Life prediction-driven collaborative control system and method for industrial dynamic seal actuators

By constructing a sensing and acquisition module, a decision-making and calculation module, and a control and verification module, a closed-loop linkage for wear prediction and control of industrial dynamic seal actuators was achieved, solving the problem of uneven wear rate and improving the system's operational reliability and economy.

CN123113755APending Publication Date: 2026-11-17VALVE BIWEI VALVE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611388424.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-08
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

When industrial dynamic seal actuators are in multi-node collaborative operation, the wear rate is uneven, which leads to excessive wear and premature failure of some nodes. The overall service life is difficult to control effectively, the allocation of operation and maintenance resources is unreasonable, and the traditional operation and maintenance mode lacks forward-looking prediction, making it difficult to meet the requirements of high reliability and low cost.

Method used

The system constructs a sensing and acquisition module, a decision-making and computation module, and a control and verification module. It builds a sensing database through multi-source data acquisition, identifies operating conditions, predicts the remaining service time of each actuator, and performs multi-node load collaborative allocation and scheduling with the goal of balancing wear rate. It generates control commands to achieve hierarchical control and graded operation and maintenance.

Benefits of technology

It achieves a balanced wear and degradation rate for each dynamic seal actuator, avoids excessive local failure, makes full use of the system's service margin, improves operational reliability and life-cycle economy, and realizes the rational allocation of operation and maintenance resources.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113755A_ABST
    Figure CN123113755A_ABST
Patent Text Reader

Abstract

This invention discloses a life prediction-driven collaborative control system and method for industrial dynamic seal actuators, belonging to the field of industrial intelligent control technology. It aims to solve the problems of uneven wear, difficulty in predicting remaining lifespan, and lagging operation and maintenance scheduling in dynamic seal actuators. It includes a sensing and acquisition module, a decision-making and computation module, and a control and verification module. This application constructs a sensing database through multi-source data acquisition, infers wear trends, and predicts remaining service time. It performs multi-node load collaborative allocation with the goal of balancing wear rates, and achieves hierarchical control and operation and maintenance scheduling based on cumulative wear. This effectively extends the overall service life and significantly improves the operational reliability and operation and maintenance efficiency of industrial dynamic seal actuators.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of industrial intelligent control technology, and more specifically to a life prediction-driven collaborative control system and method for industrial dynamic seal actuators. Background Technology

[0002] Industrial dynamic seal actuators are widely used in process industries such as petrochemicals, power generation, and metallurgy, serving as key components to ensure the tightness and operational stability of fluid transport systems. In actual industrial scenarios, multi-node dynamic seal actuators often operate in a coordinated manner. Due to multiple factors such as fluctuations in operating load, differences in installation and assembly, and uneven media environments, the wear and degradation processes of each actuator node vary significantly.

[0003] Currently, the industry mostly adopts an operation and maintenance model of independent monitoring and periodic maintenance of single nodes. This makes it difficult to coordinate the wear status and service rhythm of multiple nodes at the system level, which easily leads to an imbalance where some nodes fail prematurely due to excessive wear, while the wear margin of other nodes is not fully utilized. This increases the risk of unplanned downtime and results in the unreasonable allocation of operation and maintenance resources. At the same time, traditional reactive or periodic maintenance models lack forward-looking prediction of the remaining service life of seals, and operation and maintenance scheduling often lags behind the wear and degradation process, making it difficult to meet the dual requirements of high reliability and low-cost operation and maintenance for continuous industrial production. Therefore, in order to overcome these limitations, this invention proposes a life prediction-driven collaborative control system and method for industrial dynamic seal actuators. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a life prediction-driven collaborative control system and method for industrial dynamic seal actuators, which solves the technical problems of uneven wear rate, difficulty in effectively controlling the overall service life, and passive operation and maintenance of industrial dynamic seal actuators.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A life prediction-driven collaborative control system for industrial dynamic seal actuators includes:

[0007] The sensing and acquisition module is used to collect operating data, actuator feedback data, and wear status data, and to perform data preprocessing to build a sensing database;

[0008] The decision-making and computing module is used to identify the operating mode of each dynamic seal actuator based on the perception database and construct the wear evolution prediction trajectory to predict the remaining service time of each dynamic seal actuator; and to perform multi-node load collaborative allocation and scheduling with the goal of balancing the real-time wear rate distribution of each dynamic seal actuator, and generate multi-node load scheduling instructions.

[0009] The control and verification module is used to determine the control demand level based on the current cumulative wear of each dynamic seal actuator node within the constraint boundary formed by the multi-node load scheduling command, and to activate the calculation of the hierarchical control parameters step by step. The hierarchical control parameters are then issued for execution, and the urgency of the operation and maintenance of each dynamic seal actuator node is determined to generate hierarchical operation and maintenance scheduling commands.

[0010] Specifically, the decision computing module includes:

[0011] The working condition identification unit is used to identify the current working condition mode of each dynamic seal actuator based on the core working condition feature parameters in the perception database.

[0012] The life prediction unit is used to construct the wear evolution prediction trajectory for each working condition, generate the remaining service time prediction value and corresponding confidence interval for each dynamic seal actuator, and determine the wear risk type of each dynamic seal actuator node.

[0013] The collaborative scheduling unit is used to construct a multi-layer constraint set, with the optimization objective of minimizing the dispersion of the real-time wear rate distribution of each dynamic seal actuator. It solves the load allocation ratio of each dynamic seal actuator node and generates multi-node load scheduling instructions.

[0014] Specifically, the steps for constructing the wear evolution prediction trajectory for each operating condition mode include:

[0015] Based on the Archard wear model, combined with the performance parameters of the metal material and the geometric parameters of the sealing surface, the reference wear rate and maximum allowable wear of the sealing friction pair are calculated; and the material wear life calibration curve under preset standard working conditions is obtained.

[0016] Based on the Archard wear model, the reference wear rate is substituted to obtain a linear prediction relationship between the cumulative wear amount and the cumulative number of actions under the preset standard working conditions.

[0017] Using the material wear life calibration curve as a physical constraint, the linear prediction relationship is piecewise corrected to form a corrected benchmark wear prediction function, which serves as the baseline for wear evolution prediction.

[0018] A mapping function is established, which takes the working condition mode identifier and core working condition characteristic parameters as input and the working condition deviation correction coefficient as output, as the working condition deviation correction model.

[0019] The wear evolution prediction baseline is coupled with the working condition deviation correction coefficient output by the working condition deviation correction model to obtain the wear evolution prediction trajectory for each working condition mode.

[0020] Specifically, the remaining service life prediction value and corresponding confidence interval of each dynamic seal actuator are generated, and the wear risk type of each dynamic seal actuator node is determined, including:

[0021] Receive the current cumulative wear amount and real-time wear rate, as well as the operating mode identifier;

[0022] Within each prediction cycle, the baseline for predicting wear evolution under a preset standard working condition is derived. The working condition deviation correction model calculates the working condition deviation correction coefficient based on the current working condition mode identifier and core working condition characteristic parameters, and then couples it to obtain the wear evolution prediction trajectory.

[0023] By extrapolating along the wear evolution prediction trajectory, the cumulative number of actions that first made the wear evolution prediction trajectory greater than or equal to the maximum allowable wear amount is retrieved. Starting from the current cumulative number of actions, the predicted value of the remaining service time is calculated.

[0024] The confidence intervals corresponding to the predicted remaining service time are generated using a bootstrap resampling method.

[0025] The predicted remaining service life and confidence intervals of all dynamic seal actuators are summarized, and the predicted remaining service life are compared with the preset safety threshold to determine the wear risk type of each dynamic seal actuator node.

[0026] Specifically, the steps for generating multi-node load scheduling instructions include:

[0027] Receive the predicted remaining service time, confidence interval, and wear risk type of each dynamic seal actuator node, as well as the current operating mode identifier;

[0028] Construct a set of constraints, which includes total system load constraints, safe operating boundary constraints, and additional constraints for prediction uncertainties.

[0029] The objective function constrained optimization problem is established with the goal of minimizing the standard deviation of the real-time wear rate of each dynamic seal actuator. The decision variables are the load differential pressure distribution ratio, the flow ratio distribution ratio, and the upper limit of the maximum operating frequency per unit cycle for each dynamic seal actuator node.

[0030] The objective function constraint optimization problem is solved, and the decision variable values ​​of the convergence point are output as the update values ​​of the bearing pressure difference distribution ratio, flow ratio distribution ratio and maximum action frequency limit per unit cycle for each dynamic seal actuator node, and encapsulated as a multi-node load scheduling instruction.

[0031] Specifically, identifying the current operating mode of each dynamic seal actuator includes:

[0032] Receive the periodic sensing dataset of the current acquisition cycle from the sensing database, and extract the core working condition feature parameters from the periodic sensing dataset;

[0033] Configure corresponding operating condition thresholds for each core operating condition feature parameter, compare each core operating condition feature parameter with its corresponding operating condition threshold, determine the operating condition mode based on the combination matching result of whether each core operating condition feature parameter exceeds the corresponding operating condition threshold, and output the operating condition mode identifier.

[0034] Specifically, the control and verification module includes:

[0035] The control parameter calculation unit is used to determine the control demand level based on the cumulative wear amount within the constraint boundary formed by the multi-node load scheduling instructions, select the control parameter mapping model in combination with the operating mode identifier, activate and calculate the hierarchical control parameters level by level according to the control demand level, and send the hierarchical control parameters to the execution interaction unit for execution.

[0036] The execution interaction unit is used to receive hierarchical control parameters, drive the actuators of each dynamic seal actuator to complete the action, and at the same time, based on the predicted value of the remaining service time of each dynamic seal actuator and the confidence interval, determine the urgency of the operation and maintenance of each dynamic seal actuator node and generate hierarchical operation and maintenance scheduling instructions.

