Digital twinning emergency cut-off control method and system based on SIS system

Abstract

The invention discloses a digital twinning emergency cut-off control method and system based on an SIS system, and the method comprises the following steps: S1, building a digital twinning model system in the SIS system containing an emergency cut-off device, and taking OPC UA as a unified information communication architecture, S2, collecting the operation state data of the emergency cut-off device of the SIS system in real time through an intelligent sensor, the method comprises the steps of S1, carrying out remote data analysis and management, feeding back data to a digital twin model system, and carrying out real-time visual monitoring on the state of an instrument, and S3, automatically judging the state of the system by an intelligent algorithm according to the data of each instrument of the SIS system, and outputting risk early warning. According to the invention, the operation state data of the emergency cut-off device of the SIS system is collected in real time, remote data analysis and management are carried out, abnormity can be found in time, a control signal can be given, operation of each instrument can be cut off in an emergency mode in a quick response mode, and remote monitoring and control of the SIS system can be realized through the remote data analysis and management function.

Description

Technical Field

[0001] The present invention belongs to the technical field of chemical equipment instrument control, and specifically relates to a digital twin emergency shutoff control method and system based on an SIS system. Background Art

[0002] The SIS (Safety Instrumented System) is part of the automation of an enterprise's production processes. It's a system used to ensure safe production, with a higher safety level than DCS automation control systems. When an automated production system experiences an anomaly, the SIS intervenes to reduce the likelihood of an accident.

[0003] In the safe production of chemical equipment and facilities, the SIS system serves as a dual safeguard for the safe production of chemical equipment and facilities (such as boilers, pressure vessels, and pressure piping). In the event of an emergency, the SIS system automatically shuts off the flow of media within the pipeline. This is because the media transported within the pipeline during chemical production are often flammable, explosive, toxic, or corrosive. In the event of an emergency such as a leak or overpressure, these media can spread rapidly, causing serious accidents such as fire, explosion, or poisoning. By automatically shutting off the flow of media, the SIS system effectively curbs the spread of danger and prevents further escalation of the accident. It also quickly isolates the source of danger, reducing personnel exposure to hazardous environments and thus protecting their lives. It also prevents further damage to equipment caused by leaks or overpressure, reducing the cost of equipment repair and replacement. In other situations, the SIS system remains in a "standby" state.

[0004] However, as the equipment continues to operate and ages, the emergency shutoff devices of SIS systems, exposed to complex and changing industrial environments, are inevitably subject to erosion from environmental factors (such as temperature fluctuations, humidity changes, UV radiation, and dust pollution) as well as the media they handle (such as highly corrosive, high-temperature, high-pressure, and flammable and explosive properties). These external factors can gradually adversely affect key performance indicators such as the mechanical structure, sealing performance, material strength, and stability of electronic components of the emergency shutoff devices, leading to degradation or changes in their performance over long-term production operations.

[0005] Specifically, environmental factors can accelerate the aging process of internal components, causing corrosion of metal parts, hardening and cracking of rubber seals, and parameter drift of electronic components, thereby affecting the accuracy and response speed of the cut-off action. The slower response speed of the cut-off action, and the erosion of the medium within it, can directly damage the surface coating or protective layer of the device, exacerbating material wear and corrosion, and even leading to medium leakage, posing a direct threat to system safety.

[0006] Performance changes can not only reduce the reliability and effectiveness of the SIS system in emergency situations, increasing the risk of misoperation or failure, but can also seriously impact the safe and stable operation of the entire chemical production process, including but not limited to production interruptions, frequent safety accidents, increased environmental pollution, and increased economic losses. Therefore, necessary upgrades and modifications to the SIS system's emergency shut-off devices are crucial to ensuring safe production in chemical equipment and facilities. Summary of the Invention

[0007] In view of the deficiencies of the prior art, the object of the present invention is to provide a digital twin emergency shutdown control method based on the SIS system.

[0008] To achieve the aforementioned object of the invention, the technical solutions adopted by the present invention include: A digital twin emergency shutdown control method based on a SIS system includes the following steps: S1. Establish a digital twin model system in the SIS system containing the emergency shut-off device, use OPC UA as the unified information communication architecture, and perform data mapping between the physical entity of the SIS system and its digital twin model. The digital twin element model of the physical element model includes: raw material barrel, pipeline, pump, flow meter, shut-off valve, regulating valve, reactor, stirring device, temperature sensor, pressure sensor, liquid level sensor; the digital twin behavior model includes: pump start and stop action, flow meter action, shut-off valve action, regulating valve action, stirring device (stirring motor, stirring paddle), and liquid level gauge pointer status; S2. Use smart sensors to collect real-time operating status data of the SIS system emergency shutdown device, conduct remote data analysis and management, and feed the data back to the digital twin model system for real-time visual monitoring of the instrument status. Among them, the shut-off time of the shut-off valve is collected and analyzed, and a control signal is given to quickly respond to the emergency shut-off of the operation of various instruments. Based on convolutional neural networks, the spatial characteristics of the operating data of the emergency shut-off device in the SIS system are extracted, the temporal dependency of the sensor data of the emergency shut-off device in the SIS system is captured, and the key information in the time series is extracted to improve the accuracy of the risk warning of the emergency shut-off device in the SIS system. Rapid response emergency shut-off of various instruments and meters include: When the medium is liquid, the centrifugal pump is started to feed the mixing tank, and the electromagnetic flowmeter controls the feeding amount. When the feeding amount reaches the set value, the system alarms and gives a control signal. The centrifugal pump stops running and the electric shut-off valve is cut off. When the medium is gas, the gas is supplied to the mixing tank through the gas cylinder. When the pressure in the tank reaches the set value, the system alarms, the pressure transmitter gives a control signal, and the electric shut-off valve is cut off; The raw materials in the mixing tank are heated by an electric heating rod. When the temperature of the raw materials reaches the set value, the system alarms, the temperature sensor gives a control signal, and the power supply of the heating rod is cut off; when an over-temperature or over-pressure alarm occurs, the power supply of the stirring motor is cut off and the operation stops.

[0009] When the liquid level reaches the high level, the system alarms and gives a control signal, the centrifugal pump stops running, and the electric shut-off valve is cut off; when the liquid level reaches the low level, the system alarms and gives a control signal, the power supply of the stirring motor is cut off and it stops running.

[0010] S3. The intelligent algorithm automatically determines the system status based on the data from various instruments in the SIS system, outputs risk warnings, and provides visual prompts in the digital twin model system; Among them, the collected operating status data is used to identify abnormal medium flow, determine whether there is a deviation fault in the medium transportation, and then issue an alarm.

[0011] In the present invention, by establishing a digital twin model system, data mapping is achieved between the physical entity of the SIS system and its digital twin model, so that the operating status of the SIS system can be visualized in real time; this helps operators to intuitively understand the system status and promptly discover and deal with potential problems.

