Multi-mode cooperative control method and dynamic optimization system of high-reliability reversing valve

Abstract

The invention discloses a multi-mode cooperative control method and a dynamic optimization system of a high-reliability reversing valve, and relates to the technical field of data cooperative control, the cooperative control method comprises the following steps: collecting physical state data of the reversing valve in real time through a multi-dimensional detection module; the physical state data comprises leakage amount, displacement of a sealing block, temperature and pressure parameters, and the multi-dimensional detection module is composed of a helium mass spectrometer leak detector, a displacement sensor and a temperature sensor. A Kalman filtering algorithm and a weighted fusion technology are combined, the leakage amount, sealing displacement and temperature parameters are collected in real time, the leakage misjudgment rate is reduced to 5% or below through multi-dimensional data collaborative sensing, the early warning response time is shortened to 50 ms, and the detection sensitivity and reliability under the extreme working condition are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data collaborative control, and in particular to a multi-modal collaborative control method and dynamic optimization system for a high-reliability reversing valve. Background Art

[0002] As a core component of industrial fluid control, the reliability of the reversing valve is directly related to equipment safety and energy efficiency. Under extreme working conditions such as high pressure, high temperature and highly corrosive media, traditional control methods have significant limitations:

[0003] First, the detection dimension is single, relying on single parameter threshold monitoring such as pressure or displacement. This makes it difficult to coordinate the perception of leakage volume (such as micro-leakage <0.5ppm), seal displacement, and temperature changes, resulting in a leakage misjudgment rate exceeding 20% and delayed warning.

[0004] Second, the redundancy mechanism is rigid. The fixed dual-seal structure or single backup path relies on manual switching, and the fault compensation response time exceeds 500ms, which cannot meet the 100ms emergency response requirement.

[0005] Third, there is a mismatch in energy efficiency regulation. The static pressure-flow strategy ignores the impact of temperature on medium viscosity and seal deformation. The helium injection deviation reaches 35% at low temperatures, and high temperatures easily lead to overcompensation.

[0006] Fourth, control optimization relies on empirical models and lacks multi-physics field coupling simulation verification. The prediction error of micron-level deformation of the sealing interface exceeds 40%. Existing improvement plans, while increasing sensor density or driver redundancy, result in a 25% increase in volume, data fusion delay exceeding 200ms, and unbalanced energy consumption.

[0007] Therefore, there is an urgent need for a collaborative control method that integrates multi-dimensional perception, dynamic gradient compensation and digital twin optimization to break through the comprehensive bottleneck of detection sensitivity, response speed and energy efficiency regulation. Summary of the Invention

[0008] The object of the present invention is to provide a multi-modal coordinated control method and a dynamic optimization system for a high-reliability reversing valve.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a multi-modal coordinated control method for a high-reliability reversing valve, the coordinated control method comprising the following steps:

[0010] Step 1: Real-time acquisition of physical status data of the reversing valve using a multi-dimensional detection module. The physical status data includes leakage, seal block displacement, temperature, and pressure parameters. The multi-dimensional detection module is composed of a helium mass spectrometer leak detector, a displacement sensor, and a temperature sensor. The helium mass spectrometer leak detector is connected to the reversing valve housing via a flange and is used to detect helium leakage.

[0011] Step 2: Preprocess the physical state data based on the edge computing node to generate a dynamic feature matrix. The preprocessing includes data normalization, noise filtering and feature extraction. The noise filtering adopts the Kalman filter algorithm.

[0012] Step 3: According to the dynamic characteristic matrix, a multi-modal control strategy is matched through the decision layer center. The multi-modal control strategy includes a normal mode, a fault mode, and an energy-saving mode, wherein:

[0013] Normal mode controls the output pressure of the drive actuator based on preset parameters;

[0014] The failure mode triggers the three-level compensation mechanism of the redundant sealing components when the leakage exceeds the preset threshold;

[0015] Energy-saving mode dynamically adjusts the helium injection volume and drive power according to temperature parameters;

[0016] Step 4: In the fault mode, the three-level compensation mechanism performs the following actions in sequence:

[0017] When the main seal fails, the redundant block is pushed by the spring to fill the gap;

[0018] When the redundant block compensation fails, it switches to the backup drive path through the electromagnetic clutch;

[0019] In emergency state, the magnetic locking device is triggered to force sealing;

[0020] Step 5: Optimize control instructions based on the digital twin simulation engine. The digital twin simulation engine uses ANSYS to build a multi-physics field model of the reversing valve, simulates high-pressure, high-temperature, and leakage scenarios, generates optimized control parameters, and feeds the instructions back to the drive actuator to form a closed-loop control.

