widely used in many industrial fields such as petroleum, chemical industry, water supply, etc., significantly improving the automation level, production efficiency and safety of industrial production, reducing enterprise operating costs, and has broad application prospects and huge economic value. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a schematic diagram of the system process framework of the valve intelligent management and control platform based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, for the convenience of description below, the "upper", "lower", "left" and "right" quoted are the same as the upper, lower, left and right directions of the drawings themselves. The "first", "second" and so on in the following text are distinguished for the description and have no other special meanings.

[0051] The embodiments of the present application solve the problems in the prior art by providing a valve intelligent management and control platform based on the Internet of Things. By combining multi-source sensor data acquisition, edge computing and cloud-based intelligent analysis, it realizes real-time monitoring of valve status, accurate prediction of faults and adaptive optimization of control strategies. Compared with traditional valve control systems, it effectively solves problems such as response lag, high maintenance cost, and data isolation. It can be widely used in many industrial fields such as petroleum, chemical industry, water supply, etc., significantly improve the automation level, production efficiency and safety of industrial production, reduce enterprise operating costs, and has broad application prospects and huge economic value.

[0052] The technical solution in the embodiments of the present application is to solve the above problems, and the overall idea is as follows: Example This embodiment provides a valve intelligent management and control platform based on the Internet of Things, such as Figure 1 As shown in the figure, the valve intelligent management and control platform based on the Internet of Things includes a data acquisition module, an edge computing module, an Internet of Things communication module, a cloud platform management module, an intelligent diagnosis module, and an adaptive control module; The data acquisition module collects valve operation data in real time through multiple types of sensors, including pressure sensors, temperature sensors, flow sensors and vibration sensors; Among them, sensor selection and deployment: Pressure sensor: Use a high-precision, high-pressure piezoresistive pressure sensor. Select a product with the appropriate range and accuracy based on the valve's operating pressure range and accuracy requirements. For valves operating between 0 and 10 MPa, choose a pressure sensor with a range of 0 to 15 MPa and an accuracy of 0.1% FS. Install the pressure sensor on the valve's inlet and outlet pipes to ensure accurate measurement of pressure changes upstream and downstream of the valve.

[0053] Temperature sensor: Use either a thermocouple or RTD temperature sensor, and select the type based on the temperature range and response speed requirements of the valve environment. In high-temperature environments, thermocouple temperature sensors are preferred; in medium- and low-temperature environments with high accuracy requirements, RTD temperature sensors are preferred.

[0054] Flow sensor: Select an appropriate flow sensor, such as an electromagnetic flowmeter, vortex flowmeter, or orifice flowmeter, based on the properties of the fluid in the pipeline (e.g., liquid, gas), flow range, and installation space. For conductive liquids, an electromagnetic flowmeter is recommended; for gases or low-viscosity liquids, a vortex flowmeter is a more suitable option. Install the flow sensor on a straight pipe section downstream of the valve to ensure accurate measurement.

[0055] Vibration Sensor: A piezoelectric vibration sensor is used to sensitively detect vibration signals during valve operation. Select a vibration sensor with appropriate sensitivity and frequency response range based on the valve size and vibration frequency range. Install the vibration sensor at key locations on the valve, such as the valve body and stem, to obtain key information such as valve vibration amplitude and frequency.

[0056] Data collection process: Each type of sensor collects valve operation data in real time according to the set sampling frequency. The sampling frequency can be adjusted according to actual needs. Generally, the sampling frequency of pressure and temperature sensors is set to 1-10Hz, the sampling frequency of flow sensors is 0.1-1Hz, and the sampling frequency of vibration sensors is higher and can be set to 100-1000Hz.

[0057] The collected data is preliminarily processed by the sensor's built-in signal conditioning circuit, such as amplification and filtering, to remove noise interference and improve signal quality.

[0058] The conditioned analog signal is converted into a digital signal through an analog-to-digital conversion circuit (ADC) for subsequent processing and transmission.

