Optical fiber sensing array layout method for hydrogen storage bottle curing, process control method and system

Abstract

This invention relates to the field of intelligent control technology for the curing process of composite hydrogen storage cylinders, specifically to a method for deploying fiber optic sensor arrays, a process control method, and a system for curing hydrogen storage cylinders. The deployment method calculates regional risk weights by combining the structural stress concentration coefficient and molding defect risk coefficient of each region. A nonlinear deployment response function is used to achieve differentiated measurement point allocation, with denser distribution in high-risk areas and sparser distribution in low-risk areas. A fiber optic rotary connector is used to achieve continuous and stable transmission of optical signals under rotary curing conditions. The control method analyzes the internal curing degree and internal stress state online based on built-in sensor data, and dynamically adjusts process parameters using a phased, multi-objective control strategy, forming a closed-loop control system for the entire process. The entire technical solution can accurately capture stress characteristics in key areas, effectively suppress internal stress concentration, reduce the incidence of molding defects, and improve batch quality consistency and finished product qualification rate.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for the curing process of composite material hydrogen storage bottles, specifically a method for deploying an optical fiber sensor array for hydrogen storage bottle curing, a method for controlling the curing process of hydrogen storage bottles, and a system for controlling the curing process of hydrogen storage bottles. Background Technology

[0002] The quality of curing directly determines the mechanical properties and long-term service reliability of composite hydrogen storage cylinders. During the curing process, the chemical shrinkage effect of the resin crosslinking reaction, the inherent anisotropic thermal conductivity of the composite material, and the geometric discontinuities of the end caps, transition sections, and root of the Boss combine to form non-uniformly distributed internal molding stress within the structure. Local stress concentration can easily induce molding defects such as interlayer debonding and interface cracking, which are the core bottlenecks restricting the improvement of product performance and the consistency of mass production quality.

[0003] At the monitoring and sensing level, existing fiber optic sensing solutions generally adopt a uniformly spaced measurement point allocation logic, a design concept that has been widely applied in the field of hydrogen storage cylinder condition monitoring. For example, the hydrogen storage health monitoring device based on fiber optic sensing disclosed in patent CN220567059U lays metallized optical fibers in a spiral pattern along the outer wall of the metal inner liner. The fiber gratings in the cylindrical cylinder body and the sealing area are arranged according to a uniform spacing rule, with the distance between adjacent gratings not less than 2m. Basic monitoring of cylinder wall strain and temperature is achieved through uniform coverage across the entire area. This type of uniformly distributed scheme has a regular layout, is easy to implement, and can achieve basic sensing coverage of the entire structure, acquiring deformation and temperature data at the overall level. From the perspective of optimal configuration logic for molding monitoring, there are significant gradient differences in stress levels and molding defect risks across different structural regions of hydrogen storage cylinders. Ideally, the sensor array deployment should employ differentiated densities to match the regional risk levels: high-risk areas such as the end cap, transition section, and root of the boss, where geometry is discontinuous and stress concentration is significant, should have increased measurement point density for more frequent sampling, accurately capturing dynamic signals of stress peaks and defect initiation; low-risk areas with regular geometry and uniform stress distribution in the cylinder body can have appropriately reduced measurement point density, controlling sensor costs and additional structural disturbances while ensuring basic coverage. However, the existing solutions employ a uniform distribution logic, failing to match differentiated measurement point densities to the structural characteristics of the hydrogen storage cylinder. The measurement point density in high-risk areas is not significantly different from that in low-risk cylinder body sections, leading to missed stress concentration peak data in key areas and insufficient accuracy in capturing local defect initiation signals. Especially in curing molding scenarios, uniformly deployed sensor arrays struggle to accurately characterize the internal stress evolution process in high-risk areas, making it difficult to provide high-confidence data support for process intervention during critical stages of internal stress accumulation. This directly leads to the inability to identify abnormal stress growth trends in critical locations such as the end cap and the root of the boss in a timely manner during production, missing the process adjustment window at the defect initiation stage, and ultimately causing frequent molding defects such as interlayer debonding and interface cracking in high-risk areas, resulting in a higher rate of unqualified finished products and a significant increase in the cost of rework and scrapping of defective products. At the same time, it is difficult to pinpoint the root cause of batch-to-batch quality fluctuations through monitoring data, making it impossible to support continuous process optimization and to stably ensure the long-term mechanical performance and structural safety of hydrogen storage cylinders.