[0012] In the present invention, intelligent sensors are used to collect the operating status data of the SIS system emergency shut-off device in real time, and remote data analysis and management are performed. This can detect abnormalities in a timely manner and give control signals, quickly responding to the emergency shut-off of the operation of various instruments and meters.

[0013] This invention uses a convolutional neural network to extract spatial features from operational data, capture temporal dependencies in sensor data, and extract key information from time series, significantly improving the accuracy of risk warnings. This helps prevent accidents before they occur and reduce losses.

[0014] This invention identifies media flow anomalies based on collected operational status data, determining whether media delivery failures are occurring and issuing timely alerts. This helps prevent safety incidents caused by media delivery anomalies. An intelligent algorithm automatically determines system status based on data from various SIS instruments and outputs risk warnings. This automated risk assessment mechanism reduces the need for human intervention and mitigates the risk of accidents caused by human error.

[0015] Furthermore, in step S2, a control signal is given, and a time setting method for a quick response includes: D1. The SIS system obtains the cut-off start time and stage end time, and calculates the cut-off process time = cut-off end time - cut-off start time. The SIS system controller is a PLC, which can accurately obtain the timestamps of the cut-off start time and cut-off end time to ensure the accuracy of time measurement. D2. Analyze the emergency disconnection time threshold. The disconnection time calculation method is: T=D×1s / in, Where D is the diameter of the ball valve (in inches), and T is the cut-off time (1s / in). Based on experience or experimental data, the larger the diameter of the ball valve, the longer the cut-off time. Considering the impact of the ball valve diameter on the cut-off time makes the analysis of the cut-off time threshold more scientific and reasonable. D3. The quick response delay of the output disconnection device is ≤ 10ms, ensuring that the disconnection device responds in a very short time and meets the real-time requirements of emergency disconnection.

[0016] Furthermore, step S2 also includes an evaluation model for the operating status of the emergency shutoff device. The evaluation method of the evaluation model is as follows: T1. Extract the trigger time allowable value, response time allowable value, and pressure change allowable value corresponding to the emergency shutoff device model, and compare them with the corresponding detected values ​​respectively. When the detected corresponding value is greater than the corresponding allowable value, calculate the difference between the corresponding allowable value and the corresponding value to obtain the abnormal difference value. Match the abnormal difference value of each corresponding value within the corresponding value range. Set each value range to correspond to a fault alarm value GZ1. After matching the abnormal difference values ​​of each corresponding value, obtain the fault alarm value GZ1 of each corresponding value within the current trigger time point. T2. When the detected corresponding value is less than the corresponding allowable value, the difference between the corresponding allowable value and the corresponding value is calculated to obtain the warning difference. The warning difference of each corresponding value is matched within the corresponding value range. Each value range is set to correspond to a fault pre-value GZ2. After matching the warning differences of each corresponding value, the fault pre-value GZ2 of each corresponding value within the current trigger time point is obtained. T3. Count the number of fault warning values ​​GZ1 at the current corresponding trigger time point to obtain the fault number GK; perform normalization processing between the fault warning value GZ1 or fault prediction value GZ2 of each corresponding value at the current trigger time point and the fault number GK to obtain the fault potential assessment index at the current corresponding trigger time point: Fk=GZi×al+GZi×a2+GZi×a3+(1+GK)×a4, The fault potential assessment index Fk corresponding to the current trigger time point is obtained by calculation, where i = 1 or 2, a1, a2, a3 and a4 are the influence weight factors of the fault prediction value and fault warning value corresponding to the trigger time, response time and pressure change value respectively; T4. Compare the fault hazard assessment index corresponding to the current trigger time point with the corresponding index threshold. Mark the fault hazard assessment index as Fk, set the first threshold as F1, the second threshold as F2, and F2 > F1. When Fk < F1, it indicates that the corresponding type of cut-off device is in normal use. When F1 < Fk < F2, it indicates that the corresponding type of cut-off device is in a pending state. When F2 < Fk, it indicates that the corresponding type of cut-off device is in an abnormal state.

[0017] The prior art only focuses on a single parameter or a few parameters, making it difficult to comprehensively reflect the operating state of the device. The evaluation model for the operating state of the emergency cut-off device of the product of the present invention comprehensively considers multiple key parameters such as the trigger time, response time, and pressure change of the emergency cut-off device, ensuring the comprehensiveness and accuracy of the evaluation. By extracting the allowable trigger time, allowable response time, and allowable pressure change corresponding to the device model and comparing them with the actual measured values, it is possible to accurately identify whether the device performance meets the standards. This comparison method is direct and effective, helping to timely discover potential problems.

[0018] Further, the specific method steps for extracting the spatial features of the operating data of the emergency cut-off device in the SIS system based on the convolutional neural network (CNN) include: F1. Through multi-layer convolution and pooling operations, automatically learn the fault patterns and fault data features in the data. F2. Incorporate the long short-term memory network (LSTM) to capture the time-dependent relationships of the sensor data of the emergency cut-off device in the SIS system, enabling the long short-term memory network to handle the long-term dependence problems in the sequence data and effectively extract the key information in the time series. F3. Combine the convolutional neural network and the long short-term memory network. Take the output of the convolutional neural network as the input of the long short-term memory network, construct a convolutional neural network-long short-term memory network hybrid model, and use this model to learn the spatial features of the data and capture the time-dependent relationships, thereby improving the accuracy of the risk warning of the emergency cut-off device in the SIS system.

[0019] This hybrid model, through the deep integration of the spatial feature extraction ability of CNN and the time-dependent modeling ability of LSTM, solves the limitations of traditional methods in terms of single feature extraction dimension, weak time correlation modeling ability, and insufficient warning real-time performance. In high-risk industries such as chemical engineering and energy, its technical advantages can be transformed into direct economic benefits such as reduced accident rates, optimized maintenance costs, and improved compliance, providing key technical support for the intelligent upgrade of the SIS system.

[0020] Further, the method for identifying abnormal medium flow includes: G1. Obtain the flow data of the medium during pipeline transportation; G2. Input the flow data into a pre-trained deviation fault identification model, reconstruct the flow data using the deviation fault identification model, and calculate the deviation between the flow data and the reconstructed data; G3. If the deviation value exceeds the preset deviation threshold, it is determined that a deviation fault occurs in the medium transportation.

[0021] This method, through a deviation reconstruction model and a dynamic threshold mechanism, addresses the pain points of traditional flow anomaly detection methods, such as low automation, insufficient sensitivity, and poor adaptability. It is particularly suitable for high-risk industries with extremely high requirements for real-time performance and accuracy. Its technical advantages can directly translate into reduced safety risks and optimized operation and maintenance costs, providing key support for the intelligent upgrade of chemical equipment processes.