[0021] As a further solution of the present invention: the matching logic of the multimodal control strategy in step 3 further includes:

[0022] When the displacement of the sealing block exceeds a threshold, the redundant compensation mechanism is triggered and the sealing space closing degree is adjusted by driving the bidirectional screw of the actuator. The threshold is 80%-90% of the allowable displacement of the sealing block and is calibrated by the material deformation limit of the sealing component and the rated pressure parameter of the reversing valve.

[0023] When the temperature parameter is lower than the set range, the energy-saving mode reduces the helium injection volume to 70%-80% of the baseline value. The set range is the safe interval of the operating temperature of the reversing valve medium, specifically -20°C to 80°C, and is determined by the calibration data of the temperature sensor.

[0024] As a further solution of the present invention: the switching of the backup drive path in step 4 is achieved by controlling an electromagnetic clutch in the box, the electromagnetic clutch is connected to the main drive motor and the backup motor, and the response time is ≤10ms.

[0025] As a further solution of the present invention: the optimization parameters of the digital twin simulation engine are iteratively updated through a reinforcement learning algorithm, and the reinforcement learning algorithm adopts a deep Q network (DQN), and the training goal is to minimize the weighted sum of leakage and energy consumption.

[0026] As a further solution of the present invention: the detection sensitivity of the helium mass spectrometer leak detector is ≤0.1ppm, and a comprehensive leakage assessment index is generated by combining the data of the displacement sensor with a weighted fusion algorithm.

[0027] The present invention also proposes a dynamic optimization system for a multi-modal coordinated control method of a high-reliability reversing valve, the dynamic optimization system comprising:

[0028] Physical layer module: integrated drive actuator, multi-dimensional detection module, redundant sealing components and control box;

[0029] The driving actuator includes a bidirectional screw adjustment mechanism and an electromagnetic drive unit. The bidirectional screw drives the sleeve block to move through the rotary disk and the connecting shaft to adjust the closure of the sealing shell;

[0030] The multi-dimensional detection module includes a helium mass spectrometer leak detector, a displacement sensor and a temperature sensor, and the helium mass spectrometer leak detector is connected to the sealed housing through a hose;

[0031] The redundant sealing assembly includes a main sealing ring, a spring compensation redundant block and a magnetic emergency locking device;

[0032] The control box adopts a modular design, with partitions set inside to fix the equipment, and each module is connected through a magnetic card slot interface;

[0033] Data layer module: Built-in edge computing unit for multi-source data fusion and feature extraction; the edge computing unit is connected to the sensor through a waterproof connector;

[0034] Decision-making layer module: deploys a digital twin simulation engine and a multi-modal control strategy library to support dynamic instruction generation;

[0035] Interactive interface: Modules are connected via industrial Ethernet, supporting mechanical interface standardization and low-latency signal transmission.

[0036] As a further solution of the present invention: the bidirectional screw adjustment mechanism includes two groups of bidirectional screws, and two groups of socket blocks are movably installed on the surface of each group of bidirectional screws. A long plate is fixed on the top of the socket block and is connected to the sealed shell. The bidirectional screw is driven by rotating the turntable to control the closing of the shell.

[0037] As a further solution of the present invention: the spring compensation redundant block of the redundant sealing assembly is connected to the side of the spring, one end of the spring is fixed to the inner wall of the sealing block, and the displacement data of the redundant block is fed back to the decision layer in real time through the displacement sensor.