[0004] At the process control level, the mainstream solution for the industrial curing production of composite hydrogen storage cylinders is currently a "pre-set three-stage open-loop curing process curve, supplemented by surface thermocouple feedback temperature control." This solution executes the curing process according to a pre-calibrated heating rate, holding time, and cooling rate, collecting temperature data only through thermocouples deployed on the cylinder surface to make minor closed-loop corrections to the furnace temperature. This type of solution has a mature process path, a high degree of standardization in the production process, and simple parameter control, meeting the stable operation requirements of mass production. It is currently the most commonly used curing control method in the industry. From the optimal logic of precise control of molding internal stress, the internal stress evolution mechanism at different stages of the curing process is fundamentally different: during the heating stage, internal stress accumulates rapidly with the internal and external temperature difference and curing shrinkage; during the holding stage, internal stress gradually relaxes through the viscoelastic effect of the resin; and during the cooling stage, internal stress is easily permanently frozen due to the glass transition of the matrix. The ideal control strategy should match the stress evolution law of each stage, dynamically adjust process parameters based on the actual internal stress level inside the structure, and specifically suppress internal stress growth and freezing in stages. However, existing open-loop curing solutions rely solely on surface temperature for control, failing to perceive the actual evolution of internal curing degree and internal stress. They also lack differentiated control logic tailored to the stress formation mechanisms at different curing stages, operating with fixed process parameters throughout. This approach cannot proactively suppress interlayer thermal stress accumulation caused by excessive temperature differences during the heating phase, nor can it extend relaxation time based on real-time stress levels during the holding phase. Furthermore, it struggles to prevent rapid freezing of internal stress in the glass transition zone during the cooling phase. Ultimately, this results in frequent interlayer debonding defects in critical areas such as the end cap and the root of the boss, significant dispersion in product mechanical properties, and difficulty in ensuring batch molding quality stability. Summary of the Invention

[0005] To address the technical problem that existing fiber optic sensor arrays, which use uniformly spaced layouts without matching the different molding risks in various areas of the hydrogen storage cylinder to set differentiated measurement point densities, often miss stress peaks and defect initiation signals in high-stress concentration areas such as the end cap and the root of the boss, making it difficult to provide reliable data support for precise intervention in the curing process, and ultimately resulting in frequent molding defects in key areas, increased defect rates, and insufficient batch quality stability, this invention provides a fiber optic sensor array layout method for hydrogen storage cylinder curing.

[0006] To address the technical problems of existing open-ring curing processes that rely solely on surface temperature for feedback, fail to incorporate the evolution of internal stress at each stage of curing for precise phased control, make it difficult to perceive the true internal forming state of the structure, and ultimately lead to internal stress superposition freezing, high incidence of interlayer debonding defects in key areas, large dispersion of product mechanical properties, and insufficient batch quality stability, this invention provides a method for controlling the curing process of hydrogen storage cylinders.