[0022] Deviation fault identification model training includes the following steps: H 1. Data preparation and feature mapping (1) Data acquisition: Intelligent sensors are used to collect real-time operating status data of the SIS emergency shut-off device, including trigger time (e.g., valve action delay), response time (delay from alarm to shut-off), and pressure / flow change values ​​(dynamic parameters of pipeline media). These data are then mapped synchronously to the corresponding elements of the digital twin model (e.g., shut-off valves, flow meters, pressure sensors, etc.).

[0023] (2) Data labeling: Based on historical fault records, normal and abnormal states (such as valve sticking and sensor drift) are marked to form a labeled training set. Associated SIS fault diagnosis: This step directly corresponds to the "Media Flow Abnormality Identification" (steps G1-G3) in the manual, providing input data for subsequent models.

[0024] H 2. Model training and optimization (1) Loss function design: Weighted binary cross-entropy is used. According to the class imbalance characteristic of the SIS system (normal samples far outnumber fault samples), the weight of positive samples is set to the number of negative samples / the number of positive samples to improve the sensitivity to rare faults.

[0025] (2) Optimization strategy: Use the Nadam optimizer (learning rate lr = 1e-4, gradient clipping clipnorm = 1.0) to accelerate convergence and avoid gradient explosion. Use ReduceLROnPlateau to dynamically adjust the learning rate (attenuation factor = 0.5, patience value = 5) to adapt to the nonlinear characteristics of SIS system data.

[0026] (3) Training and Validation: 5-fold cross-validation: 100 epochs per fold, batch_size = 64, 20% validation set, and isolated test set to ensure generalization. Early stopping mechanism: Monitor the fault potential assessment index (val_fi) of the validation set. If there is no improvement after 15 epochs, terminate the training and restore the optimal weights. Associated SIS fault diagnosis: The trained model is used to calculate the fault potential assessment index Fk in real time (step T3) and judge the device status through the abnormal difference of parameters such as trigger time and response time (steps T1-T2).

[0027] H 3. Spatial-temporal feature extraction (1) CNN extracts spatial features: Through multi-layer convolution and pooling operations, it learns the local patterns of SIS system sensor data (such as the waveform characteristics of pressure mutations and the time series images of valve actions).

[0028] (2) LSTM captures temporal dependencies: Analyzes the long-term trends of sensor data (such as the slow drift of flow meter readings and the cumulative aging effect of temperature sensors) to solve long-term dependency problems.

[0029] (3) Hybrid model construction: The CNN output is used as the LSTM input to construct a CNN-LSTM hybrid model, which integrates spatial and temporal features and outputs the fault probability. Associated SIS fault diagnosis: This model directly supports the "deviation fault identification model" (step G2) in the manual. By reconstructing the deviation between the flow data and the measured value, an over-threshold alarm is triggered (step G3).

[0030] H 4. Real-time fault diagnosis and response (1) Fault warning generation: The model outputs the fault probability in real time, combined with the fault potential assessment index Fk (step T3). If Fk exceeds the threshold F2 (abnormal state), the visual alarm of the digital twin system is immediately triggered.

[0031] (2) Control command issuance: The warning signal is transmitted to the controller via the OPC UA protocol, and an emergency shutdown (e.g., valve closing, pump stopping) is executed, with a response delay of ≤10ms (step D3). Associated SIS fault diagnosis**: This step corresponds to the "rapid response emergency shutdown" (step S2) in the manual, achieving closed-loop control from diagnosis to action.

[0032] H 5. Continuous optimization and simulation verification (1) Interactive training unit update: The state data during the actual cutting process (such as the actual cutting time and pressure change) is fed back to the deep neural network unit to iteratively optimize the model parameters.

[0033] (2) Simulation platform verification: Fault scenarios (e.g., sensor failure, media leakage) are injected into the SIS system disconnect device simulation environment unit to verify the model robustness. Related SIS fault diagnosis: Supports the "dynamic adjustment of the evaluation model" (step T4) in the manual to ensure that the system continues to adapt as the equipment ages.

[0034] A digital twin emergency shutdown control system based on the SIS system, comprising: OPC UA communication protocol, based on the unified architecture of OPC UA communication protocol; A controller, which controls key components and collects physical sensor signals, feeds back the physical sensor signals to the OPC UA communication protocol, and shuts down various instruments of the device in an emergency; The host computer collects physical sensor signals and controls the host computer to send control commands to the controller to emergency shut down various instruments of the device; A control system simulator, which connects physical entities, digital models, and data by modeling the digital twin system. The control system simulator feeds simulated sensor information back to the OPC UA communication protocol. Database, used to store and be called by the controller for the operating status data of the SIS system emergency shut-off device; The SIS system disconnect device simulation environment unit provides a virtualized, secure, and controllable simulation platform for the design, testing, verification, and optimization of the SIS system. The SIS system disconnect device simulation environment unit communicates with the OPC UA communication protocol. A deep neural network unit that interacts in real time with the SIS system shutoff device simulation environment unit to extract spatial features of operating data of the emergency shutoff device in the SIS system; The interactive training unit feeds back status, strategy information, and fault identification information to the deep neural network unit; The OPC UA communication protocol seamlessly accesses the SIS system cutting device simulation environment unit through the database, performs data mapping between the SIS system physical entity and its digital twin model, and realizes remote online visual status detection and control; the simulation results of the SIS system cutting device simulation environment unit are converted into control strategies through the OPC UA communication protocol, and the control system simulator forms control instructions and feeds them back to the controller through the OPC UA communication protocol. The controller displays the information to the host computer and controls the host computer to issue control instructions.

[0035] Compared with the prior art, the advantages of the present invention include: (1) The present invention provides a digital twin emergency shut-off control method based on the SIS system. This technical solution integrates advanced technologies such as digital twin technology, intelligent sensor technology, and convolutional neural network into the SIS system, promoting technological innovation and upgrading. The digital twin model system has good scalability and adaptability and can be customized and optimized according to actual needs. This helps to meet the safety production needs of different chemical companies. (2) The present invention provides a digital twin emergency shut-off control method based on the SIS system, which covers key physical elements such as raw material barrels, pipelines, pumps, flow meters, shut-off valves and their behaviors (such as pump start and stop, flow meter action, etc.), ensuring that the virtual model is highly consistent with the actual system, and improving the comprehensiveness and accuracy of monitoring; (3) The present invention provides a digital twin emergency shutoff control method based on the SIS system. This method uses intelligent sensors to collect real-time operating status data of the SIS system's emergency shutoff device, performs remote data analysis and management, and can promptly detect anomalies and issue control signals, quickly responding to the emergency shutoff of various instruments and meters. Through remote data analysis and management functions, remote monitoring and control of the SIS system can be achieved. This improves the flexibility and efficiency of system management and reduces maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is a digital twin system flow chart of a digital twin emergency shutoff control method based on the SIS system in the present invention; Figure 2 This is a virtual-real interaction flow chart of a digital twin emergency shutdown control system based on the SIS system in the present invention; Figure 3 This is a schematic diagram of the overall architecture of a digital twin emergency shutdown control system based on the SIS system in the present invention. DETAILED DESCRIPTION

[0038] In view of the deficiencies in the prior art, the inventors of this case have proposed the technical solution of the present invention after long-term research and extensive practice. The following will further explain the technical solution, its implementation process and principles, etc. in conjunction with the drawings in the embodiments of this application and specific implementation cases.