[0037] Specifically, the steps for calculating the stratified control parameters include:

[0038] Receive multi-node load scheduling instructions and extract their load differential pressure distribution ratio, current ratio distribution ratio, maximum action frequency limit per unit cycle, and operating mode identifier;

[0039] The bearing pressure difference distribution ratio, the flow rate distribution ratio, and the upper limit of the maximum frequency of action per unit cycle constitute the constraint boundary;

[0040] The system receives the predicted remaining service time and cumulative wear of the dynamic seal actuator node, and determines the current control demand level of the dynamic seal actuator node based on the ratio of the cumulative wear to the maximum allowable wear. The control demand level includes mild control level, moderate control level, and severe control level.

[0041] Based on the combination of the current control demand level and the operating mode identifier, the corresponding control parameter mapping model is retrieved; the control parameter mapping model is used to store the reference mapping coefficients of each layer of control parameters under different control demand levels and different operating mode combinations.

[0042] By combining the benchmark mapping coefficients and calculating the control parameters step by step according to the level of control demand, the hierarchical control parameters are obtained.

[0043] Specifically, by combining the benchmark mapping coefficients and calculating the control parameters level by level according to the level of control demand, the hierarchical control parameters are obtained, including:

[0044] When the level of regulation demand reaches the level of mild regulation or above, the first layer of control parameter calculation is activated to decompose the single opening and closing action of the opening and closing drive unit into an initial acceleration stage, an intermediate constant speed stage and a final deceleration stage. Combined with the reference mapping coefficient, the segmented target speed value of the opening and closing drive unit and the speed change rate limit between each stage are output.

[0045] When the regulation demand level reaches the medium regulation level or above, the second layer of control parameter calculation is activated on the basis of the first layer of control parameter calculation. This is used to correct the constraint boundary within the constraint boundary formed by the multi-node load scheduling instructions according to the benchmark mapping coefficient, and calculate the operating constraint conditions item by item.

[0046] When the regulation demand level reaches the severe regulation level, the third layer of control parameter calculation is superimposed and activated, and the sealing gap compensation stroke that the active preload drive subunit should apply along the normal direction of the sealing surface is calculated in combination with the reference mapping coefficient.

[0047] A life prediction-driven collaborative control method for industrial dynamic seal actuators includes:

[0048] Collect operating data, actuator feedback data, and wear status data, and perform data preprocessing to build a perception database;

[0049] Identify the operating modes of each dynamic seal actuator based on a perception database;

[0050] Construct wear evolution prediction trajectories to predict the remaining service life of each dynamic seal actuator;

[0051] With the goal of balancing the real-time wear rate distribution of each dynamic seal actuator, multi-node load collaborative allocation and scheduling is performed to generate multi-node load scheduling instructions;

[0052] Within the constraint boundary formed by the multi-node load scheduling command, the control demand level is determined based on the current cumulative wear of each dynamic seal actuator node, and the solution of the hierarchical control parameters is activated step by step.

[0053] The hierarchical control parameters are issued and executed, and hierarchical operation and maintenance scheduling instructions are generated based on the urgency of operation and maintenance of each dynamic seal actuator node.

[0054] The beneficial effects of this invention are:

[0055] This invention constructs a perception database through multi-source data acquisition, and combines operating condition pattern recognition and wear evolution trend inference to accurately predict the remaining service time of each dynamic seal actuator, providing a reliable forward-looking basis for system operation decisions. Based on this, it conducts multi-node load collaborative allocation and scheduling with the goal of balancing the real-time wear rate distribution of each node, effectively smoothing the wear degradation rhythm of each dynamic seal actuator, avoiding premature failure of local nodes, and fully utilizing the overall system's service margin. Simultaneously, relying on the cumulative wear amount to determine the control demand level and gradually activate the hierarchical control parameters, it can implement differentiated and refined control for different wear states. Combined with hierarchical operation and maintenance scheduling instructions, it achieves reasonable allocation of operation and maintenance resources, comprehensively improving the operational reliability and life-cycle economy of the industrial dynamic seal actuator system. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the industrial dynamic seal actuator collaborative control system driven by life prediction according to the present invention;

[0057] Figure 2 This is a schematic diagram of the decision-making calculation module of the present invention;

[0058] Figure 3 A flowchart for constructing wear evolution prediction trajectories under various working conditions in this invention;

[0059] Figure 4 This is a flowchart of the control and verification module of the present invention. Detailed Implementation

[0060] Example 1

[0061] In the current process of wear control of dynamic seal actuators in industrial fluid pipeline systems, existing solutions generally suffer from problems such as a lack of closed-loop linkage between wear prediction and active control, disconnect between single-device control and system-level load scheduling, and a lack of a unified long-term iterative benchmark leading to a continuous decline in control accuracy. Specifically, wear life prediction results cannot automatically trigger control decisions, and the control effect cannot correct the prediction model in reverse, resulting in an isolated open-loop state between prediction and control; each dynamic seal actuator operates, wears, and is repaired independently, leading to a severe imbalance in wear rates at different nodes within the system, and premature failure of individual nodes forcing the entire system to shut down; furthermore, time-varying factors such as sensor zero-point drift and changes in operating environment cause the monitoring and prediction accuracy to continuously decline over time, and existing solutions lack the ability to self-calibrate the benchmark throughout the entire life cycle. To solve the above technical problems, this embodiment provides a life prediction-driven collaborative control system for industrial dynamic seal actuators. The dynamic seal actuator is a type of actuator that relies on the relative motion between sealing friction pairs to open, close, or regulate fluid channels, including, for example, any one or more of metal hard-seal fixed ball valves, metal hard-seal floating ball valves, metal-seal butterfly valves, and plug valves. In this embodiment, a metal hard-seal ball valve is used as a representative dynamic sealing actuator for explanation. The core sealing structure of the metal hard-seal ball valve is a pair of directly contacting metal sealing friction pairs. During the opening and closing process, sliding friction occurs between the ball surface and the valve seat sealing surface, resulting in continuous material wear.

[0062] Please see Figure 1 The collaborative control system includes a sensing and acquisition module, a decision-making and calculation module, and a control and verification module; the modules establish a bidirectional data communication link through an industrial fieldbus or an industrial Ethernet network.

[0063] The sensing and acquisition module is used to collect operating condition data, actuator feedback data, and wear status data at the physical sensing front-end level of the collaborative control system, and to perform data preprocessing to build a sensing database.

[0064] The sensing and acquisition module includes a sensing and acquisition unit and a preprocessing unit;

[0065] The sensing and acquisition unit is used to collect operating data, actuator feedback data, and wear status data that reflect the changes in the physical state of the sealing friction pair, the action status of the actuator, and the operating conditions of the pipeline fluid, through sensor groups deployed on each dynamic sealing actuator body and pipeline, according to a preset acquisition cycle.

[0066] Furthermore, the execution steps of the sensing and acquisition unit include:

[0067] According to the preset acquisition cycle, a synchronous trigger signal is sent to the data acquisition channel of all dynamic seal actuator nodes to synchronously acquire operating data, actuator feedback data, and wear status data:

[0068] The operating data includes: pressure difference of the medium in the upstream and downstream pipelines of the dynamic seal actuator, that is, the difference in medium pressure between the inlet and outlet pipelines of the dynamic seal actuator, which is obtained by installing a pressure transmitter at each end and calculating the difference; real-time temperature of the pipeline medium, which is obtained by a temperature sensor installed on the pipe wall near the dynamic seal actuator or directly immersed in the medium; and frequency of action per unit time, that is, the ratio of the total number of opening, closing or regulating actions actually performed by the dynamic seal actuator within the current statistical time window to the duration of the time window.

[0069] The actuator feedback data includes: the output torque of the opening / closing drive unit, which is obtained directly through the current feedback signal of the actuator motor driver or a torque sensor, and is used to reflect the frictional resistance level of the sealing friction pair during operation; and the stroke feedback signal of the opening / closing drive unit, which is obtained through the actuator motor encoder or displacement sensor, and is used to record the actual stroke and switch position status for each operation. The opening / closing drive unit refers to the actuator assembly that drives the dynamic seal actuator to complete opening / closing or adjustment actions, including a drive motor, a transmission reduction mechanism, and a position feedback device. Both the output torque and the stroke feedback signal are output in real time by the opening / closing drive unit during operation.

[0070] The wear status data includes cumulative wear and real-time wear rate. The cumulative wear and real-time wear rate are acquired by a front-end wear sensing sensor. Specifically, the front-end wear sensing sensor is installed on the body of each dynamic seal actuator, forming a physical measurement coupling relationship with the sealing friction pair. It directly senses the physical state changes of the sealing friction pair and outputs a raw clearance change signal. The raw clearance change signal is corrected by the zero-point calibration reference signal value collected by each sensor during the factory calibration stage from the wear-free calibration reference surface, and then converted into the cumulative wear. The real-time wear rate is obtained by performing time difference calculation on the cumulative wear. For example, the front-end wear sensing sensor includes one or more eddy current displacement sensors arranged circumferentially along the sealing surface, which output the raw clearance change signal by measuring the real-time clearance change between the valve ball surface and the valve seat sealing surface. The working principle of the eddy current displacement sensor is as follows: The sensor probe coil generates an alternating magnetic field under high-frequency current excitation. When the probe approaches the metal surface being measured, eddy currents are induced on the metal surface. The counter-magnetic field generated by the eddy currents changes the equivalent impedance of the probe coil. The change in equivalent impedance is proportional to the distance from the probe to the metal surface, thus achieving non-contact gap measurement. By uniformly arranging multiple measurement points along the circumference of the sealing surface, the wear distribution along the circumference of the sealing surface can be obtained, eliminating measurement deviations caused by uneven local wear on the sealing surface.

[0071] The preprocessing unit is used to perform outlier removal and time-series alignment processing on the operating data, actuator feedback data, and wear status data, construct a perception database, and distribute it to the decision-making and computing module.