[0059] The digital signal is transmitted to the edge computing module via the data bus for further processing; The edge computing module is deployed at the valve terminal to implement data preprocessing, anomaly detection, and localized control. The edge computing module performs preprocessing operations such as denoising and filling missing values ​​on the raw data obtained by the data acquisition module. Among them, the hardware platform is built: Choose a high-performance, low-power embedded processor as the core of the edge computing module, such as an ARM architecture-based processor, which has powerful computing capabilities and rich interface resources to meet data processing and communication needs.

[0060] Equipped with an appropriate amount of memory (such as 2GB-4GB of DDR4 memory) and storage devices (such as 16GB-64GB of eMMC flash memory) to store program code, intermediate data, configuration files, etc.

[0061] It integrates multiple communication interfaces, including SPI, I2C, UART, Ethernet interface, etc., to facilitate data interaction with various sensors, IoT communication modules and other devices.

[0062] Data preprocessing: Data cleaning: Check the validity of the collected sensor data and remove data points with obvious errors or anomalies.

[0063] Data normalization: Normalize data with different dimensions and value ranges collected by different types of sensors to make them comparable.

[0064] Feature extraction: Based on the characteristics of valve operating data, key features that reflect the valve's operating status are extracted. For example, for vibration data, features such as vibration amplitude, frequency, and phase are extracted; for pressure data, features such as pressure change rate and pressure fluctuation amplitude are extracted. These features serve as an important basis for subsequent anomaly detection and localized control.

[0065] Anomaly Detection: Threshold-based detection: Set the normal threshold range for various sensor data based on the valve's historical operating data and design parameters.

[0066] Machine learning-based detection: Machine learning algorithms, such as support vector machines (SVMs) and Gaussian mixture models (GMMs), are used to model and train data after normalization and feature extraction. The trained model learns the distribution characteristics of data under normal operating conditions. When new data significantly differs from the normal distribution predicted by the model, it is identified as an anomaly. For example, when using GMM to model valve vibration data, if a new vibration data point falls outside the probability density function predicted by the model, the valve vibration is considered abnormal.

[0067] Localization Control: When an abnormal situation is detected, the edge computing module takes corresponding control measures locally according to the preset control strategy.

[0068] For some common operating status adjustment requirements, such as adjusting the valve opening according to the flow set value, the edge computing module can calculate the appropriate valve opening locally in real time based on a pre-set control algorithm (such as the PID control algorithm) and send control instructions to the valve actuator to achieve automatic adjustment of the valve opening; The IoT communication module uses LoRa, NB-IoT or 5G protocols to achieve remote data transmission and command issuance; Among them, communication module integration and configuration: Based on the selected communication protocol, select the corresponding communication module, such as LoRa module (such as SX1278), NB-IoT module (such as BC95-B8), or 5G communication module (such as Quectel RM500Q-GN8).

[0069] Configure the communication module, set network parameters (such as LoRa frequency band, spreading factor, NB-IoT access network parameters, 5G APN, etc.), device identification (IMEI, etc.), and data transmission format.

[0070] The IoT communication module monitors the control instructions issued by the cloud platform in real time. When receiving the instructions, it parses them and forwards them to the edge computing module. The edge computing module performs corresponding operations based on the instructions, such as adjusting the valve opening and querying the valve status. The cloud platform management module includes data storage, analysis engine, visualization interface and API interface, supporting multi-user access. The data storage module of the cloud platform management module stores the collected valve operation data for a long time, and the analysis engine uses big data analysis and machine learning algorithms to conduct in-depth data mining. Among them, the cloud platform architecture is built: Choose cloud computing resources provided by mature cloud service providers (such as Alibaba Cloud and Tencent Cloud) to build the cloud platform infrastructure. This includes computing resources (such as virtual machine instances), storage resources (such as object storage (OSS) and cloud databases), and network resources (such as virtual private clouds (VPCs).