[0039] It should be noted that the embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be understood as limiting the present invention. The embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, the present invention covers any substitution, modification, equivalent method and scheme made within the spirit, principle and scope of the present invention defined by the claims. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0040] In the description of this application, "first", "second", "third" and similar words do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, "a" or "an" and other similar words do not indicate a quantity limitation, but rather indicate the existence of at least one. "Include" or "comprising" and other similar words mean that the elements or objects appearing before "include" or "comprising" include the elements or objects listed after "include" or "comprising" and their equivalents, and do not exclude other elements or objects. "Connected" or "connected" and other similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0041] In the description of this application, the terms "center," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are intended only to facilitate the description of this application and simplify the description. They are not intended to indicate or imply that the devices or components referred to must have a specific direction, be constructed, or operate in a specific direction. Therefore, they should not be construed as limitations on this application. Furthermore, when positional terms such as "both sides," "outside," "upper," and "lower" are used, they should be understood to be used solely to facilitate understanding and description, taking into account that the structure may be oriented in other directions.

[0042] In the description of this application, unless otherwise clearly specified and limited, the technical or scientific terms used should have the usual meanings understood by persons with ordinary skills in the field to which this application belongs. Terms such as "install", "connect", and "connect" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, a conflicting connection, or an integrated connection. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0043] The embodiment of the present invention is intended to introduce and illustrate the structural composition of the digital twin emergency shut-off control method based on the SIS system and the coordination relationship between the various components. Unless otherwise specified, the dimensions, materials, and manufacturing processes of the various components in the digital twin emergency shut-off control method based on the SIS system in the embodiment of the present invention can be selected according to specific circumstances and are not specifically limited or explained here.

[0044] Furthermore, in order to provide the public with a better understanding of the present invention, some specific details are described in detail in the following detailed description of the present invention, but those skilled in the art can fully understand the present invention without the description of these details.

[0045] Example 1 See also Figure 1 、 Figure 2 and Figure 3 , a digital twin emergency shutoff control method based on SIS system, comprising the following steps: S1. Real-time visual monitoring: Establish a digital twin model system in the SIS system containing the emergency shut-off device, use OPC UA as a unified information communication architecture, and perform data mapping between the physical entity of the SIS system and its digital twin model. Comprehensive element mapping, in which the digital twin element model of the physical element model includes: raw material barrels, pipelines, pumps, flow meters, shut-off valves, regulating valves, reactors, stirring devices, temperature sensors, pressure sensors, and liquid level sensors; the digital twin behavior model includes: pump start and stop actions, flow meter actions, shut-off valve actions, regulating valve actions, stirring devices (stirring motors, stirring paddles), and liquid level gauge pointer status; covering key physical elements such as raw material barrels, pipelines, pumps, flow meters, shut-off valves and their behaviors (such as pump start and stop, flow meter actions, etc.), ensuring that the virtual model is highly consistent with the actual system, and improving the comprehensiveness and accuracy of monitoring.

[0046] S2. Improve risk warning and response speed. Use smart sensors to collect real-time operating status data of the SIS system emergency shutdown device, conduct remote data analysis and management, and feed the data back to the digital twin model system for real-time visual monitoring of the instrument status. This can detect anomalies in a timely manner and give control signals, quickly responding to emergency shutdowns of various instruments and meters. Among them, the method of collecting and analyzing the shut-off time of the shut-off valve, giving a control signal, and quickly responding to the emergency shut-off of the operation of each instrument and giving a control signal and the time setting method of the quick response include: D1. The SIS system obtains the cut-off start time and stage end time, and calculates the cut-off process time = cut-off end time - cut-off start time. The SIS system controller is a PLC, which can accurately obtain the timestamps of the cut-off start time and cut-off end time to ensure the accuracy of time measurement. D2. Analyze the emergency disconnection time threshold. The disconnection time calculation method is: T=D×1s / in, Where D is the diameter of the ball valve (in inches), and T is the cut-off time (1s / in). Based on experience or experimental data, the larger the diameter of the ball valve, the longer the cut-off time. Considering the impact of the ball valve diameter on the cut-off time makes the analysis of the cut-off time threshold more scientific and reasonable. D3. The quick response delay of the output disconnection device is ≤ 10ms, ensuring that the disconnection device responds in a very short time and meets the real-time requirements of emergency disconnection.

[0047] Ball valves of different diameters experience varying levels of fluid resistance and mechanical inertia during the closing process. Larger diameter valves require longer to overcome fluid inertia, while smaller diameter valves close faster. Dynamically adjusting the shutoff time threshold using a linear formula ensures efficient shutoff within a safe range for valves of varying diameters. This approach can directly reduce safety risks and improve system stability in high-risk industries such as chemical, oil and gas, and water treatment.

[0048] Based on convolutional neural networks, the spatial characteristics of the operating data of the emergency shut-off device in the SIS system are extracted, the temporal dependency of the sensor data of the emergency shut-off device in the SIS system is captured, and the key information in the time series is extracted to improve the accuracy of the risk warning of the emergency shut-off device in the SIS system. The specific method steps for extracting spatial features of emergency shut-off device operation data in SIS system based on convolutional neural network (CNN) include: F1. Through multi-layer convolution and pooling operations, it automatically learns the fault modes and fault data features in the data. It automatically learns local features such as high-frequency fluctuations and spikes in sensor data (such as pressure / temperature / flow curves) through multi-layer convolution kernels (such as 3×3 and 5×5), accurately identifying early characteristic signals of mechanical faults such as valve body sticking and seal leakage. The pooling layer (such as maximum pooling) downsamples the feature data to eliminate the influence of data acquisition noise (such as sensor electromagnetic interference), retaining key features while reducing the amount of calculation. F2. The integration of a long short-term memory (LSTM) network captures the temporal dependencies of sensor data from the emergency shutoff devices in the SIS system, enabling the LSTM to handle long-term dependencies in sequential data and effectively extract key information from time series. The LSTM unit controls information flow through input, forget, and output gates, addressing the vanishing gradient problem of traditional RNNs. This effectively addresses the chain reaction within 10 minutes of a sudden pressure change (e.g., pressure increase → shutoff valve actuation → sudden flow drop). The gating mechanism automatically filters redundant data (e.g., steady-state fluctuations) and focuses on abnormal events (e.g., the response time within 5 seconds after a pressure threshold is exceeded). Traditional time series models such as ARIMA (time series model) can only process linear trends, while LSTM can learn the nonlinear coupling relationship between valve actuation and upstream and downstream pressure changes.