[0072] Furthermore, the execution steps of the preprocessing unit include:

[0073] For operating condition data, actuator feedback data, and wear status data, an out-of-bounds judgment is performed according to a preset physical reasonableness range. Data points exceeding the physical reasonableness range are marked as outliers and removed from the data frames of the current acquisition cycle. The corresponding data values ​​from the previous valid acquisition cycle are then used for interpolation to fill the gaps. The valid acquisition cycle refers to the acquisition cycle in which no data frames from any data channel are marked as outliers. The physical reasonable range is preset based on the physical meaning of each data type and the actual working condition characteristics. Specifically, the upper limit of the physical reasonable range of pipeline medium pressure difference is set according to the rated pressure level of each dynamic seal actuator, for example, it is set to 120% of the rated pressure level; the lower limit is zero, and the pressure difference should not be negative. The physical reasonable range of pipeline medium real-time temperature is set according to the design temperature range of the process flow and the ambient temperature range, for example, it is set from the ambient temperature minus 10°C to the upper limit of the design temperature plus 20°C. The upper limit of the physical reasonable range of the frequency of action per unit time is set according to the maximum allowable frequency of action. The upper limit of the physical reasonable range of the output torque is set according to the maximum allowable frequency of action of the actuator. The upper limit of the physical reasonable range of the cumulative wear is set according to the corresponding sensor range calculated from the maximum allowable wear of the sealing friction pair.

[0074] The system performs time alignment processing on operating condition data, actuator feedback data, and wear status data. This process unifies and aligns time stamp deviations caused by differences in sampling frequency or transmission delay between different acquisition channels, ensuring that the time stamp deviation of all data frames belonging to the same acquisition cycle does not exceed a preset time alignment tolerance. The preset time alignment tolerance is set to 10% of the preset acquisition cycle. Specifically, the time alignment processing involves arranging the data frames acquired from each channel in ascending order of time stamps. Using the arrival time of the slowest channel as a reference, the time stamps of other channels are aligned to this reference. If the deviation of a data frame's time stamp from the reference exceeds the preset time alignment tolerance, the data frame is determined not to belong to the current acquisition cycle and is relegated to the previous or next acquisition cycle.

[0075] After the above processing is completed, the data is distributed to the decision-making and computing module. At the same time, all operating data, actuator feedback data and wear status data are continuously accumulated to form a perception database. The perception database provides a continuously expanding training data foundation for the online incremental training of the subsequent life prediction model.

[0076] Specifically, the sensing and acquisition module is designed to address core issues in industrial environments, such as heterogeneous multi-source data channels, inconsistent acquisition cycles, significant differences in transmission delays, lack of a unified spatiotemporal reference, and susceptibility of sensor signals to electromagnetic interference. In industrial pipeline environments, pressure transmitters output 4-20mA analog signals, temperature sensors output RTD signals, eddy current displacement sensors output high-frequency carrier-modulated signals, and actuator motor drivers output digital communication signals. Each signal has a different physical interface, communication protocol, and sampling frequency. Traditional acquisition methods only perform simple multi-channel signal access without quality control or timing alignment, resulting in poor reliability of input data for subsequent operational condition identification and lifespan prediction. This module provides high-quality, standardized, and traceable data input to the upper-level decision-making and computation module through a unified synchronous triggering mechanism, automatic outlier detection and interpolation, high-precision timing alignment, and continuous construction of the sensing database.

[0077] The decision-making and computation module receives the sensing database distributed by the sensing and acquisition module, identifies the current operating mode of each dynamic seal actuator as the operating condition constraint for subsequent decisions, and performs wear evolution trend deduction on the sealing friction pair based on the operating mode and historical operating data in the sensing database, constructs wear evolution prediction trajectory, predicts the remaining service time of each dynamic seal actuator and generates corresponding confidence intervals. Based on the global wear risk distribution revealed by the remaining service time prediction results, it performs multi-node load collaborative allocation scheduling with the goal of balancing the real-time wear rate distribution of each dynamic seal actuator in the entire system, and outputs the scheduling results as the constraint boundary for subsequent control parameter calculation to the control verification module.

[0078] Please see Figure 2 The decision-making module includes a working condition identification unit, a lifespan prediction unit, and a collaborative scheduling unit.

[0079] The operating condition identification unit is used to identify the current operating condition mode of each dynamic seal actuator based on the distribution position of the core operating condition feature parameters in the multi-dimensional parameter space of the perception database. The identified operating condition mode identifier is then synchronously provided as an operating condition constraint to the life prediction unit and the collaborative scheduling unit, enabling the aforementioned decision-making processes to adaptively adjust based on the current operating conditions during their respective execution. The core operating condition feature parameters refer to key parameters extracted from the perception database that characterize the severity of the operating conditions of the dynamic seal actuator and directly affect the wear rate of the sealing friction pair.

[0080] Furthermore, the execution steps of the operating condition identification unit include:

[0081] The system receives the periodic sensing dataset from the sensing database transmitted in real-time by the preprocessing unit. This dataset has undergone outlier removal and time-series alignment processing within the current acquisition cycle. The periodic sensing dataset is a structured data set consisting of all operating condition data, actuator feedback data, and wear status data processed by the preprocessing unit within a single acquisition cycle. It includes time stamps and all fields of the operating condition data, actuator feedback data, and wear status data corresponding to the acquisition cycle. Four core operating condition characteristic parameters are extracted from the periodic sensing dataset: upstream and downstream pipeline medium pressure difference, real-time pipeline medium temperature, frequency of action per unit time, and output torque of the start / stop drive unit.

[0082] Based on the rated design parameters of each dynamic seal actuator and the wear sensitivity characteristics of the sealing friction pair, a corresponding operating condition threshold is configured for each core operating condition characteristic parameter. The operating condition threshold refers to a preset grading limit value for the core operating condition characteristic parameter, used to divide the continuous parameter value range into discrete operating condition level intervals. Specifically, a high differential pressure threshold is configured for the pressure difference between the upstream and downstream pipeline media, exemplarily set to 60% of the rated pressure level of the dynamic seal actuator; a high temperature threshold is configured for the real-time temperature of the pipeline media, exemplarily set to the upper limit of the process design temperature minus 20°C; a high frequency threshold is configured for the frequency of action per unit time, exemplarily set to 70% of the maximum allowable frequency of action of the actuator; and a high torque threshold is configured for the output torque, exemplarily set to 80% of the maximum allowable drive torque of the actuator.

[0083] The core operating condition feature parameters are compared with the corresponding operating condition thresholds. The operating condition mode is determined based on the combination matching result of whether each core operating condition feature parameter exceeds the corresponding operating condition threshold, and the operating condition mode identifier is output. For example, the determination of the operating mode includes: when all core operating mode characteristic parameters do not exceed the corresponding operating mode threshold, it is determined to be a normal temperature and pressure low frequency operating mode; when the real-time temperature of the pipeline medium exceeds the high temperature threshold, the pressure difference between the upstream and downstream pipeline media exceeds the high pressure difference threshold, and the frequency of action per unit time exceeds the high frequency threshold, it is determined to be a high temperature and high pressure high frequency operating mode; when the pressure difference between the upstream and downstream pipeline media exceeds the high pressure difference threshold but does not meet all the conditions of the high temperature and high pressure high frequency operating mode, it is determined to be a high pressure difference start-up and shut-down operating mode; when the change amplitude of any core operating mode characteristic parameter between two adjacent acquisition cycles exceeds the preset transition change rate threshold, it is determined to be a variable operating mode transition mode. The variable operating mode transition mode indicates that the system is transitioning from one steady-state operating mode to another steady-state operating mode. At this time, the operating mode identifier remains as a variable operating mode transition mode until the change amplitude of all core operating mode characteristic parameters recovers to below the preset transition change rate threshold.

[0084] The operating condition mode identifier is synchronously transmitted to the life prediction unit as an operating condition constraint input for wear evolution trend prediction; and to the collaborative scheduling unit as an operating condition constraint for load scheduling decision.

[0085] The life prediction unit is used to obtain the material performance parameters and geometric parameters of the sealing friction pair as the basis for model construction, construct the wear evolution prediction trajectory of each working condition mode in combination with the physical wear mechanism, generate the remaining service time prediction value and corresponding confidence interval of each dynamic seal actuator, and determine the wear risk type of each dynamic seal actuator node in the system based on the comparison result of the remaining service time prediction value and the preset safety threshold, and output the prediction result to the collaborative scheduling unit and the control verification module.

[0086] Please see Figure 3 Furthermore, the execution steps of the lifetime prediction unit include:

[0087] The system receives the current cumulative wear amount and real-time wear rate transmitted in real time by the preprocessing unit, as well as the working condition mode identifier output by the working condition identification unit.

[0088] Obtain the metal material performance parameters and sealing surface geometric parameters of the sealing friction pair. The metal material performance parameters include material hardness and elastic modulus. The sealing surface geometric parameters are obtained by performing standard hardness tests on samples of the sealing friction pair material; the sealing surface geometric parameters include the sealing surface width. , sealing surface contact area and the normal projected area of ​​the sealing surface .

[0089] Based on the Archard wear model, and combining the performance parameters of the metal material and the geometric parameters of the sealing surface, the reference wear rate and maximum allowable wear amount of the sealing friction pair are calculated. The basic expression of the Archard wear model is:

[0090]

[0091] in For wear volume, The wear coefficient is a dimensionless coefficient. For the contact surface normal load, The sliding distance, The hardness of the sealing friction pair is given by the material. The dimensionless wear coefficient k is obtained by measuring the mating metal materials of the sealing friction pair in a standard pin-disc wear test, characterizing the inherent wear characteristics of a specific material pair under given lubrication conditions, and its value ranges from 10. -5 Up to 10 -3 The contact surface normal load The contact area of ​​the sealing surface during a single opening and closing action. The resultant normal force on the surface, under the preset standard working conditions, ,in The rated pressure value of the pipeline medium corresponding to the preset standard operating conditions. The normal projected area of ​​the sealing surface; the sliding distance The total relative sliding stroke between the two contact surfaces of the sealing friction pair during a single complete opening and closing action. ,in The nominal diameter of the valve ball sealing surface. This refers to the valve ball rotation angle corresponding to a single opening and closing action. The preset standard operating condition refers to the steady-state operating condition of the dynamic seal actuator under the rated pressure level, the median of the design temperature range, and the rated operating frequency.