[0038] As a further solution of the present invention: the magnetic card slot interface of the control box includes a magnet 1 arranged on the side of the box and a magnet 2 arranged on the side of the telescopic frame, both of which are quickly disassembled and assembled through magnetic attraction. A clamping ring is provided at the bottom of the control box, and the clamping ring is rotatably connected to the fixing ring of the base through a card slot.

[0039] As a further solution of the present invention: the digital twin simulation engine shares model parameters across devices through a federated learning algorithm, the federated learning algorithm uses differential privacy technology to protect data security and supports generalized optimization under multiple working conditions.

[0040] By adopting the above technical solution, compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. This invention uses a multi-dimensional detection module of a helium mass spectrometer leak detector, a displacement sensor, and a temperature sensor, combined with a Kalman filter algorithm and weighted fusion technology, to collect leakage volume, seal displacement, and temperature parameters in real time to generate a comprehensive leakage assessment index. Compared with traditional single-parameter monitoring, multi-dimensional data collaborative perception reduces the leakage misjudgment rate to below 5% and shortens the warning response time to 50ms, significantly improving the detection sensitivity and reliability under extreme working conditions.

[0042] 2. To address the problem of redundant mechanism rigidity, the present invention adopts a three-level compensation mechanism consisting of a spring compensation redundant block, an electromagnetic clutch switching backup drive path, and a magnetic emergency locking device. By triggering a gradient response through a dynamic characteristic matrix, the redundant compensation startup time is ≤80ms when the main seal fails, and the emergency locking response is ≤20ms. This improves the fault compensation efficiency by five times compared to traditional solutions, meeting the 100ms-level high reliability requirement and effectively avoiding system paralysis caused by high-pressure leakage.

[0043] 3. The present invention builds a digital twin simulation engine based on the ANSYS multi-physics field model, combines it with a deep Q network reinforcement learning algorithm, takes minimizing the weighted sum of leakage and energy consumption as the optimization goal, dynamically generates control parameters, and uses a federated learning algorithm to share differentially privately protected model parameters across devices, thereby reducing the helium injection deviation from 35% to 8%, and reducing the high-temperature overcompensation rate by 60%. It achieves adaptive energy efficiency regulation and generalized optimization under multiple working conditions, and improves the overall energy efficiency of the system by more than 25%. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of the multi-modal collaborative control method for a high-reliability reversing valve in the present invention. DETAILED DESCRIPTION

[0045] The specific embodiments of the present invention will be further described below in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0046] In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0047] Please see the attached Figure 1 The present invention provides a multi-modal collaborative control method for a high-reliability reversing valve, and the collaborative control method includes the following steps:

[0048] Step 1: Use a multi-dimensional detection module to collect real-time physical status data of the reversing valve. The physical status data includes leakage, sealing block displacement, temperature, and pressure parameters. The multi-dimensional detection module consists of a helium mass spectrometer leak detector, a displacement sensor, and a temperature sensor. The helium mass spectrometer leak detector is connected to the reversing valve housing via a flange to detect helium leakage.

[0049] Step 2: Preprocess the physical state data based on the edge computing node to generate a dynamic feature matrix. The preprocessing includes data normalization, noise filtering, and feature extraction. Noise filtering uses the Kalman filter algorithm.

[0050] Step 3: According to the dynamic characteristic matrix, the decision layer center matches the multi-modal control strategy, which includes normal mode, fault mode and energy-saving mode, where:

[0051] Normal mode controls the output pressure of the drive actuator based on preset parameters;

[0052] The failure mode triggers the three-level compensation mechanism of the redundant sealing components when the leakage exceeds the preset threshold;

[0053] Energy-saving mode dynamically adjusts the helium injection volume and drive power according to temperature parameters;

[0054] Step 4: In fault mode, the three-level compensation mechanism performs the following actions in sequence:

[0055] When the main seal fails, the redundant block is pushed by the spring to fill the gap;

[0056] When the redundant block compensation fails, it switches to the backup drive path through the electromagnetic clutch;

[0057] In emergency state, the magnetic locking device is triggered to force sealing;

[0058] Step 5: Optimize control instructions based on the digital twin simulation engine. The digital twin simulation engine uses ANSYS to build a multi-physics field model of the reversing valve, simulates high-pressure, high-temperature, and leakage scenarios, generates optimized control parameters, and feeds the instructions back to the drive actuator to form a closed-loop control.