[0049] F3. Combining the convolutional neural network and the long short-term memory network, the output of the convolutional neural network is used as the input of the long short-term memory network to construct a convolutional neural network-long short-term memory network hybrid model. This model is used to learn the spatial characteristics of the data and capture the temporal dependencies, thereby improving the accuracy of the risk warning of the emergency shutdown device in the SIS system.

[0050] The hybrid model can simultaneously process multi-source data such as pressure, temperature, and vibration, capture the coordinated changes of multiple parameters during valve operation (such as pressure increase accompanied by abnormal vibration), solve the false alarm problem of single parameter warning, and automatically adjust the weight of each sensor data through the attention mechanism (Attention), strengthening the influence of key parameters (such as the current of the shut-off valve actuator). The traditional threshold method requires setting the alarm threshold of each parameter separately, while the hybrid model can learn the complex coupling relationship between parameters.

[0051] It also includes an evaluation model for the operating status of the emergency shut-off device. This model comprehensively considers multiple key parameters such as the emergency shut-off device's trigger time, response time, and pressure changes to ensure comprehensive and accurate evaluation. Existing technologies may only focus on a single parameter or a few parameters, making it difficult to fully reflect the operating status of the device. The evaluation model's evaluation method is as follows: T1. Extract the trigger time allowable value, response time allowable value, and pressure change allowable value corresponding to the emergency shutoff device model and compare them with the corresponding detected values. When the detected corresponding value is greater than the corresponding allowable value, calculate the difference between the corresponding allowable value and the corresponding value to obtain the abnormal difference value. Match the abnormal difference value of each corresponding value within the corresponding value range. Set each value range to correspond to a fault alarm value GZ1. After matching the abnormal difference values ​​of each corresponding value, obtain the fault alarm value GZ1 for each corresponding value within the current trigger time point. By extracting the trigger time allowable value, response time allowable value, and pressure change allowable value corresponding to the device model and comparing them with the actual detected values, it is possible to accurately identify whether the device performance meets the standard. This comparison method is direct and effective, helping to promptly identify potential problems.

[0052] T2. When the corresponding value detected is less than the corresponding allowable value, the difference between the corresponding allowable value and the corresponding value is calculated to obtain the warning difference. The warning difference of each corresponding value is matched within the corresponding value range. Each value range is set to correspond to a fault warning value GZ2. After matching the warning difference of each corresponding value, the fault warning value GZ2 of each corresponding value within the current trigger time point is obtained. The model calculates the abnormal difference and warning difference based on the difference between the actual detection value and the allowable value, providing a quantitative basis for fault warning and alarm. The existing technology may lack such a clear difference calculation mechanism, resulting in inaccurate or in-time fault judgment. By setting different value ranges to correspond to different fault warning values ​​and fault warning values, the model can more finely divide the operating status of the device and provide guidance for subsequent maintenance and processing.

[0053] T3. Count the number of fault warning values ​​GZ1 at the current corresponding trigger time point to obtain the number of faults GK. Based on the fault warning values ​​GZ1 or fault prediction values ​​GZ2 and the number of faults GK of each corresponding value within the current trigger time point, perform normalization processing to obtain the fault hidden danger assessment index at the current corresponding trigger time point: Fk=GZi×al+GZi×a2+GZi×a3+(1+GK)×a4. The model normalizes the fault warning value, fault prediction value and number of faults to obtain the fault hidden danger assessment index. This processing method makes different parameters and indicators comparable, facilitating comprehensive evaluation. The fault potential assessment index Fk corresponding to the current trigger time point is obtained by calculation, where i=1 or 2, a1, a2, a3 and a4 are the influence weight factors of the fault prediction value and fault warning value corresponding to the trigger time, response time and pressure change value, respectively; by introducing the influence weight factors of the fault prediction value and fault warning value corresponding to the trigger time, response time and pressure change value, the model can more accurately reflect the influence of each parameter on the operating status of the device, thereby improving the accuracy of the assessment.

[0054] T4. Compare the potential hazard assessment index corresponding to the current trigger time point with the corresponding index threshold. Mark the potential hazard assessment index as Fk, set the first threshold as F1, and the second threshold as F2, where F2 > F1. When Fk < F1, it indicates that the corresponding type of cutoff device is in normal use. When F1 < Fk < F2, it indicates that the corresponding type of cutoff device is in a pending processing state. When F2 < Fk, it indicates that the corresponding type of cutoff device is in an abnormal state.

[0055] Quickly respond and urgently cut off all instruments and meters, for example, including: When the medium is liquid, start the centrifugal pump to feed the blending tank, and the electromagnetic flowmeter controls the feeding volume. When the feeding volume reaches the set value, the system alarms, gives a control signal, the centrifugal pump stops running, and the electric cut-off valve cuts off. When the medium is gas, supply gas to the blending tank through the gas cylinder. When the pressure in the tank reaches the set value, the system alarms, the pressure transmitter gives a control signal, and the electric cut-off valve cuts off. Heat the raw materials in the blending tank with an electric heating rod. When the raw material temperature reaches the set value, the system alarms, the temperature sensor gives a control signal, and the power supply of the heating rod is cut off. When over-temperature or over-pressure alarms occur, the power supply of the stirring motor is cut off and it stops running.

[0056] When the liquid level reaches the high liquid level, the system alarms, gives a control signal, the centrifugal pump stops running, and the electric cut-off valve cuts off. When the liquid level reaches the low liquid level, the system alarms, gives a control signal, the power supply of the stirring motor is cut off and it stops running.

[0057] Extract the spatial features of the operation data based on the convolutional neural network, capture the time-dependent relationship of the sensor data, and extract the key information in the time series, significantly improving the accuracy of risk early warning. This helps to take preventive measures before accidents occur and reduce losses.

[0058] S3. Improve the system safety and reliability. The intelligent algorithm automatically judges the system state based on the data of each instrument and meter in the SIS system, outputs a risk early warning, and at the same time gives a visual prompt in the digital twin model system. Among them, identify the abnormal medium flow in the collected operation status data, determine whether there is a deviation fault in the medium transportation, and then alarm. By identifying the abnormal medium flow in the collected operation status data, it is possible to determine whether there is a deviation fault in the medium transportation and alarm in time. This helps to prevent safety accidents caused by abnormal medium transportation.

[0059] This automated risk judgment mechanism reduces the need for human intervention and reduces the accident risk caused by human errors.