[0092] Specifically, the steps for calculating the reference wear rate are as follows:

[0093] The dimensionless wear coefficient The contact surface normal load The sliding distance and the hardness of the material Substituting the basic expression of the Archard wear model, the theoretical volumetric wear amount corresponding to a single opening and closing action is obtained. Then the theoretical volumetric wear amount Divide by the contact area of ​​the sealing surface The wear depth increment corresponding to a single opening and closing action is obtained. The reference wear rate is the wear depth increment corresponding to a single opening and closing action under the preset standard operating conditions. .

[0094] Specifically, the steps for calculating the maximum permissible wear amount are as follows:

[0095] Based on the width of the sealing surface and preset minimum sealing pressure requirements Calculate the maximum allowable wear depth of the sealing friction pair while maintaining an effective seal. The preset minimum sealing pressure requirement... This refers to the minimum contact pressure per unit area required to maintain a leak-free sealing surface, determined by the sealing structure design of the sealing friction pair. When the wear depth of the sealing surface reaches... At this time, the effective contact width of the sealing surface is reduced to the point that the actual sealing specific pressure drops to The state is thus determined. This refers to the maximum permissible wear of the sealing friction pair. For example, for a planar contact type sealing friction pair, For spherical contact type sealing friction pairs, The sealing surface curvature radius and sealing surface width are jointly determined and calibrated through sealing performance tests at the factory stage.

[0096] The material wear life calibration curves of each sealing friction pair material were obtained through laboratory wear tests under the preset standard operating conditions. The material wear life calibration curves consist of a set of discrete data points. ,in To accumulate the number of actions, The curve represents the corresponding measured cumulative wear amount and is used to characterize the evolution of the cumulative wear amount with the number of operations under the preset standard operating conditions.

[0097] Based on the benchmark wear rate The maximum permissible wear amount Based on the material wear life calibration curve, the wear evolution prediction calculation model is constructed. The wear evolution prediction calculation model is a cascaded coupled calculation model that uses the physical wear mechanism of the sealing friction pair as the framework, actual working condition data as the basis for deviation correction, and outputs the wear evolution prediction trajectory of the cumulative wear amount with the number of operations as the objective.

[0098] Furthermore, the steps for constructing the wear evolution prediction computational model include:

[0099] Based on the linear wear relationship of the Archard wear model, the reference wear rate is... Substituting the values, we obtain a linear prediction relationship between the cumulative wear amount and the cumulative number of operations under the preset standard working conditions. .

[0100] Discrete data points of the material wear life calibration curve As a physical constraint, the linear prediction relationship is... Perform segmented correction: Divide the range of the number of operations of the material wear life calibration curve into several equal intervals, and calculate the slope of the tangent of the calibration curve and the reference wear rate in each interval. The ratio is used as a nonlinear correction factor. An interpolation mapping table is established between the correction factor in each interval and the corresponding interval of action number to form the corrected benchmark wear prediction function. The This is the baseline for predicting wear evolution under the preset standard operating conditions.

[0101] A working condition deviation correction model is constructed. This model is a mapping function that receives the working condition mode identifier and the core working condition characteristic parameters, and outputs a working condition deviation correction coefficient. The working condition deviation correction coefficient is a dimensionless correction quantity that characterizes the comprehensive impact of the deviation of the current working condition mode from the preset standard working condition on the wear rate. A positive value indicates that the wear rate under the current working condition is higher than the preset standard working condition; a negative value indicates that it is lower than the preset standard working condition; and a value of zero indicates that it is consistent with the preset standard working condition. Its internal structure is as follows: for each working condition mode... Define a set of operating condition characteristic parameters and an operating condition deviation correction coefficient vector. For example, for four core operating condition characteristic parameters—downstream pipeline medium differential pressure, real-time pipeline medium temperature, frequency of action per unit time, and output torque—the operating condition deviation correction coefficient vector is defined. These correspond to the independent correction contributions to the wear rate when the four core operating condition characteristic parameters—pressure difference between upstream and downstream pipeline media, real-time temperature of pipeline media, frequency of action per unit time, and output torque—deviate from the preset standard operating condition. In any of the aforementioned operating condition modes... Below, the working condition deviation correction coefficient The calculation method is as follows:

[0102] The relative deviations of the actual values ​​of the core operating condition characteristic parameters in the current acquisition period are calculated by comparing them with the corresponding benchmark values ​​under the preset standard operating conditions. The relative deviations are then multiplied by the corresponding correction coefficient vector. The working condition deviation correction coefficient is obtained by summing the components. The reference value refers to the value of each core operating condition characteristic parameter under the preset standard operating condition, including the rated pressure value of the pipeline medium, the median of the design temperature range, the rated operating frequency, and the theoretical output torque under the preset standard operating condition. The relative deviation is the dimensionless ratio obtained by dividing the difference between the actual value of the core operating condition characteristic parameter in the current acquisition period and the corresponding reference value by the corresponding reference value; it is used to characterize the degree of deviation of the current operating condition from the preset standard operating condition. During the initial operation phase, the correction coefficient vector... Each component adopts a conservative preset value based on physical boundary estimation. The conservative preset value is set based on:

[0103] As shown in the Archard wear model, the wear volume is directly proportional to the normal load on the contact surface and inversely proportional to the material hardness. The normal load on the contact surface is determined by the pipeline medium pressure, while the material hardness decreases with increasing temperature. Simultaneously, the sliding distance is determined by the frequency of action, and the output torque reflects the actual frictional resistance level of the sealing friction pair. Therefore, the theoretical sensitivity of each core operating condition characteristic parameter to the wear rate is determined by the following relationship: the pressure difference between the upstream and downstream pipeline media has the highest sensitivity by directly affecting the normal load on the contact surface; the frequency of action per unit time has the second highest sensitivity by directly affecting the sliding distance; the real-time temperature of the pipeline medium has a moderate sensitivity by affecting the material hardness; and the output torque, as a comprehensive reflection parameter of frictional resistance, has an auxiliary sensitivity. The higher the theoretical sensitivity of each parameter, the larger the preset value of the corresponding correction coefficient component. For example, under the high-temperature, high-pressure, and high-frequency operating mode, the conservative preset value for the pressure difference between the upstream and downstream pipeline media is set to 0.4 to 0.6; the conservative preset value for the real-time temperature of the pipeline media is set to 0.3 to 0.5; the conservative preset value for the frequency of action per unit time is set to 0.2 to 0.4; and the conservative preset value for the output torque is set to 0.1 to 0.3. If this operating mode has been accumulated in the sensing database... The historical operating data is then used to adjust the correction coefficient vector. Regression fitting pre-training is performed, using the actual values ​​of the core working condition feature parameters of each historical acquisition period as input, and the actual wear rate of that period recorded in the perception database relative to the benchmark wear rate. The relative deviation is the label, and the least squares method is used to calculate it. The optimal estimate replaces the conservative preset value.

[0104] The wear evolution prediction baseline output by the benchmark wear model The working condition deviation correction coefficient output by the working condition deviation correction model Coupling is performed to obtain the operating mode under the specified conditions. Wear evolution prediction trajectory Where N is the cumulative number of actions. The wear evolution prediction trajectory refers to the wear evolution prediction computational model under the given operating condition mode. A prediction sequence generated under constraints, describing the evolution of cumulative wear as a function of cumulative action count.

[0105] Within each prediction cycle, the wear evolution prediction computation model is invoked to perform cascaded calculations. First, the baseline wear model is used to deduce the wear evolution prediction baseline under the preset standard operating conditions. Subsequently, the operating condition deviation correction model calculates the operating condition deviation correction coefficient based on the current operating condition mode identifier and the core operating condition characteristic parameters. ;Will and By coupling, the wear evolution prediction trajectory corresponding to the current operating condition is obtained. The prediction period refers to the time interval between two consecutive calls by the lifetime prediction unit to the wear evolution prediction calculation model to perform prediction calculations. The prediction period is set to an integer multiple of the preset acquisition period, for example, 10 to 100 times the preset acquisition period, in order to achieve a balance between prediction timeliness and computational resource consumption.

[0106] Along the wear evolution prediction trajectory Forward deduction, retrieval enables The first time it is greater than or equal to the maximum allowable wear amount Cumulative number of actions Based on the current cumulative number of actions Starting from this point, the predicted remaining service time is... ,in The average daily action frequency is calculated based on historical operational data of the action frequency per unit time.

[0107] A bootstrap resampling method is used to generate the confidence interval corresponding to the predicted value of the remaining service time. The bootstrap resampling method refers to performing B random resamplings with replacement on a training sample set consisting of complete data records from all historical acquisition cycles extracted from the sensing database. Each resampling draws a sample of the same size as the training sample set to form a bootstrap sample subset. B represents the total number of bootstrap resampling operations, i.e., the number of bootstrap sample subsets generated and the corresponding number of wear evolution prediction model instances. The value of B is determined based on the size of the training sample set; the larger the size of the training sample set, the larger the value of B to ensure the statistical stability of the confidence interval estimation. For example, B is between 200 and 1000. The training sample set refers to the set of all historical cycle data records stored in the sensing database, including the operating condition data, the actuator feedback data, the wear status data, and the corresponding operating condition mode identifiers. Using each of the bootstrap sample subsets, one instance of the wear evolution prediction computation model is independently trained, resulting in a total of B model instances. The B wear evolution prediction computation model instances are then used to generate B predicted values ​​for remaining service time. These B predicted values ​​are arranged in ascending order, and the 2.5th percentile is taken as the lower limit of the confidence interval, and the 97.5th percentile as the upper limit. These upper and lower limits constitute the confidence interval at a 95% confidence level.

[0108] The system aggregates the predicted remaining service time values ​​and confidence intervals of all dynamic seal actuators within the system. The predicted remaining service time values ​​are compared with preset safety thresholds to determine the wear risk type of each dynamic seal actuator node. The preset safety thresholds are tiered thresholds for judging the urgency of wear risk for each dynamic seal actuator, including a first preset safety threshold and a second preset safety threshold, where the first preset safety threshold is lower than the second preset safety threshold. Nodes with predicted remaining service time values ​​lower than the first preset safety threshold are classified as first wear risk type; nodes with predicted remaining service time values ​​between the first and second preset safety thresholds are classified as second wear risk type; and the remaining nodes are classified as third wear risk type, thereby generating a global wear risk distribution result. For example, the first preset safety threshold is set to 15% of the design service life of the dynamic seal actuator, and the second preset safety threshold is set to 40% of the design service life. The design service life refers to the expected cumulative operating time of the dynamic seal actuator from commissioning to the point where the sealing friction pair reaches the maximum permissible wear amount under preset standard operating conditions.