[0059] In one embodiment of the present invention, the matching logic of the multimodal control strategy in step 3 further includes:

[0060] When the displacement of the sealing block exceeds the threshold, the redundant compensation mechanism is triggered, and the closing degree of the sealing space is adjusted by driving the bidirectional screw of the actuator. The threshold is 80%-90% of the allowable displacement of the sealing block, which is calibrated by the material deformation limit of the sealing component and the rated pressure parameter of the reversing valve.

[0061] When the temperature parameter is lower than the set range, the energy-saving mode reduces the helium injection volume to 70%-80% of the baseline value. The set range is the safe operating temperature interval of the reversing valve medium, specifically -20℃ to 80℃, and is determined by the calibration data of the temperature sensor.

[0062] In one embodiment of the present invention: the switching of the backup drive path in step 4 is achieved by controlling an electromagnetic clutch in the box, the electromagnetic clutch is connected to the main drive motor and the backup motor, and the response time is ≤10ms.

[0063] In one embodiment of the present invention: the optimization parameters of the digital twin simulation engine are iteratively updated through a reinforcement learning algorithm, the reinforcement learning algorithm adopts a deep Q network (DQN), and the training goal is to minimize the weighted sum of leakage and energy consumption.

[0064] In one embodiment of the present invention: the detection sensitivity of the helium mass spectrometer leak detector is ≤0.1 ppm, and the data of the displacement sensor is combined with the data through a weighted fusion algorithm to generate a comprehensive leakage assessment index.

[0065] The present invention also proposes a dynamic optimization system for a multi-modal coordinated control method of a high-reliability reversing valve, the dynamic optimization system comprising:

[0066] Physical layer module: integrated drive actuator, multi-dimensional detection module, redundant sealing components and control box;

[0067] The drive actuator includes a bidirectional screw adjustment mechanism and an electromagnetic drive unit. The bidirectional screw drives the sleeve block to move through the turntable and the connecting shaft to adjust the closing of the sealing shell.

[0068] The multi-dimensional detection module includes a helium mass spectrometer leak detector, a displacement sensor, and a temperature sensor. The helium mass spectrometer leak detector is connected to the sealed housing via a hose.

[0069] The redundant sealing assembly includes a main sealing ring, a spring compensation redundant block and a magnetic emergency locking device;

[0070] The control box adopts a modular design, with partitions inside to fix the equipment, and each module is connected through a magnetic card slot interface;

[0071] Data layer module: Built-in edge computing unit for multi-source data fusion and feature extraction; the edge computing unit is connected to the sensor through a waterproof connector;

[0072] Decision-making layer module: deploys a digital twin simulation engine and a multi-modal control strategy library to support dynamic instruction generation;

[0073] Interactive interface: Modules are connected via industrial Ethernet, supporting mechanical interface standardization and low-latency signal transmission.

[0074] In one embodiment of the present invention: the bidirectional screw adjustment mechanism includes two groups of bidirectional screws, and two groups of socket blocks are movably installed on the surface of each group of bidirectional screws. A long plate is fixed on the top of the socket block and connected to the sealed shell. The bidirectional screw is driven by rotating the turntable to control the closing of the shell.

[0075] In one embodiment of the present invention: the spring compensation redundant block of the redundant sealing assembly is connected to the side of the spring, one end of the spring is fixed to the inner wall of the sealing block, and the displacement data of the redundant block is fed back to the decision layer in real time through the displacement sensor.

[0076] In one embodiment of the present invention: the magnetic card slot interface of the control box includes a magnet 1 arranged on the side of the box and a magnet 2 arranged on the side of the telescopic frame, both of which are quickly disassembled and assembled through magnetic attraction. A clamping ring is provided at the bottom of the control box, and the clamping ring is rotatably connected to the fixing ring of the base through the card slot.

[0077] In one embodiment of the present invention: the digital twin simulation engine shares model parameters across devices through a federated learning algorithm. The federated learning algorithm uses differential privacy technology to protect data security and supports generalized optimization under multiple working conditions.