[0060] In step S2, it also includes an evaluation model for the operating state of the emergency cut-off device. The evaluation method of the evaluation model is as follows: T1. Extract the trigger time allowable value, response time allowable value, and pressure change allowable value corresponding to the model of the emergency cut-off device, and compare them with the detected corresponding values respectively. When the detected corresponding value is greater than the corresponding allowable value, calculate the difference between the corresponding allowable value and the corresponding value to obtain an abnormal difference. Match the abnormal differences of each corresponding value within the corresponding value range. Set each value range to correspond to a fault warning value GZ1 respectively. After matching the abnormal differences of each corresponding value, obtain the fault warning value GZ1 of each corresponding value at the current trigger time point; T2. When the detected corresponding value is less than the corresponding allowable value, calculate the difference between the corresponding allowable value and the corresponding value to obtain a warning difference. Match the warning differences of each corresponding value within the corresponding value range. Set each value range to correspond to a fault prediction value GZ2 respectively. After matching the warning differences of each corresponding value, obtain the fault prediction value GZ2 of each corresponding value at the current trigger time point; T3. Count the number of fault warning values GZ1 at the current corresponding trigger time point to obtain the number of faults GK; perform normalization processing among the fault warning value GZ1 or fault prediction value GZ2 of each corresponding value, the number of faults GK at the current trigger time point to obtain the fault hidden danger evaluation index at the current corresponding trigger time point: Fk = GZi×al + GZi×a2 + GZi×a3 + (1 + GK)×a4, Calculate the fault hidden danger evaluation index Fk at the current corresponding trigger time point, where i = l or 2, and al, a2, a3, and a4 are the influence weight factors of the fault prediction value and fault warning value corresponding to the trigger time, response time, and pressure change value respectively; T4. Compare the fault hidden danger evaluation index corresponding to the current trigger time point with the corresponding index threshold. Mark the fault hidden danger evaluation index as Fk, set the first threshold as F1; the second threshold as F2, and F2 > F1; when Fk < F1, it indicates that the corresponding model cut-off device is in normal use state. When F1 < Fk < F2, it indicates that the corresponding model cut-off device is in a pending processing state. When F2 < Fk, it indicates that the corresponding model cut-off device is in an abnormal state.

[0061] The operating state of the device is divided into three types: normal use state, pending processing state, and abnormal state. This division method is clear and easy for operators to understand and execute. By setting the first threshold and the second threshold, the model can accurately judge which state the device is in and provide clear guidance for subsequent maintenance and processing. The prior art may lack such a clear state division and threshold setting mechanism, resulting in difficulty for operators to judge the actual operating state of the device.

[0062] The steps of the estimation model of the present invention are clear and easy to understand and implement. The operator only needs to follow the steps to complete the evaluation of the operating status of the device.

[0063] Furthermore, the medium flow anomaly identification method includes: G1. Obtain the flow data of the medium during pipeline transportation; G2. Input the flow data into a pre-trained deviation fault identification model, reconstruct the flow data using the deviation fault identification model, and calculate the deviation between the flow data and the reconstructed data; G3. If the deviation value exceeds the preset deviation threshold, it is determined that a deviation fault occurs in the medium transportation.

[0064] Through unsupervised learning, a normal distribution model for flow data is constructed, encoding the input flow data into low-dimensional features before decoding and reconstructing them. If the flow data contains anomalies (such as leaks or blockages), the reconstruction error will increase significantly. The deviation threshold can be dynamically adjusted based on the reconstruction error distribution of historical data (e.g., the 3σ principle) to adapt to flow fluctuations under different operating conditions. The model directly processes raw flow data, avoiding the tedious manual extraction of statistical features such as mean and variance required in traditional methods. Through online learning or a sliding window mechanism, the deviation between the current flow and the reconstructed data can be calculated in real time, meeting the millisecond-level response requirements of industrial scenarios.

[0065] H 1. Data preparation and feature mapping (1) Data acquisition: Intelligent sensors are used to collect real-time operating status data of the SIS emergency shut-off device, including trigger time (e.g., valve action delay), response time (delay from alarm to shut-off), and pressure / flow change values ​​(dynamic parameters of pipeline media). These data are then mapped synchronously to the corresponding elements of the digital twin model (e.g., shut-off valves, flow meters, pressure sensors, etc.).

[0066] (2) Data labeling: Based on historical fault records, normal and abnormal states (such as valve sticking and sensor drift) are marked to form a labeled training set. Associated SIS fault diagnosis: This step directly corresponds to the "Media Flow Abnormality Identification" (steps G1-G3) in the manual, providing input data for subsequent models.

[0067] H 2. Model training and optimization (1) Loss function design: Weighted binary cross-entropy is used. According to the class imbalance characteristic of the SIS system (normal samples far outnumber fault samples), the weight of positive samples is set to the number of negative samples / the number of positive samples to improve the sensitivity to rare faults.

[0068] (2) Optimization strategy: Use the Nadam optimizer (learning rate lr = 1e-4, gradient clipping clipnorm = 1.0) to accelerate convergence and avoid gradient explosion. Use ReduceLROnPlateau to dynamically adjust the learning rate (attenuation factor = 0.5, patience value = 5) to adapt to the nonlinear characteristics of SIS system data.

[0069] (3) Training and Validation: 5-fold cross-validation: 100 epochs per fold, batch_size = 64, 20% validation set, and isolated test set to ensure generalization. Early stopping mechanism: Monitor the fault potential assessment index (val_fi) of the validation set. If there is no improvement after 15 epochs, terminate the training and restore the optimal weights. Associated SIS fault diagnosis: The trained model is used to calculate the fault potential assessment index Fk in real time (step T3) and judge the device status through the abnormal difference of parameters such as trigger time and response time (steps T1-T2).

[0070] H 3. Spatial-temporal feature extraction (1) CNN extracts spatial features: Through multi-layer convolution and pooling operations, it learns the local patterns of SIS system sensor data (such as the waveform characteristics of pressure mutations and the time series images of valve actions).

[0071] (2) LSTM captures temporal dependencies: Analyzes the long-term trends of sensor data (such as the slow drift of flow meter readings and the cumulative aging effect of temperature sensors) to solve long-term dependency problems.

[0072] (3) Hybrid model construction: The CNN output is used as the LSTM input to construct a CNN-LSTM hybrid model, which integrates spatial and temporal features and outputs the fault probability. Associated SIS fault diagnosis: This model directly supports the "deviation fault identification model" (step G2) in the manual. By reconstructing the deviation between the flow data and the measured value, an over-threshold alarm is triggered (step G3).

[0073] H 4. Real-time fault diagnosis and response (1) Fault warning generation: The model outputs the fault probability in real time, combined with the fault potential assessment index Fk (step T3). If Fk exceeds the threshold F2 (abnormal state), the visual alarm of the digital twin system is immediately triggered.

[0074] (2) Control command issuance: The warning signal is transmitted to the controller via the OPC UA protocol, and an emergency shutdown (e.g., valve closing, pump stopping) is executed, with a response delay of ≤10ms (step D3). Associated SIS fault diagnosis**: This step corresponds to the "rapid response emergency shutdown" (step S2) in the manual, achieving closed-loop control from diagnosis to action.

[0075] H 5. Continuous optimization and simulation verification (1) Interactive training unit update: The state data during the actual cutting process (such as the actual cutting time and pressure change) is fed back to the deep neural network unit to iteratively optimize the model parameters.