[0109] The overall wear risk distribution results, the predicted remaining service time of each dynamic seal actuator node, the confidence interval, and the wear risk type are synchronously distributed to the collaborative scheduling unit and the control and verification module. If the correction coefficient vector of the operating condition deviation correction model in this prediction cycle... If an update occurs during online incremental training, the updated correction coefficient vector will be persistently stored according to the operating mode.

[0110] The collaborative scheduling unit is used to receive the remaining service time prediction value, confidence interval and wear risk type of each dynamic seal actuator node output by the life prediction unit, and the working condition mode identifier of each dynamic seal actuator node output by the working condition identification unit. Based on this, it constructs a multi-layer constraint condition set including total system load constraint, safe operation boundary constraint and additional constraint of prediction uncertainty. With minimizing the real-time wear rate distribution dispersion of each dynamic seal actuator in the whole system as the optimization objective, it solves the load allocation ratio of each dynamic seal actuator node. The solution result is encapsulated into a multi-node load scheduling command and sent to the control and verification module. After receiving the global wear balance effect deviation data returned by the control and verification module, it iteratively corrects the scheduling strategy weight parameters.

[0111] Furthermore, the execution steps of the coordinated scheduling unit include:

[0112] The system receives the global wear risk distribution results output by the life prediction unit, the predicted remaining service time of each dynamic seal actuator node, the confidence interval and the wear risk type, and the current operating condition mode identifier of each dynamic seal actuator node output by the operating condition identification unit.

[0113] Construct a set of constraints, which includes total system load constraints, safe operation boundary constraints, and additional constraints for prediction uncertainties.

[0114] The total system load constraint is represented in data form as follows:

[0115]

[0116] in This represents the total number of nodes in the dynamic seal actuator. For the first The current upstream and downstream pipeline medium pressure difference at each node For the first The distribution ratio of throughput to each node This represents the minimum total flow capacity requirement that the industrial fluid pipeline must maintain under given process conditions. The flow allocation ratio is... satisfy and .

[0117] The safety operation boundary constraints include the maximum permissible drive torque, maximum permissible operating speed, and maximum permissible preload compensation stroke for each dynamic seal actuator node. The maximum permissible drive torque is determined by the smaller of the rated output torque of the actuator motor and the rated strength of the transmission mechanism; the maximum permissible operating speed is calculated and determined by the maximum speed of the actuator drive motor and the mechanical transmission ratio; the maximum permissible preload compensation stroke is determined by the smaller of the elastic deformation range of the sealing friction pair and the mechanical stroke of the preload mechanism.

[0118] The aforementioned additional constraint on predictive uncertainty refers to applying additional safety margins of varying strengths to the safe operating boundary of each dynamic seal actuator node based on its wear risk type. Specifically, for nodes belonging to the first wear risk type, the limits related to actuation frequency and bearing pressure difference in their safe operating boundary constraints are multiplied by a first compression factor. For nodes belonging to the second wear risk type, multiply by the second compression factor. ,in For nodes belonging to the third wear risk type, no additional constraints are applied. The first compression coefficient... and the second compression coefficient The value is also dynamically adjusted based on the width of the confidence interval for each dynamic seal actuator node. The larger the normalized width of the confidence interval, the smaller the contraction coefficient, and the stricter the constraint. For example, for nodes belonging to the first wear risk type... The maximum frequency of actions per unit period after adjustment for uncertainty-related constraints is: ,in This is the upper limit of the maximum frequency of actions per unit cycle in the original safe operation boundary constraints of this node. It is the ratio of the width of the confidence interval of the node to the predicted value of the remaining service time. A preset risk penalty coefficient is used. The preset risk penalty coefficient... It is a globally adjustable parameter that controls the degree to which forecast uncertainty reduces the safety margin, and its setting is based on: when If the value is too small, the width of the confidence interval is insufficient to reduce the safe operating boundary, and high-risk nodes cannot be effectively protected; when... When the value is too large, even a moderate level of prediction uncertainty will excessively tighten the safe operating boundary, leading to a reduction in available system nodes and a decrease in overall throughput capacity; therefore The value of must strike a balance between sufficient safety margin and system availability. For example, when the normalized width of the confidence interval... When it is 0.5, A reduction ratio of 0.3 to 0.5 corresponds to a reduction ratio of 15% to 25%. This reduction ratio can ensure that high-risk nodes receive additional protection without excessively sacrificing the system's throughput capacity.

[0119] With minimizing the standard deviation of the real-time wear rate of each dynamic seal actuator as the optimization objective, the following objective function-constrained optimization problem is established:

[0120]

[0121] in For the first The current real-time wear rate of each node. , which is the average real-time wear rate of the entire system.

[0122] The decision variables are the distribution ratio of bearing pressure difference and the distribution ratio of flow rate at each dynamic seal actuator node. and the maximum frequency of actions per unit period .

[0123] A sequential quadratic programming algorithm is used to solve the aforementioned objective function-constrained optimization problem. This algorithm approximates the original nonlinear constraint optimization problem as a quadratic programming subproblem at the current iteration point, and iteratively solves a series of these subproblems to gradually approach the optimal solution of the original problem. Specifically, the solution process is as follows:

[0124] The algorithm starts with the current upstream and downstream pipeline medium pressure difference, flow ratio allocation, and maximum unit cycle frequency limit for each node. In each iteration, the gradient of the objective function and the approximation of the Hessian matrix at the current iteration point are calculated. The system total load constraint, safe operation boundary constraint, and prediction uncertainty additional constraint are linearized at the current iteration point to construct a quadratic programming subproblem. Solving this quadratic programming subproblem yields the search direction and step size, which are used to update the decision variables. The iteration is repeated until the change in the objective function value is below a preset convergence threshold for two consecutive iterations or reaches a preset maximum number of iterations. The final convergence point's decision variable value is output as the update value for the load-bearing pressure difference allocation, flow ratio allocation, and maximum unit cycle frequency limit for each node. The preset convergence threshold refers to the lower limit of the change in the objective function value used to determine the convergence of the sequential quadratic programming algorithm. The algorithm is considered convergent when the change in the objective function value is below this threshold for two consecutive iterations. For example, it is set to 10. -6 The preset maximum number of iterations refers to the maximum number of iterations allowed by the sequential quadratic programming algorithm. When the iteration count reaches this value and convergence is still not achieved, the iteration terminates and the current optimal solution is output. For example, this is set to 100 to 200 iterations. The scheduling strategy weight parameters refer to the set of adjustable parameters used by the collaborative scheduling unit in the load allocation solution process, including the first compaction coefficient. The second compression coefficient and the preset risk penalty coefficient The above parameters use preset values ​​during initial operation. After receiving the global wear balance effect deviation data returned by the control and verification module, the parameters are iteratively corrected according to the direction and magnitude of the deviation.

[0125] The solution results are encapsulated into multi-node load scheduling instructions. These instructions include the differential pressure distribution ratio, flow rate distribution ratio, maximum operating frequency limit per unit cycle, and operating mode identifier for each dynamic seal actuator node. These instructions are then sent to the corresponding control parameter calculation unit of each dynamic seal actuator in the control verification module. The multi-node load scheduling instructions constitute the upper-level constraint boundary for the corresponding node control parameter calculation.

[0126] Specifically, the decision-making and computation module is designed to address two major structural deficiencies in existing technologies: the lack of closed-loop linkage between wear prediction and proactive control, and the disconnect between single-device management and system-level load scheduling. This module cascades and couples a physical wear mechanism model of the sealing friction pair with a data-driven operating condition deviation correction model. This allows wear evolution prediction to be based on physical mechanisms and corrected using actual operating condition data. The resulting remaining service life prediction directly drives system-level collaborative scheduling, creating a causal link between prediction and scheduling rather than isolation. Simultaneously, the collaborative scheduling unit aims for a balanced wear rate distribution across the entire system. It quantifies the prediction uncertainty of each dynamic seal actuator node as an additional constraint in the scheduling solution, directly linking system-level load allocation with the wear state of individual nodes, thus resolving the problem of uneven system wear caused by independent operation of each device.

[0127] The control and verification module is used to receive the multi-node load scheduling instructions and the remaining service time prediction results of each dynamic seal actuator output by the decision calculation module. Within the constraint boundary formed by the multi-node load scheduling instructions, it determines the control demand level based on the current cumulative wear of each dynamic seal actuator node and activates the calculation of hierarchical control parameters level by level. The calculated hierarchical control parameters are sent out for execution and the execution results are collected and fed back. At the same time, based on the remaining service time prediction value and the confidence interval, it determines the urgency of the operation and maintenance of each dynamic seal actuator node and generates hierarchical operation and maintenance scheduling instructions. The deviation data between the single-node control effect and the global wear balance effect after the control execution is sent back to the collaborative scheduling unit in the decision calculation module to drive the iterative correction of the collaborative scheduling strategy and prediction model.

[0128] Please see Figure 4 The control verification module includes a control parameter calculation unit and an execution interaction unit.

[0129] The control parameter calculation unit is used to determine the control demand level based on the current cumulative wear of the dynamic seal actuator node within the constraint boundary formed by the multi-node load scheduling instructions issued by the collaborative scheduling unit, and select an appropriate control parameter mapping model in combination with the working condition mode identifier extracted from the multi-node load scheduling instructions. The unit then activates and calculates the hierarchical control parameters step by step according to the three progressive control demand levels of mild, moderate and severe, and sends them to the execution interaction unit for execution.