[0078] Example 1: Verification of multi-dimensional perception and dynamic compensation mechanism

[0079] This embodiment verifies the practical effectiveness of the multi-dimensional data fusion and three-level gradient compensation mechanism based on an industrial scenario of a certain type of high-pressure reversing valve.

[0080] 1. Multi-sensor data fusion model

[0081] Definition of comprehensive leakage assessment index I leak for:

[0082] I leak =α·L He +β·ΔS+γ·Tdev

[0083] Among them, L He The leakage rate (unit: ppm) measured by the helium mass spectrometer leak detector, after sensitivity calibration, is σ L ≤0.1, ΔS is the displacement deviation of the sealing block detected by the displacement sensor (unit: mm), the allowable threshold S max =1.2mm, T dev is the deviation between the medium temperature measured by the temperature sensor and the set value (unit: °C), with weight coefficients α = 0.6, β = 0.3, and γ = 0.1, which are determined by optimizing the covariance matrix after Kalman filtering;

[0084] Experimental data show that when I leak When ≥0.8, the system triggers the fault mode. After testing, the misjudgment rate of traditional single-parameter monitoring is 21%, while the misjudgment rate of this solution is reduced to 4.3%, verifying the effectiveness of multi-dimensional fusion.

[0085] 2. Calculation of three-level compensation response time

[0086] When the main seal fails, the dynamic response time t1 of the spring compensation redundant block satisfies:

[0087]

[0088] Where m is the mass of the redundant block (0.5 kg), k is the spring stiffness coefficient (200 N / m), and substituting ΔS = 1.0 mm, we calculate t1 = 68 ms. The measured average value is 72 ms, which meets the design requirement (t1 ≤ 80 ms).

[0089] 3. Electromagnetic clutch switching verification

[0090] The backup drive path switching time t2 is determined by the response equation of the electromagnetic clutch:

[0091]

[0092] Where inductance L = 5mH, resistance R = 2Ω, driving voltage V in =24V, maintaining current I hold =2A, calculated t2=8.3ms, measured t2=9.1ms, which meets the requirement of t2≤10ms.

[0093] Example 2: Digital Twin and Reinforcement Learning Optimization Verification

[0094] This example uses a directional valve optimization scenario for a high-temperature helium delivery system to verify the energy efficiency improvement effect of digital twin simulation and the DQN algorithm:

[0095] 1. Multi-physics coupling simulation modeling

[0096] Construct the sealing interface deformation equation based on ANSYS:

[0097]

[0098] Where P = 20 MPa, r = 50 mm, v = 0.3, the simulation predicted deformation δ sim =

[0099] The actual value is 1.8μm, and the actual value is 1.9μm, with an error of only 5.3%, which is significantly better than the traditional empirical model (error>40%).

[0100] 2. DQN algorithm energy consumption optimization

[0101] Define the reward function R as:

[0102] R=-(w1·L He +w2·P power )

[0103] Among them, w1=0.7,w2=0.3,P power is the driving power (unit: kW). After training, the helium injection deviation of the system at -20°C was reduced from 35% to 7.5%, and the overcompensation rate at high temperature (80°C) was reduced from 22% to 8.6%.

[0104] 3. Generalization Verification of Federated Learning

[0105] Update model parameters using federated learning with differential privacy protection:

[0106]

[0107] The learning rate η = 0.01, the noise variance σ 2 =0.1. After training across 5 devices, the leakage control error of the model is still less than 10% under the unseen high-pressure fluctuation condition, which verifies the generalization ability.

[0108] Example 3: Magnetic Locking Device and Control Box Modular Design Verification

[0109] This embodiment aims to verify the forced sealing performance of the magnetic locking device and the efficiency improvement effect of the modular design in response to emergency sealing requirements and rapid maintenance scenarios of the control box under extreme working conditions.