[0076] (2) Simulation platform verification: Fault scenarios (e.g., sensor failure, media leakage) are injected into the SIS system disconnect device simulation environment unit to verify the model robustness. Related SIS fault diagnosis: Supports the "dynamic adjustment of the evaluation model" (step T4) in the manual to ensure that the system continues to adapt as the equipment ages.

[0077] Features of the above solution: (1) Comprehensive fault diagnosis: covering multiple parameters such as trigger time, response time, pressure / flow, etc., to solve the limitation of single parameter monitoring of traditional SIS system.

[0078] (2) Real-time guarantee: The entire process delay from data collection to cutting action is ≤10ms, meeting the safety requirements of chemical production.

[0079] (3) Adaptability: Adapt to device performance degradation (such as response delay caused by valve wear) through digital twins and online learning.

[0080] A digital twin emergency shutdown control system based on the SIS system, comprising: OPC UA communication protocol, based on the unified architecture of OPC UA communication protocol, performs data mapping between the physical entities of the SIS system and its digital twin model; A controller, which controls key components and collects physical sensor signals, feeds back the physical sensor signals to the OPC UA communication protocol, and shuts down various instruments of the device in an emergency; The host computer collects physical sensor signals and controls the host computer to send control commands to the controller to emergency shut down various instruments of the device; A control system simulator, which models the digital twin system, connects the physical entity, digital model, and data, and feeds simulated sensor information back to the OPC UA communication protocol; Database, used to store, query and call the operating status data of the SIS system emergency shut-off device; The SIS system disconnect device simulation environment unit provides a virtualized, secure, and controllable simulation platform for the design, testing, verification, and optimization of the SIS system. The SIS system disconnect device simulation environment unit communicates with the OPC UA communication protocol. The deep neural network unit interacts in real time with the SIS system shutdown device simulation environment unit to extract the spatial characteristics of the SIS system's emergency shutdown device operating data. Using convolutional neural networks to extract spatial characteristics of operating data, the unit captures the temporal dependencies of sensor data and extracts key information from time series, significantly improving the accuracy of risk warnings. This helps prevent accidents before they occur and reduce losses.

[0081] The interactive training unit feeds back status, strategy information, and fault identification information to the deep neural network unit; The OPC UA communication protocol seamlessly accesses the SIS system disconnection device simulation environment unit through the database, performs data mapping between the SIS system physical entity and its digital twin model, and realizes remote online visual status detection and control; the simulation results of the SIS system disconnection device simulation environment unit are converted into control strategies through the OPC UA communication protocol and fed back to the control system simulator through the OPC UA communication protocol, and the control system simulator forms control instructions and feeds them back to the controller through the OPC UA communication protocol. The controller displays the information to the host computer and controls the host computer to issue control instructions.

[0082] Working principle: The flow sensor collects the flow data of the medium in the pipeline in real time, inputs it into the pre-trained deviation fault identification model, calculates the deviation between the flow data and the reconstructed data, and if the deviation exceeds the preset threshold, it is determined to be a flow abnormality (such as leakage or blockage), triggering a cut-off signal. Obtain the ball valve diameter D (inches) that triggers the shutoff, substitute it into the formula T = D × 1 s / in to calculate the shutoff time threshold T, start the timer, and trigger a timeout alarm if the valve is not completely closed within T seconds. Flow data → anomaly detection model → cut-off signal → dynamic threshold calculation → valve control → timeout alarm; The technical solution of integrating the flow monitoring module, anomaly recognition model, and shut-off control module into the SIS system and extracting the operating data features of the emergency shut-off device in the SIS system based on a hybrid model of convolutional neural network (CNN) and long short-term memory network (LSTM) can effectively solve the problems of insufficient spatial feature extraction and difficulty in capturing temporal dependencies of the operating data of the emergency shut-off device in the SIS system (safety instrument system), and significantly improve the accuracy of risk warnings.

[0083] In this way, by synergizing flow anomaly detection with dynamic shutoff time thresholds, the problems of disconnected detection and response, as well as poor threshold adaptability, in traditional solutions are resolved. This method significantly improves safety by rapidly shutting off leak sources and suppressing pressure surges; significantly reduces engineering costs through a fully automated process, minimizing manual intervention; improves economic efficiency by reducing accident losses and maintenance costs; and enhances versatility by adapting to multiple scenarios and valve types.

[0084] It should be understood that the above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with this technology to understand the content of the present invention and implement it accordingly. It cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, they can make some simple deductions or substitutions without departing from the concept of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A digital twin emergency shutdown control method based on SIS system, characterized in that: The following steps are involved: S1. Establish a digital twin model system in the SIS system containing the emergency shut-off device, use the OPC UA communication protocol as a unified information communication architecture, and perform data mapping between the physical entity of the SIS system and its digital twin model. The digital twin element model of the physical element model includes the raw material barrel, pipeline, pump, flow meter, shut-off valve, regulating valve, reactor, stirring device, temperature sensor, pressure sensor, and liquid level sensor. The digital twin behavior model includes the start and stop actions of the pump, flow meter action, shut-off valve action, regulating valve action, stirring device, and liquid level gauge pointer status. S2. Use smart sensors to collect real-time operating status data of the SIS system emergency shutdown device, conduct remote data analysis and management, and feed the data back to the digital twin model system for real-time visual monitoring of the instrument status. Among them, the shut-off time of the shut-off valve is collected and analyzed, and a control signal is given to quickly respond to the emergency shut-off of the operation of various instruments. Based on convolutional neural networks, the spatial characteristics of the operating data of the emergency shut-off device in the SIS system are extracted, the temporal dependency of the sensor data of the emergency shut-off device in the SIS system is captured, and the key information in the time series is extracted to improve the accuracy of the risk warning of the emergency shut-off device in the SIS system. S3. The intelligent algorithm automatically determines the system status based on the data from various instruments in the SIS system, outputs risk warnings, and provides visual prompts in the digital twin model system; Among them, the collected operating status data is used to identify abnormal medium flow, determine whether there is a deviation fault in the medium transportation, and then issue an alarm; In step S2, a control signal is given, and a method for setting time for a quick response includes: D1. The SIS system obtains the cut-off start time and stage end time, and calculates the cut-off process time = cut-off end time - cut-off start time; D2. Analyze the emergency disconnection time threshold. The disconnection time calculation method is: T=D×1s / in, Among them, D is the diameter of different ball valves, and T is the cut-off time; D3. The fast response delay of the output cut-off device is ≤ 10ms; The step S2 also includes an evaluation model for the operating status of the emergency shutoff device. The evaluation method of the evaluation model is as follows: T1. Extract the trigger time allowable value, response time allowable value, and pressure change allowable value corresponding to the emergency shutoff device model, and compare them with the corresponding detected values ​​respectively. When the detected corresponding value is greater than the corresponding allowable value, calculate the difference between the corresponding allowable value and the corresponding value to obtain the abnormal difference value. Match the abnormal difference value of each corresponding value within the corresponding value range. Set each value range to correspond to a fault alarm value GZ1. After matching the abnormal difference values ​​of each corresponding value, obtain the fault alarm value GZ1 of each corresponding value within the current trigger time point. T2. When the detected corresponding value is less than the corresponding allowable value, calculate the difference between the corresponding allowable value and the corresponding value to obtain the warning difference. Match the warning differences of each corresponding value within the corresponding value range. Set each value range to correspond to a fault threshold GZ2 respectively. After matching the warning differences of each corresponding value, obtain the fault threshold GZ2 of each corresponding value at the current trigger time point. T3. Count the number of fault warning values GZ1 at the current corresponding trigger time point to obtain the number of faults GK. Perform normalization processing among the fault warning values GZ1 or fault thresholds GZ2 of each corresponding value, the number of faults GK at the current trigger time point to obtain the fault hidden danger assessment index at the current corresponding trigger time point: Fk = GZi×al + GZi×a2 + GZi×a3 + (1 + GK)×a4. Calculate the fault hidden danger assessment index Fk at the current corresponding trigger time point, where i = 1 or 2, and al, a2, a3, and a4 are the influence weight factors of the fault thresholds and fault warning values corresponding to the trigger time, response time, and pressure change value respectively. T4. Compare the fault hidden danger assessment index corresponding to the current trigger time point with the corresponding index threshold. Mark the fault hidden danger assessment index as Fk, set the first threshold as F1; the second threshold as F2, and F2 > F1; when Fk < F1, it indicates that the corresponding type of cut-off device is in normal use; when F1 < Fk < F2, it indicates that the corresponding type of cut-off device is in a pending processing state; when F2 < Fk, it indicates that the corresponding type of cut-off device is in an abnormal state.