[0130] Furthermore, the execution steps of the control parameter calculation unit include:

[0131] The system receives the multi-node load scheduling command for the dynamic seal actuator node issued by the collaborative scheduling unit, and extracts the following parameters: the differential pressure distribution ratio, the flow proportion distribution ratio, the maximum upper limit of the number of actions per unit cycle, and the operating mode identifier. The differential pressure distribution ratio, the flow proportion distribution ratio, and the maximum upper limit of the number of actions per unit cycle constitute constraint boundaries. Specifically, the differential pressure distribution ratio, as the absolute value limit of the maximum upstream and downstream pipeline medium pressure difference borne by this node, defines the upper limit of the differential pressure that the dynamic seal actuator can withstand in actual operation; the flow proportion distribution ratio, as the flow share that this node must bear in the total flow demand of the system, defines the proportion of fluid flow allocated to this dynamic seal actuator in the parallel branches of the pipeline; and the maximum upper limit of the number of actions per unit cycle, as the maximum number of opening and closing actions that this node is allowed to perform within a unit statistical time window. These three parameters together constitute the operating boundaries of the dynamic seal actuator node in terms of load-bearing capacity, flow share, and operating frequency.

[0132] The system receives the predicted remaining service time and cumulative wear of the dynamic seal actuator node from the life prediction unit. Based on the ratio of the cumulative wear to the maximum allowable wear, the system determines the current control requirement level of the dynamic seal actuator node. The control requirement level includes a light control level, a medium control level, and a heavy control level. For example, when the cumulative wear does not exceed 30% of the maximum allowable wear, it is a light control level; when it exceeds 30% but does not exceed 60%, it is a medium control level; and when it exceeds 60%, it is a heavy control level.

[0133] Based on the combination of the current control demand level and the operating condition mode identifier extracted from the scheduling instruction, the corresponding control parameter mapping model is retrieved. The control parameter mapping model internally stores the reference mapping coefficients of each layer of control parameters under different combinations of control demand levels and operating conditions. These reference mapping coefficients differ under different operating conditions. The reference mapping coefficient refers to the scaling ratio of each layer of control parameters relative to the corresponding reference value under a specific control demand level and operating condition mode combination. It is pre-calibrated using wear-sensitive characteristic test data of the sealing friction pair and motion characteristic data of the actuator. The reference mapping coefficients are different for different control demand levels and operating conditions. For example, under high temperature, high pressure, and high frequency operating conditions, the reference mapping coefficient for the same control demand level is higher than the corresponding reference mapping coefficient under normal temperature, normal pressure, and low frequency operating conditions, in order to apply stronger motion characteristic constraints under harsh operating conditions; under the same operating condition mode, the reference mapping coefficient for a heavy control level is higher than the corresponding reference mapping coefficient for a light control level, in order to apply stricter protective restrictions when wear increases.

[0134] The control parameter calculation unit activates the three-layer control parameter calculation step by step according to the control demand level to obtain the layered control parameters, specifically including:

[0135] When the control demand level reaches the mild control level or above, the first layer of control parameter calculation is activated. The single opening and closing action of the opening and closing drive unit is decomposed into three continuous intervals: an initial acceleration phase, an intermediate constant speed phase, and a final deceleration phase. Based on the combination of the current control demand level and the operating mode identifier, the reference mapping coefficients for each phase are retrieved from the control parameter mapping model. The corresponding reference mapping coefficients are multiplied by the maximum allowable operating speed to obtain the target speed value for each phase. The speed change rate between adjacent phases is limited to below a preset smoothness threshold. The preset smoothness threshold is based on 40% to 60% of the acceleration corresponding to the maximum allowable operating speed, and the acceleration slope is further reduced under high pressure differential conditions to reduce the starting torque impact. The segmented target speed values ​​of the opening and closing drive unit and the speed change rate limit between each phase are output. The opening and closing drive subunit controls the speed curve of the drive motor according to the above parameters, replacing abrupt speed changes with smooth speed transitions to reduce the impact load on the sealing friction pair during the transient process of operation.

[0136] When the regulation demand level reaches the moderate regulation level or above, the second layer of control parameter calculation is activated on the basis of the first layer of control parameters. The second layer of control parameter calculation is used to perform hierarchical correction of the operating constraints of the dynamic seal actuator within the constraint boundary formed by the multi-node load scheduling command, based on the regulation demand level at which the cumulative wear amount is located. This causes the action permission conditions to gradually narrow as the cumulative wear amount increases, so as to avoid the sealing friction pair triggering accelerated wear effect under adverse working conditions. Based on the combination of the current regulation demand level and the working condition mode identifier, the reference mapping coefficients of each constraint item are retrieved from the regulation parameter mapping model to correct the constraint boundary, and the operating constraints are calculated item by item. Specifically, one is to multiply the high pressure differential action trigger condition, that is, the upper limit of the upstream and downstream pipeline medium pressure difference required for the opening and closing drive subunit to perform the opening and closing action, by... The trigger threshold is determined by the corresponding reference mapping coefficient, which is exemplarily set to 30% of the rated pressure level of the dynamic seal actuator. Secondly, the maximum operating frequency limit per unit cycle is obtained by multiplying the maximum operating frequency limit per unit cycle by the corresponding reference mapping coefficient. Under the medium control level, the reference mapping coefficient is exemplarily set to 0.7 to 0.85. Thirdly, the minimum interval time for continuous operation is determined by multiplying the temperature recovery characteristic time constant of the sealing friction pair after operation by the corresponding reference mapping coefficient. The high pressure differential operation trigger condition threshold, the maximum operating frequency limit per unit cycle, and the minimum interval time for continuous operation are output. The above three operating constraints together limit the permissible range of operation of the dynamic seal actuator under harsh working conditions.

[0137] When the regulation demand level reaches the severe regulation level and the corresponding dynamic seal actuator is equipped with the active pre-tightening drive subunit, the third layer of control parameters is activated by superimposing the first two layers of control parameters. The third-layer control parameter calculation is used to calculate the additional preload displacement that the active preload drive subunit should apply along the normal direction of the sealing surface when the cumulative wear causes the effective contact pressure of the sealing friction pair to drop to a level where it can no longer maintain sealing reliability through its own elastic deformation. This is to compensate online for the increase in sealing gap caused by accumulated wear. Using the cumulative wear as the input variable, the reference mapping coefficient of the sealing surface geometric compensation coefficient is retrieved from the control parameter mapping model based on the combination of the current control demand level and the operating mode identifier. The sealing gap compensation stroke is calculated according to the relationship: compensation stroke = cumulative wear × sealing surface geometric compensation coefficient × reference mapping coefficient. The sealing surface geometric compensation coefficient is pre-calibrated at the factory according to the geometric configuration of the sealing friction pair. The calculation of the compensation stroke must be constrained by the maximum allowable preload compensation stroke and the calculation result must not exceed this upper limit. The sealing gap compensation stroke value is output, and the active preload drive subunit applies the preload displacement of the compensation stroke value along the normal direction of the sealing surface to restore the effective contact pressure of the sealing friction pair caused by the cumulative wear.

[0138] The execution interaction unit is used to receive the hierarchical control parameters issued by the control parameter calculation unit, drive the actuators of each dynamic seal actuator to complete the opening and closing action or the pre-tightening compensation action and feed back the execution result. At the same time, based on the remaining service time prediction value and confidence interval of each dynamic seal actuator output by the life prediction unit, it determines the urgency of operation and maintenance of each dynamic seal actuator node and generates hierarchical operation and maintenance scheduling instructions to be uploaded to the remote monitoring terminal.

[0139] The execution interaction unit includes an on / off drive subunit, an active pre-tightening drive subunit, and a remote monitoring terminal.

[0140] Furthermore, the execution steps of the interactive unit include:

[0141] The system receives the first-level and second-level control parameters from the control parameter calculation unit, and analyzes the segmented target speed values, speed change rate limits for each stage, high pressure differential action triggering conditions, maximum action frequency limit per unit cycle, and minimum interval time for continuous action. The analyzed action characteristic parameters and action constraint parameters are then sent to the corresponding opening / closing drive subunits of each dynamic seal actuator. These subunits control the drive motor's speed curve according to the segmented target speed values, executing opening / closing actions within the range that satisfies the action constraint parameters. After each action, the opening / closing drive subunit sends back feedback signals of the actual output torque and actual action stroke.

[0142] If the current control requirement level is a severe control level and the dynamic sealing actuator is equipped with an active pre-tightening drive subunit, then it receives the third-level control parameters sent by the control parameter calculation unit, analyzes the sealing gap compensation stroke value therein, and sends it to the active pre-tightening drive subunit for execution. The active pre-tightening drive subunit loads the pre-tightening displacement of the sealing gap compensation stroke value along the normal direction of the sealing surface, and after loading is completed, it sends back the actual pre-tightening displacement feedback signal.

[0143] After the control is executed, the cumulative wear amount and real-time wear rate of the first acquisition cycle after the control are obtained from the sensing and acquisition module. The real-time wear rate after the control is compared with the real-time wear rate before the control, and the single-node control effect deviation of this dynamic seal actuator node is calculated. This refers to the change in the real-time wear rate after adjustment relative to the real-time wear rate before adjustment; summarizing the changes at each adjustment node. Calculate the deviation of the global wear equalization effect. ,in and These represent the standard deviations of the real-time wear rate of the entire system before and after the control execution. The deviation of the single-node control effect is then... Deviation from the global wear equalization effect The data is sent back to the collaborative scheduling unit in the decision-making module to drive the iterative correction of the scheduling strategy weight parameters.

[0144] The system receives the predicted remaining service time and confidence interval of each dynamic seal actuator node from the life prediction unit, determines the urgency of maintenance for each dynamic seal actuator node, and generates a tiered maintenance scheduling instruction. The generation rule for the tiered maintenance scheduling instruction is as follows: nodes whose predicted remaining service time is lower than a first maintenance threshold and whose upper limit of the confidence interval of the predicted remaining service time is lower than a second maintenance threshold are marked as requiring emergency shutdown, generating an emergency response instruction containing a node identifier and an emergency shutdown instruction; nodes whose predicted remaining service time is lower than the first maintenance threshold but do not meet the emergency shutdown conditions are marked as requiring planned maintenance, generating a planned maintenance instruction containing a node identifier, a recommended maintenance time window, and the predicted remaining service time. The recommended maintenance time window is expanded by a preset proportion before and after the predicted remaining service time, for example, by 20% of the predicted remaining service time; the remaining nodes are marked as requiring continuous monitoring. The first maintenance threshold is exemplarily set to 10% of the designed service life of the dynamic seal actuator, and the second maintenance threshold is exemplarily set to 5% of the designed service life. The hierarchical maintenance scheduling instructions are uploaded to the remote monitoring terminal for centralized display.