[0071] Visual interface: Interface design and development: We utilize modern front-end development technologies (e.g., HTML5, CSS3, and JavaScript) combined with visualization libraries (e.g., Echarts and D3.js) to develop an intuitive and user-friendly visualization interface. This interface design adheres to the principles of simplicity and clarity, using charts and graphs to display valve operating status, health assessment results, fault warning information, and relevant production system indicators.

[0072] Real-time data display: The key operating parameters of the valve (such as pressure, temperature, flow, valve opening, etc.) are displayed in real time on the visual interface. By dynamically updating the data, users can understand the working status of the valve in real time. At the same time, different colors and icons are used to intuitively display whether the valve operating status is normal, such as green for normal and red for abnormal. The intelligent diagnosis module analyzes historical data based on machine learning models to predict valve failures and remaining life. The machine learning models used by the intelligent diagnosis module include long short-term memory networks (LSTMs) and random forest models. Among them, data preparation: The historical operation data of the valve is extracted from the database of the cloud platform, including sensor data such as pressure, temperature, flow, vibration, and valve operation records (such as opening and closing time, opening adjustment records, etc.).

[0073] The preprocessed data is divided into training set, validation set and test set according to the time series.

[0074] Model selection and training: LSTM Model: Build an LSTM-based time series prediction model for predicting valve failures and remaining life. The LSTM model consists of an input layer, multiple LSTM unit layers, and an output layer. The input layer encodes preprocessed time series data. The LSTM unit layer uses memory cells and a gating mechanism to learn long-term dependencies in the time series. The output layer makes predictions based on the output of the LSTM unit layer. The LSTM model is trained using training data, and the backpropagation algorithm adjusts the model's weights and biases to minimize the loss function (such as the mean squared error) between the predicted and actual values.

[0075] Model evaluation and optimization: The trained LSTM and random forest models are evaluated using test data, using metrics such as accuracy, recall, and root mean square error (RMSE) to assess model performance. For example, for fault prediction tasks, accuracy represents the proportion of correctly predicted fault samples to the total number of fault samples; recall represents the proportion of correctly predicted fault samples to the total number of actual fault samples; and RMSE measures the degree of error between the predicted and actual values.

[0076] Optimize the model based on the evaluation results; The adaptive control module combines process parameters and environmental data to dynamically adjust valve opening to optimize energy efficiency; Among them, data acquisition and fusion: The operating data of the valve body (such as pressure, temperature, flow, etc.) is obtained from the data acquisition module, the environmental data (such as temperature, humidity, air pressure, etc.) is obtained from the environmental sensor, and the process parameters (such as pipeline pressure setting value, flow setting value, production task requirements, etc.) are obtained from the production system.

[0077] The acquired multi-source data is fused and processed, and different types of data are integrated into a comprehensive data set through operations such as data cleaning, feature extraction and data association, providing comprehensive information support for subsequent control decisions.

[0078] Control strategy development: Rule-based control: Develop a series of rule-based control strategies based on the valve's working principle, production process requirements, and historical operating experience.

[0079] Model-based control: Establish a mathematical model of the valve, combine the production process and environmental factors, and calculate the optimal valve opening through methods such as model predictive control (MPC).

[0080] Control execution and feedback adjustment: The calculated valve opening control instruction is sent to the edge computing module through the Internet of Things communication module, and the edge computing module then forwards the instruction to the valve actuator to adjust the valve opening.

[0081] Real-time monitoring of operating data after valve opening adjustment and feedback from the production system, such as changes in pipeline pressure and flow, allows for adjustment and optimization of control strategies based on this feedback.

[0082] It also includes multi-source data fusion function, integrating valve body data, environmental data and production system data to generate multi-dimensional health assessment indicators.

[0083] The IoT-based intelligent valve management and control platform also includes digital twin modeling capabilities, which build real-time mapping between the valve's physical entity and virtual model, supporting simulation testing and optimization strategy verification.