2. A digital twin emergency shutdown control method based on a SIS system according to claim 1, characterized in that: Among them, The specific method steps for extracting the spatial features of the operation data of the emergency cut-off device in the SIS system based on the convolutional neural network include: F1. Automatically learn the fault patterns and fault data features in the data through multi-layer convolution and pooling operations. F2. Incorporate the long short-term memory network to capture the time-dependent relationship of the sensor data of the emergency cut-off device in the SIS system, enabling the long short-term memory network to handle the long-term dependence problems in the sequence data and effectively extract the key information in the time series. F3. Combine the convolutional neural network and the long short-term memory network. Use the output of the convolutional neural network as the input of the long short-term memory network to construct a convolutional neural network-long short-term memory network hybrid model. Adopt this model to learn the spatial features of the data and capture the time-dependent relationship, thereby improving the accuracy of the risk warning of the emergency cut-off device in the SIS system.

3. The digital twin emergency shutdown control method based on the SIS system according to claim 1 is characterized in that: The method for identifying abnormal medium flow includes: G1. Obtain the flow data of the medium during pipeline transportation. G2. Input the flow data into a pre-trained deviation fault identification model, and reconstruct the flow data through the deviation fault identification model to calculate the deviation value between the flow data and the reconstructed data. G3. If the deviation value exceeds the pre-set deviation threshold, it is determined that a deviation fault occurs in the medium transportation.

4. A digital twin emergency shutdown control method based on a SIS system according to claim 5, characterized in that: The training of the deviation fault identification model includes the following steps: H1. Data preparation and feature mapping Smart sensors collect real-time operating status data of the SIS emergency shut-off device, label normal and abnormal states based on historical fault records, and form a labeled training set. This is then linked to SIS fault diagnosis to provide input data for subsequent models. H2. Model training and optimization H2.

1. Using weighted binary cross entropy, based on the class imbalance characteristic of the SIS system, the weight of positive samples is set to the number of negative samples / the number of positive samples to improve the sensitivity to rare faults. H2.

2. Use the Nadam optimizer to accelerate convergence and avoid gradient explosion, and use ReduceLROnPlateau to dynamically adjust the learning rate to adapt to the nonlinear characteristics of SIS system data; H2.3, 5-fold cross-validation, 100 epochs per fold, batch size = 64, 20% validation set, isolated test set to ensure generalization; early stopping mechanism: monitor the fault risk assessment index of the validation set, and terminate training if there is no improvement after 15 epochs, and restore the optimal weights; H3. Spatial-temporal feature extraction The spatial features of the emergency shut-off device operation data in the SIS system are extracted based on convolutional neural networks; The output of the convolutional neural network (CNN) is used as the input to a fused long short-term memory (LSTM) network to construct a CNN-LSTM hybrid model. This model integrates spatial and temporal features to output the fault probability. Related SIS fault diagnosis: This model directly supports the "deviation fault identification model" described in the manual. By reconstructing the deviation between flow data and the measured value, an over-threshold alarm is triggered. G4, Real-time fault diagnosis and response G4.

1. Fault warning generation: The model outputs the fault probability in real time. Combined with the fault potential assessment index, abnormal conditions immediately trigger visual alarms in the digital twin system. G5, continuous optimization and simulation verification G5.

1. Interactive training unit update: Feedback the actual cutting process status data to the deep neural network unit to iteratively optimize the model parameters; G5.

2. Simulation platform verification: Inject fault scenarios into the SIS system disconnect device simulation environment unit to verify the model robustness.

5. A digital twin emergency shutoff control system based on a SIS system according to any one of claims 1 to 4, characterized in that: include: OPC UA communication protocol, based on the unified architecture of OPC UA communication protocol; . A controller, which controls key components and collects physical sensor signals, feeds back the physical sensor signals to the OPC UA communication protocol, and shuts down various instruments of the device in an emergency; The host computer collects physical sensor signals and controls the host computer to send control commands to the controller to emergency shut down various instruments of the device; A control system simulator, which models the digital twin system, connects the physical entity, digital model, and data, and feeds simulated sensor information back to the OPC UA communication protocol; Database, used to store, query and call the operating status data of the SIS system emergency shut-off device; The SIS system disconnect device simulation environment unit provides a virtualized, secure, and controllable simulation platform for the design, testing, verification, and optimization of the SIS system. The SIS system disconnect device simulation environment unit communicates with the OPC UA communication protocol. A deep neural network unit that interacts in real time with the SIS system shutoff device simulation environment unit to extract spatial features of operating data of the emergency shutoff device in the SIS system; The interactive training unit feeds back status, strategy information, and fault identification information to the deep neural network unit; The OPC UA communication protocol seamlessly accesses the SIS system cutting device simulation environment unit through the database, performs data mapping between the SIS system physical entity and its digital twin model, and realizes remote online visual status detection and control; the simulation results of the SIS system cutting device simulation environment unit are converted into control strategies through the OPC UA communication protocol and fed back to the control system simulator through the OPC UA communication protocol, and the control system simulator forms control instructions and feeds them back to the controller through the OPC UA communication protocol. The controller displays the information to the host computer and controls the host computer to issue control instructions.

Images