[0145] The remote monitoring terminal, deployed in a centralized monitoring center or on a mobile monitoring device physically separated from the industrial site, receives and centrally displays the following information via an industrial Ethernet network: hierarchical operation and maintenance scheduling instructions, with different colors indicating different handling categories; green indicates continuous monitoring handling nodes, yellow indicates planned maintenance handling nodes, and red indicates emergency shutdown handling nodes, displaying the predicted remaining service time, confidence interval, and recommended maintenance time window for each dynamic seal actuator; the overall wear risk distribution results, displayed as a heat map showing the wear risk level distribution of all dynamic seal actuators in the system; the multi-node load scheduling instructions, displayed in a table format as the current load differential pressure distribution ratio, flow rate distribution ratio, and maximum maximum action frequency limit per unit cycle for each node; and the sensor anomaly warning signal, where an alarm icon is superimposed on the corresponding device icon when a sensor at a certain node is determined to have abnormal drift.

[0146] Specifically, the control and verification module is designed to address the structural deficiency in existing technologies where the lack of a unified long-term iterative benchmark leads to a continuous decline in control accuracy. This module achieves closed-loop correction by calculating and implementing hierarchical control parameters, providing execution feedback, and analyzing operation and maintenance scheduling. This is done through a two-dimensional deviation calculation (single node and global) and transmitting the actual effect of the control execution back to the collaborative scheduling unit, thus upgrading operation and maintenance scheduling from passive fault triggering to proactive lifespan prediction-driven operation. This maintains the long-term stability of the control benchmark and the reliability of decision-making throughout the entire lifespan of the dynamic seal actuator.

[0147] Example 2

[0148] This embodiment introduces a life prediction-driven collaborative control method for industrial dynamic seal actuators, including:

[0149] Step S1: At the physical sensing front-end level of the collaborative control method, a synchronous trigger signal is sent to the data acquisition channels of all dynamic sealing actuator nodes according to a preset acquisition cycle. Wear state data reflecting the physical state changes of the sealing friction pair, actuator feedback data reflecting the action state of the actuator, and operating condition data reflecting the pipeline fluid operating conditions are collected synchronously. Outlier removal processing is performed on all collected data according to a preset physical rationality range, and time alignment processing is performed according to a preset time alignment tolerance. The current acquisition cycle data after the above processing is distributed to subsequent steps. At the same time, all operating condition data, actuator feedback data, and wear state data are continuously accumulated to form a sensing database, providing a continuously expanding training data foundation for the online incremental training of the subsequent life prediction model.

[0150] Step S2: Extract the pressure difference of the upstream and downstream pipeline media, the real-time temperature of the pipeline media, the frequency of action per unit time, and the output torque of the opening and closing drive unit from the perception database constructed in Step S1 as core operating condition characteristic parameters. Based on the rated design parameters of each dynamic seal actuator and the wear sensitivity characteristics of the sealing friction pair, configure a corresponding operating condition threshold for each core operating condition characteristic parameter. Compare the core operating condition characteristic parameters with the corresponding operating condition thresholds respectively. Determine the current operating condition mode of each dynamic seal actuator based on the combination matching result of whether each core operating condition characteristic parameter exceeds the corresponding operating condition threshold and output the corresponding operating condition mode identifier. The operating condition mode identifier is provided as an operating condition constraint to the subsequent life prediction step and collaborative scheduling step, so that the above decision-making links can make adaptive adjustments based on the current operating conditions when they are executed.

[0151] Step S3: Obtain the metal material performance parameters and sealing surface geometric parameters of the sealing friction pair. Based on the Archard wear model, combine the metal material performance parameters and sealing surface geometric parameters to calculate the reference wear rate and maximum allowable wear amount of the sealing friction pair. Obtain the material wear life calibration curve of each sealing friction pair material obtained through laboratory wear tests under preset standard working conditions. Based on the reference wear rate, the maximum allowable wear amount, and the material wear life calibration curve, construct a wear evolution prediction calculation model with physical wear mechanism as the framework and actual working condition data as the deviation correction basis. Call the wear evolution prediction calculation model in each prediction cycle. By combining the current operating condition mode identifier and core operating condition characteristic parameters, cascade calculations are performed to extrapolate along the generated wear evolution prediction trajectory to determine the predicted remaining service time of each dynamic seal actuator. A self-lifting resampling method is used to generate the confidence interval corresponding to the predicted remaining service time. The predicted remaining service time of all dynamic seal actuators is compared with a preset safety threshold to determine the wear risk type of each dynamic seal actuator node and generate a global wear risk distribution result. The global wear risk distribution result, the predicted remaining service time of each dynamic seal actuator node, the confidence interval, and the wear risk type are synchronously distributed to subsequent collaborative scheduling steps and control verification steps.

[0152] Step S4: Receive the predicted remaining service time, confidence interval, and wear risk type of each dynamic seal actuator node output from Step S3, and the operating condition mode identifier of each dynamic seal actuator node output from Step S2. Based on this, construct a multi-layer constraint set including total system load constraints, safe operation boundary constraints, and additional constraints for prediction uncertainties. With minimizing the real-time wear rate distribution dispersion of each dynamic seal actuator in the entire system as the optimization objective, use a sequential quadratic programming algorithm to solve for the load allocation ratio of each dynamic seal actuator node. Encapsulate the solution results into a multi-node load scheduling command containing the load differential pressure allocation ratio, flow ratio allocation ratio, maximum operating frequency limit per unit cycle, and operating condition mode identifier of each dynamic seal actuator node, and send it to subsequent control steps. After receiving the global wear balance effect deviation data returned from subsequent steps, iteratively correct the scheduling strategy weight parameters.

[0153] Step S5: Receive the multi-node load scheduling command for the dynamic seal actuator node issued in Step S4, extract the load differential pressure distribution ratio, flow ratio distribution ratio, maximum operating frequency limit per unit cycle, and operating mode identifier to form constraint boundaries. Simultaneously, receive the predicted remaining service time and cumulative wear of the dynamic seal actuator node output in Step S3. Determine the current control demand level of the dynamic seal actuator node based on the ratio of the cumulative wear to the maximum allowable wear. The control demand level includes mild control, moderate control, and severe control. Based on the combination of the current control demand level and the operating mode identifier extracted from the scheduling command, retrieve the wear sensitivity characteristic test data of the sealing friction pair and the actuator... The pre-calibrated control parameter mapping model of the structure's motion characteristic data is activated step by step according to the control requirement level, and the three-layer control parameter calculation is activated. The first layer control parameter calculation is activated for mild control level and above to output the segmented target speed value of the opening and closing drive unit and the speed change rate limit between each stage. The second layer control parameter calculation is activated on the basis of the previous layer for moderate control level and above to output the high pressure differential action trigger condition threshold, the maximum action frequency limit value per unit cycle and the minimum interval time of continuous action. When the control level is severe and the dynamic seal actuator is equipped with an active pre-tightening drive sub-unit, the third layer control parameter calculation is activated on the basis of the first two layers to output the sealing gap compensation stroke value. The calculated layered control parameters are then sent to the subsequent execution steps.

[0154] Step S6: Receive the hierarchical control parameters issued in Step S5, drive the opening and closing drive subunits of each dynamic seal actuator to perform opening and closing actions according to the segmented target speed value and action constraint parameters, and drive the active pre-tightening drive subunit to load the pre-tightening displacement of the sealing gap compensation stroke value along the normal direction of the sealing surface and feed back the execution results. After the control execution is completed, obtain the cumulative wear amount and real-time wear rate of the first acquisition cycle after control from Step S1, calculate the single-node control effect deviation and global wear balance effect deviation of this dynamic seal actuator node and send them back to Step S4 to drive the iterative correction of the scheduling strategy weight parameters. At the same time, based on the predicted value of the remaining service time of each dynamic seal actuator output in Step S3 and the confidence interval, determine the urgency of the operation and maintenance of each dynamic seal actuator node, generate hierarchical operation and maintenance scheduling instructions including emergency shutdown, planned maintenance and continuous monitoring, and upload them to the remote monitoring terminal for centralized display.

[0155] Working principle and its effects:

[0156] This invention addresses the industry pain points of difficult wear control of multi-node dynamic seal actuators in industrial fluid pipeline systems, short overall system service life, and passive operation and maintenance decisions by carrying out full life cycle wear leveling control.

[0157] The sensing and acquisition module continuously collects and manages multi-source heterogeneous operating data synchronously to build a sensing database. The operating condition identification unit in the decision-making and computation module identifies the operating condition modes of each dynamic seal actuator in real time. The life prediction unit calls the wear evolution prediction model to generate the predicted remaining service time and confidence interval for each dynamic seal actuator, and determines the wear risk type of each node accordingly, generating a global wear risk distribution result. The collaborative scheduling unit, with the optimization objective of minimizing the dispersion of the real-time wear rate distribution across the entire system, solves the load allocation scheme for each node under multiple constraints. The control parameter calculation unit in the control and verification module calculates the control parameter in the multi-node load scheduling... Under the rigid constraints of the command, the three-layer control parameters are activated and calculated step by step according to the mild control level, the moderate control level, and the severe control level, and then sent to the execution interaction unit for execution. The execution interaction unit drives the start-stop drive subunit and the active pre-tightening drive subunit to complete the control action, collects the execution results for feedback, and calculates the single-node control effect deviation and the global wear balance effect deviation and sends them back to the collaborative scheduling unit. Simultaneously, iterative correction of the wear evolution prediction calculation model and the scheduling strategy weight parameters is triggered, forming a complete closed loop of prediction, scheduling, control, verification and correction of prediction. At the same time, the execution interaction unit generates the hierarchical operation and maintenance scheduling command based on the remaining service time prediction value and displays it through the remote monitoring terminal.