[0110] Sealing force modeling and response verification of magnetic locking device:

[0111] 1. Magnetic force F mag According to Maxwell stress formula:

[0112]

[0113] Among them, B = 1.2T is the magnetic induction intensity (using NdFeB permanent magnets), A = 50mm 2 is the magnetic contact area, μ0=4π×10 -7 F is calculated from H / m mag =286N The measured sealing force is 275N, with an error of 3.8%, which meets the high-pressure sealing requirement (≥250N). The emergency locking response time t3 is determined by the magnetic attraction stroke d=2mm and the acceleration a:

[0114] (Ignore air resistance)

[0115] Electromagnetic drive acceleration a=F mag / m lock , where the mass of the locking block is m lock = 0.1kg, so a = 2750m / s 2 , substituting into t3 = 1.2ms, the measured average value is 1.5ms, which fully meets the emergency locking response requirements (t3 ≤ 5ms).

[0116] 2. Analysis of the disassembly and assembly efficiency of the modular magnetic card slot

[0117] Control box magnetic card slot interface disassembly and assembly time t dis It can be modeled as:

[0118]

[0119] Among them, t align =0.5s is the positioning alignment time, F attach =50N is the magnetic attraction force (design value of magnet 1 and magnet 2), μ=0.3 is the friction coefficient, v=0.2m / s is the manual disassembly speed, and t is calculated. dis =0.5+0.83=1.33s, and the measured average value is 1.4s, which is 95% more efficient than the traditional bolt fixing solution (≥30s).

[0120] 3. Industrial Ethernet low-latency transmission verification

[0121] Total signal transmission delay T total Including transmission delay T trans With processing delay T proc :

[0122]

[0123] Among them, the data packet size D = 512 bytes, the bandwidth B = 100Mbps, and T trans =0.041ms, number of nodes N = 3, single node processing delay T node = 0.1ms, and T total=0.041+0.3=0.341ms, the measured end-to-end delay is 0.38ms, which is significantly lower than the traditional RS485 bus (≥2ms) and supports millisecond-level closed-loop control.

[0124] Although the present invention is disclosed above with reference to preferred embodiments, this is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications without departing from the spirit and scope of the present invention. Therefore, any modifications, equivalent variations, and modifications made to the above embodiments in accordance with the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection defined by the claims of the present invention.

Claims

1. A multi-modal coordinated control method for a high-reliability reversing valve, characterized by: The collaborative control method comprises the following steps: Step 1: Real-time acquisition of physical status data of the reversing valve using a multi-dimensional detection module. The physical status data includes leakage, seal block displacement, temperature, and pressure parameters. The multi-dimensional detection module is composed of a helium mass spectrometer leak detector, a displacement sensor, and a temperature sensor. The helium mass spectrometer leak detector is connected to the reversing valve housing via a flange and is used to detect helium leakage. Step 2: Preprocess the physical state data based on the edge computing node to generate a dynamic feature matrix. The preprocessing includes data normalization, noise filtering and feature extraction. The noise filtering adopts the Kalman filter algorithm. Step 3: According to the dynamic characteristic matrix, a multi-modal control strategy is matched through the decision layer center. The multi-modal control strategy includes a normal mode, a fault mode, and an energy-saving mode, wherein: Normal mode controls the output pressure of the drive actuator based on preset parameters; The failure mode triggers the three-level compensation mechanism of the redundant sealing components when the leakage exceeds the preset threshold; Energy-saving mode dynamically adjusts the helium injection volume and drive power according to temperature parameters; Step 4: In the fault mode, the three-level compensation mechanism performs the following actions in sequence: When the main seal fails, the redundant block is pushed by the spring to fill the gap; When the redundant block compensation fails, it switches to the backup drive path through the electromagnetic clutch; In emergency state, the magnetic locking device is triggered to force sealing; Step 5: Optimize control instructions based on the digital twin simulation engine. The digital twin simulation engine uses ANSYS to build a multi-physics field model of the reversing valve, simulates high-pressure, high-temperature, and leakage scenarios, generates optimized control parameters, and feeds the instructions back to the drive actuator to form a closed-loop control.