[0084] The adaptive control module dynamically adjusts the valve opening according to the following formula: =

[0085] in, is the valve opening, for The process pressure parameters at the moment, for The process temperature parameters at the moment, for Traffic demand parameters at the moment, for Environmental data parameters at the moment, is a nonlinear mapping function trained based on historical data.

[0086] The nonlinear mapping function f is trained by the following formula:

[0087] in, For the The actual valve opening in the group history data, For the The process pressure parameters, process temperature parameters, flow demand parameters and environmental data parameters in the group historical data, is the number of historical data samples.

[0088] Intelligent diagnostic module predicts remaining valve life The formula is: =

[0089] in, For valves The operating state characteristic vector at the moment, including pressure, temperature, and vibration parameters; is the characteristic matrix of historical fault data; is the prediction function obtained based on LSTM model training.

[0090] By adopting the above technical solutions: Scenario Description A large petrochemical enterprise has a complex pipeline network with numerous different types of valves controlling the flow of crude oil, chemical raw materials, and products. These valves operate in harsh environments characterized by high temperatures, high pressures, and flammable and explosive conditions, making accurate monitoring and intelligent control of their operating status crucial. This application of the platform is illustrated using a regulating valve on a crude oil pipeline connecting a refinery to a storage tank farm as an example.

[0091] System deployment and operation Data acquisition module: High-precision pressure sensors with a range of 0-20 MPa and an accuracy of 0.05% FS are installed on the inlet and outlet pipes of the regulating valve to monitor pressure changes in real time. A thermocouple temperature sensor with a measurement range of 0-600°C is installed on the valve body to monitor the valve's operating temperature. A vortex flowmeter is installed in the downstream pipeline to measure crude oil flow. Piezoelectric vibration sensors are also installed at key locations on the valve to monitor valve vibration. Each sensor collects data at a set frequency: the pressure and temperature sensors collect data once per second, the flow sensor collects data every 10 seconds, and the vibration sensor collects data every 100 milliseconds.

[0092] The collected data is transmitted to the edge computing module after signal conditioning and analog-to-digital conversion.

[0093] Edge computing module: It uses an embedded processor based on the ARM Cortex-A53 architecture, paired with 2GB DDR4 memory and 32GB eMMC flash memory, and integrates SPI, I2C, UART and Ethernet interfaces.

[0094] The collected data was cleaned to remove data points that fell outside the normal range by ±30%. Data of different dimensions were processed through linear normalization. Key features such as pressure change rate and vibration amplitude were extracted.

[0095] A threshold-based and SVM-based anomaly detection method is used. If the pressure exceeds the 15-18 MPa range or the SVM model determines that the vibration data is abnormal, local control measures are immediately taken, such as reducing the valve opening.

[0096] IoT Communication Module: Considering the large area of ​​the petrochemical plant and weak signal coverage in some areas, a LoRa communication module was selected, operating at 433 MHz with a spreading factor of 6. It was connected to the edge computing module via a UART interface, and network parameters and device identification were configured. The edge computing module encapsulated and sent pre-processed data to the cloud platform every five minutes, while also receiving commands from the cloud platform.

[0097] Cloud platform management module: The cloud platform is built on Alibaba Cloud, using a distributed architecture to deploy multiple ECS instances for load balancing. Structured data is stored in the RDS for MySQL database, and unstructured data is stored in OSS object storage.

[0098] We used association rules to analyze the relationships between pressure, temperature, and flow, and used TensorFlow to train LSTM and random forest models. The training set included valve operation data and fault records from the past two years.

[0099] Develop a visual interface to display valve pressure, temperature, flow, and opening parameters in real time, and use line graphs to show historical data trends. Provide an API interface to connect with enterprise MES and ERP systems to share valve operation and maintenance data.

[0100] Intelligent diagnostic module: The historical operation data of valves in the past three years were extracted and divided into training set, validation set and test set in the ratio of 70%, 15% and 15%.