[0158] In summary, this invention, through a three-module architecture and a multi-unit progressive decision-making chain, organically integrates wear perception, operating condition identification, life prediction, global scheduling, hierarchical control, effect verification, operation and maintenance decision-making, and benchmark self-calibration into a closed-loop collaborative system. This effectively balances wear control accuracy with overall system reliability and can adapt to the full life cycle collaborative management needs of multi-node dynamic seal actuators in industrial fluid pipeline systems of different scales.

[0159] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A life-prediction-driven, industrial dynamic sealing actuator co-control system, characterized in that, include: The sensing and acquisition module is used to collect operating data, actuator feedback data, and wear status data, and to perform data preprocessing to build a sensing database; The decision-making and computing module is used to identify the operating mode of each dynamic seal actuator based on the perception database and construct the wear evolution prediction trajectory to predict the remaining service life of each dynamic seal actuator. And with the goal of balancing the real-time wear rate distribution of each dynamic seal actuator, multi-node load collaborative allocation and scheduling is performed to generate multi-node load scheduling instructions; The control and verification module is used to determine the control demand level based on the current cumulative wear of each dynamic seal actuator node within the constraint boundary formed by the multi-node load scheduling command, and to activate the calculation of the hierarchical control parameters step by step. The hierarchical control parameters are then issued for execution, and the urgency of the operation and maintenance of each dynamic seal actuator node is determined to generate hierarchical operation and maintenance scheduling commands.

2. The life prediction drive coordinated control system for industrial dynamic seal actuators of claim 1, wherein, The decision computation module includes: The working condition identification unit is used to identify the current working condition mode of each dynamic seal actuator based on the core working condition feature parameters in the perception database. The life prediction unit is used to construct the wear evolution prediction trajectory for each working condition, generate the remaining service time prediction value and corresponding confidence interval for each dynamic seal actuator, and determine the wear risk type of each dynamic seal actuator node. The collaborative scheduling unit is used to construct a multi-layer constraint set, with the optimization objective of minimizing the dispersion of the real-time wear rate distribution of each dynamic seal actuator. It solves the load allocation ratio of each dynamic seal actuator node and generates multi-node load scheduling instructions.

3. The life-predicted, drive-based, industrial dynamic seal actuator co- control system of claim 2, wherein, The steps for constructing the wear evolution prediction trajectory for each operating condition mode include: Based on the Archard wear model, combined with the performance parameters of the metal material and the geometric parameters of the sealing surface, the reference wear rate and maximum allowable wear of the sealing friction pair are calculated; and the material wear life calibration curve under preset standard working conditions is obtained. Based on the Archard wear model, the reference wear rate is substituted to obtain a linear prediction relationship between the cumulative wear amount and the cumulative number of actions under the preset standard working conditions. Using the material wear life calibration curve as a physical constraint, the linear prediction relationship is piecewise corrected to form a corrected benchmark wear prediction function, which serves as the baseline for wear evolution prediction. A mapping function is established, which takes the working condition mode identifier and core working condition characteristic parameters as input and the working condition deviation correction coefficient as output, as the working condition deviation correction model. The wear evolution prediction baseline is coupled with the working condition deviation correction coefficient output by the working condition deviation correction model to obtain the wear evolution prediction trajectory for each working condition mode.

4. The life prediction-driven industrial dynamic seal actuator collaborative control system as described in claim 3, characterized in that, The process involves generating predicted remaining service time values ​​and corresponding confidence intervals for each dynamic seal actuator, and determining the wear risk type of each dynamic seal actuator node, including: Receive the current cumulative wear amount and real-time wear rate, as well as the operating mode identifier; Within each prediction cycle, the baseline for predicting wear evolution under a preset standard working condition is derived. The working condition deviation correction model calculates the working condition deviation correction coefficient based on the current working condition mode identifier and core working condition characteristic parameters, and then couples it to obtain the wear evolution prediction trajectory. By extrapolating along the wear evolution prediction trajectory, the cumulative number of actions that first made the wear evolution prediction trajectory greater than or equal to the maximum allowable wear amount is retrieved. Starting from the current cumulative number of actions, the predicted value of the remaining service time is calculated. The confidence intervals corresponding to the predicted remaining service time are generated using a bootstrap resampling method. The predicted remaining service life and confidence intervals of all dynamic seal actuators are summarized, and the predicted remaining service life are compared with the preset safety threshold to determine the wear risk type of each dynamic seal actuator node.

5. The life prediction-driven industrial dynamic seal actuator collaborative control system as described in claim 2, characterized in that, The step of generating multi-node load scheduling instructions includes: Receive the predicted remaining service time, confidence interval, and wear risk type of each dynamic seal actuator node, as well as the current operating mode identifier; Construct a set of constraints, which includes total system load constraints, safe operating boundary constraints, and additional constraints for prediction uncertainties. The objective function constrained optimization problem is established with the goal of minimizing the standard deviation of the real-time wear rate of each dynamic seal actuator. The decision variables are the load differential pressure distribution ratio, the flow ratio distribution ratio, and the upper limit of the maximum operating frequency per unit cycle for each dynamic seal actuator node. The objective function constraint optimization problem is solved, and the decision variable values ​​of the convergence point are output as the update values ​​of the bearing pressure difference distribution ratio, flow ratio distribution ratio and maximum action frequency limit per unit cycle for each dynamic seal actuator node, and encapsulated as a multi-node load scheduling instruction.

6. The life prediction-driven industrial dynamic seal actuator collaborative control system as described in claim 2, characterized in that, The identification of the current operating mode of each dynamic seal actuator includes: Receive the periodic sensing dataset of the current acquisition cycle from the sensing database, and extract the core working condition feature parameters from the periodic sensing dataset; Configure corresponding operating condition thresholds for each core operating condition feature parameter, compare each core operating condition feature parameter with its corresponding operating condition threshold, determine the operating condition mode based on the combination matching result of whether each core operating condition feature parameter exceeds the corresponding operating condition threshold, and output the operating condition mode identifier.

7. The life prediction-driven industrial dynamic seal actuator collaborative control system as described in claim 1, characterized in that, The control and verification module includes: The control parameter calculation unit is used to determine the control demand level based on the cumulative wear amount within the constraint boundary formed by the multi-node load scheduling instructions, select the control parameter mapping model in combination with the operating mode identifier, activate and calculate the hierarchical control parameters level by level according to the control demand level, and send the hierarchical control parameters to the execution interaction unit for execution. The execution interaction unit is used to receive hierarchical control parameters, drive the actuators of each dynamic seal actuator to complete the action, and at the same time, based on the predicted value of the remaining service time of each dynamic seal actuator and the confidence interval, determine the urgency of the operation and maintenance of each dynamic seal actuator node and generate hierarchical operation and maintenance scheduling instructions.

8. The life prediction-driven industrial dynamic seal actuator collaborative control system as described in claim 7, characterized in that, The steps for calculating the hierarchical control parameters include: Receive multi-node load scheduling instructions and extract their load differential pressure distribution ratio, current ratio distribution ratio, maximum action frequency limit per unit cycle, and operating mode identifier; The bearing pressure difference distribution ratio, the flow rate distribution ratio, and the upper limit of the maximum frequency of action per unit cycle constitute the constraint boundary; The system receives the predicted remaining service time and cumulative wear of the dynamic seal actuator node, and determines the current control demand level of the dynamic seal actuator node based on the ratio of the cumulative wear to the maximum allowable wear. The control demand level includes mild control level, moderate control level, and severe control level. Based on the combination of the current control demand level and the operating mode identifier, the corresponding control parameter mapping model is retrieved; the control parameter mapping model is used to store the reference mapping coefficients of each layer of control parameters under different control demand levels and different operating mode combinations. By combining the benchmark mapping coefficients and calculating the control parameters step by step according to the level of control demand, the hierarchical control parameters are obtained.

9. The life prediction-driven industrial dynamic seal actuator collaborative control system as described in claim 8, characterized in that, The process involves combining the benchmark mapping coefficients and, according to the level of control demand, activating control parameters level by level to obtain tiered control parameters, including: When the level of regulation demand reaches the level of mild regulation or above, the first layer of control parameter calculation is activated to decompose the single opening and closing action of the opening and closing drive unit into an initial acceleration stage, an intermediate constant speed stage and a final deceleration stage. Combined with the reference mapping coefficient, the segmented target speed value of the opening and closing drive unit and the speed change rate limit between each stage are output. When the regulation demand level reaches the medium regulation level or above, the second layer of control parameter calculation is activated on the basis of the first layer of control parameter calculation. This is used to correct the constraint boundary within the constraint boundary formed by the multi-node load scheduling instructions according to the benchmark mapping coefficient, and calculate the operating constraint conditions item by item. When the regulation demand level reaches the severe regulation level, the third layer of control parameter calculation is superimposed and activated, and the sealing gap compensation stroke that the active preload drive subunit should apply along the normal direction of the sealing surface is calculated in combination with the reference mapping coefficient.

10. A life prediction-driven collaborative control method for industrial dynamic seal actuators, applied to the life prediction-driven collaborative control system for industrial dynamic seal actuators according to any one of claims 1-9, characterized in that, include: Collect operating data, actuator feedback data, and wear status data, and perform data preprocessing to build a perception database; Identify the operating modes of each dynamic seal actuator based on a perception database; Construct wear evolution prediction trajectories to predict the remaining service life of each dynamic seal actuator; With the goal of balancing the real-time wear rate distribution of each dynamic seal actuator, multi-node load collaborative allocation and scheduling is performed to generate multi-node load scheduling instructions; Within the constraint boundary formed by the multi-node load scheduling command, the control demand level is determined based on the current cumulative wear of each dynamic seal actuator node, and the solution of the hierarchical control parameters is activated step by step. The hierarchical control parameters are issued and executed, and hierarchical operation and maintenance scheduling instructions are generated based on the urgency of operation and maintenance of each dynamic seal actuator node.