2. A multi-modal coordinated control method for a high-reliability reversing valve according to claim 1, characterized in that: The matching logic of the multimodal control strategy in step 3 further includes: When the displacement of the sealing block exceeds a threshold, the redundant compensation mechanism is triggered and the sealing space closing degree is adjusted by driving the bidirectional screw of the actuator. The threshold is 80%-90% of the allowable displacement of the sealing block and is calibrated by the material deformation limit of the sealing component and the rated pressure parameter of the reversing valve. When the temperature parameter is lower than the set range, the energy-saving mode reduces the helium injection volume to 70%-80% of the baseline value. The set range is the safe interval of the operating temperature of the reversing valve medium, specifically -20°C to 80°C, and is determined by the calibration data of the temperature sensor.

3. The multi-modal coordinated control method for a high-reliability reversing valve according to claim 1, characterized in that: The switching of the backup drive path in step 4 is achieved by controlling an electromagnetic clutch in the box. The electromagnetic clutch is connected to the main drive motor and the backup motor, and the response time is ≤10ms.

4. The multi-modal coordinated control method for a high-reliability reversing valve according to claim 1, characterized in that: The optimization parameters of the digital twin simulation engine are iteratively updated through a reinforcement learning algorithm. The reinforcement learning algorithm adopts a deep Q network, and the training goal is to minimize the weighted sum of leakage and energy consumption.

5. The multi-modal coordinated control method for a high-reliability reversing valve according to claim 1, characterized in that: The detection sensitivity of the helium mass spectrometer leak detector is ≤0.1ppm, and the data of the displacement sensor is combined with the data to generate a comprehensive leakage assessment index through a weighted fusion algorithm.

6. A dynamic optimization system applicable to the collaborative control method described in any one of 1-5, characterized in that: The dynamic optimization system comprises: Physical layer module: integrated drive actuator, multi-dimensional detection module, redundant sealing components and control box; The driving actuator includes a bidirectional screw adjustment mechanism and an electromagnetic drive unit. The bidirectional screw drives the sleeve block to move through the rotary disk and the connecting shaft to adjust the closure of the sealing shell; The multi-dimensional detection module includes a helium mass spectrometer leak detector, a displacement sensor and a temperature sensor, and the helium mass spectrometer leak detector is connected to the sealed housing through a hose; The redundant sealing assembly includes a main sealing ring, a spring compensation redundant block and a magnetic emergency locking device; The control box adopts a modular design, with partitions set inside to fix the equipment, and each module is connected through a magnetic card slot interface; Data layer module: Built-in edge computing unit for multi-source data fusion and feature extraction; the edge computing unit is connected to the sensor through a waterproof connector; Decision-making layer module: deploys a digital twin simulation engine and a multi-modal control strategy library to support dynamic instruction generation; Interactive interface: Modules are connected via industrial Ethernet, supporting mechanical interface standardization and low-latency signal transmission.

7. The dynamic optimization system of the multi-modal coordinated control method for a high-reliability reversing valve according to claim 6 is characterized in that: The bidirectional screw adjustment mechanism includes two groups of bidirectional screws, and two groups of socket blocks are movably installed on the surface of each group of bidirectional screws. A long plate is fixed on the top of the socket block and is connected to the sealed shell. The bidirectional screw is driven by rotating the turntable to control the closing of the shell.

8. The dynamic optimization system of a multi-modal coordinated control method for a high-reliability reversing valve according to claim 6 is characterized in that: The spring compensation redundant block of the redundant sealing assembly is connected to the side of the spring, one end of the spring is fixed to the inner wall of the sealing block, and the displacement data of the redundant block is fed back to the decision layer in real time through the displacement sensor.

9. The dynamic optimization system of a multi-modal coordinated control method for a high-reliability reversing valve according to claim 6, characterized in that: The magnetic card slot interface of the control box includes a magnet 1 arranged on the side of the box and a magnet 2 arranged on the side of the telescopic frame. The two are quickly disassembled and assembled through magnetic attraction. A clamping ring is provided at the bottom of the control box, and the clamping ring is rotatably connected to the fixing ring of the base through the card slot.

10. The dynamic optimization system of a multi-modal coordinated control method for a high-reliability reversing valve according to claim 6, characterized in that: The digital twin simulation engine shares model parameters across devices through a federated learning algorithm that uses differential privacy technology to protect data security and supports generalized optimization under multiple working conditions.

Images