[0101] Build an LSTM model with three layers of LSTM units, 128 neurons per layer, and train using the mean squared error loss function and the Adam optimizer. Build a random forest model with 500 decision trees and a maximum depth of 10.

[0102] Evaluations on the test set showed that the LSTM model achieved an 85% accuracy rate in predicting valve failures, while the random forest model achieved an 88% accuracy rate in classifying valve operating status. Real-time data is fed into the model, and if a valve internal leakage failure is predicted within 72 hours, an alert is sent to maintenance personnel.

[0103] Adaptive Control Module: The system obtains valve operation data, tank liquid level (production system data), and plant ambient temperature (environmental data). If the tank liquid level approaches the upper limit and the ambient temperature rises, causing the crude oil viscosity to decrease, the valve opening is reduced based on the rule-based control strategy.

[0104] A dynamic model of the valve-pipeline system was established, and the optimal valve opening was calculated using the MPC algorithm, which was adjusted every 10 minutes. The control strategy was optimized based on the pressure and flow feedback after the valve opening adjustment.

[0105] Finally, it should be noted that the above embodiments are merely examples for the purpose of illustrating the present invention and are not intended to limit the embodiments. Those skilled in the art will readily appreciate that other variations or modifications based on the above description are possible. It is not necessary and impossible to provide an exhaustive list of all embodiments. However, obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A valve intelligent management and control platform based on the Internet of Things, characterized by: The IoT-based valve intelligent management and control platform includes a data acquisition module, an edge computing module, an IoT communication module, a cloud platform management module, an intelligent diagnosis module, and an adaptive control module; The data acquisition module collects valve operation data in real time through multiple types of sensors; The edge computing module is deployed at the valve terminal to achieve data preprocessing, anomaly detection and localized control; The IoT communication module uses LoRa, NB-IoT or 5G protocols to achieve remote data transmission and command issuance; The cloud platform management module includes data storage, analysis engine, visualization interface and API interface, supporting multi-user access; The intelligent diagnosis module analyzes historical data based on machine learning models to predict valve failures and remaining life; The adaptive control module combines process parameters and environmental data to dynamically adjust valve opening to optimize energy efficiency.

2. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The multiple types of sensors include pressure sensors, temperature sensors, flow sensors and vibration sensors.

3. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The edge computing module performs preprocessing operations such as denoising and filling missing values ​​on the original data obtained by the data acquisition module.

4. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The data storage module of the cloud platform management module stores the collected valve operation data for a long time, and the analysis engine uses big data analysis and machine learning algorithms to deeply mine the data.

5. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The machine learning models used in the intelligent diagnosis module include long short-term memory network LSTM and random forest model.

6. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: It also includes multi-source data fusion function, integrating valve body data, environmental data and production system data to generate multi-dimensional health assessment indicators.

7. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The IoT-based valve intelligent management and control platform also includes a digital twin modeling function, which builds a real-time mapping between the valve physical entity and the virtual model, and supports simulation testing and optimization strategy verification.

8. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The adaptive control module dynamically adjusts the valve opening according to the following formula: = in, is the valve opening, for The process pressure parameters at the moment, for The process temperature parameters at the moment, for Traffic demand parameters at the moment, for Environmental data parameters at the moment, is a nonlinear mapping function trained based on historical data.

9. The valve intelligent management and control platform based on the Internet of Things according to claim 8 is characterized in that: The nonlinear mapping function f is obtained by training through the following formula: in, For the The actual valve opening in the group history data, For the The process pressure parameters, process temperature parameters, flow demand parameters and environmental data parameters in the group historical data, is the number of historical data samples.

10. The valve intelligent management and control platform based on the Internet of Things according to claim 1 is characterized in that: The intelligent diagnostic module predicts the remaining life of the valve The formula is: = in, For valves The operating state characteristic vector at the moment, including pressure, temperature, and vibration parameters; is the characteristic matrix of historical fault data; is the prediction function obtained based on LSTM model training.

Images