[0007] Based on the fiber optic sensor array deployment method for hydrogen storage bottle curing and the hydrogen storage bottle curing process control method, this invention also provides a hydrogen storage bottle curing process control system.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A method for deploying a fiber optic sensor array for hydrogen storage cylinder solidification includes the following deployment steps: The hydrogen storage cylinders to be deployed are divided into four regions: the cylinder body, the transition section, the end cap region, and the root of the boss. The structural stress concentration factor K for each region is obtained. σ,i With curing and molding risk coefficient R i , where i is the region number; K in the corresponding region σ,i With R i Multiply by each other to obtain the regional risk weight w. i Normalization process yields the corresponding normalized value. ; Through the corresponding Calculate the deployment response value for each area. p, q, A, λ b β b The parameters of the pre-calibrated layout function are: exp[·] is an exponential function with the natural constant as its base; F j This represents the deployment response value for the j-th region. According to F in each region i Combined with the total number of sensor units N that can be deployed tot The effective length L of each area to be tested along the layout direction i The minimum number of sensing units N required to ensure observability of a single area min The number of sensor units deployed in each region was calculated. Spacing between adjacent sensor units `max{·}` is the maximum value operation; `round[·]` is the rounding function. According to the calculated N i With d i Sensing units are pre-embedded at the interface between adjacent winding layers in the corresponding area along the current area's layout direction until the pre-embedded layout of the entire bottle's sensing array is completed.

[0009] As a further improvement to the above scheme: structural stress concentration factor K σ,i The acquisition process is as follows: By establishing a three-dimensional finite element model of the hydrogen storage cylinder to be deployed and importing the preset curing process curve, a thermo-mechanical coupling simulation of the entire curing process was carried out, and the ratio of the maximum Mises equivalent stress in the corresponding region to the average Mises equivalent stress in the reference region of the cylinder was extracted. Curing and molding risk factor Ri The acquisition process is as follows: Historical production samples with the same resin matrix, fiber reinforcement material, winding lay-up structure and curing process as the hydrogen storage cylinders to be deployed are selected. Molding defects in each region are counted, and the regional defect incidence rate of the i-th region is calculated. The regional defect incidence rate is the proportion of the number of samples with molding defects in that region to the total number of valid statistical samples. Using the regional defect incidence rate of the cylinder section as the benchmark value, the ratio of the regional defect incidence rate of each region to the benchmark value is determined as the curing and molding risk coefficient of the corresponding region.

[0010] As a further improvement to the above scheme: The function parameters p, q, A, and λ are set. b β b Satisfying 0 < q ≤ p < 1, 0 < A < 1, 0 < λ b <1, 0<β b <1, and obtained through the following calibration steps: Establish a thermo-mechanical coupling simulation benchmark model of the curing process with the same structural type, material system and layup process as the hydrogen storage cylinder to be deployed. Calculate the true Mises equivalent stress distribution of each region of the entire cylinder during the curing process, and use it as the benchmark true value for evaluating the deployment effect. The total number of sensor units N that can be deployed per fixed bottle tot Multiple different regional measurement point allocation schemes are set up, with each scheme corresponding to a different allocation ratio of the number of sensing units in four regions: the cylinder section, the transition section, the end cap region, and the root of the boss. For each group of regional measurement point allocation schemes, the stress value of each regional measurement point is used as the sampling point for interpolation fitting to obtain the monitoring and reconstruction stress distribution; the average relative error between the reconstruction stress of the head region and the root of the Boss and the reference true value is calculated as the evaluation index of the current regional measurement point allocation scheme. With the objective of minimizing the deviation between the area measurement point allocation result output by the deployment response function and the area measurement point allocation scheme with the optimal evaluation index, a nonlinear least squares fitting algorithm is used to solve for the deployment function parameters p, q, A, and λ. b β b The initial optimal value; Multiple independent samples with different total number of sensing units are selected to verify the generalization of the initial optimal value. If the average relative error between the head region and the root of Boss of all samples does not exceed the preset error threshold, the initial optimal value is determined to be the final calibration value. If the error exceeds the threshold, the sample of the area measurement point allocation scheme is supplemented and refitted until the accuracy requirements are met.

[0011] As a further improvement to the above plan, the specific deployment direction for each area is as follows: The sensing units in the cylinder body section are arranged along the axial direction of the hydrogen storage cylinder; the sensing units in the head area are arranged along the meridian direction of the head; and the sensing units in the transition section and the root of the Boss are arranged along the circumference of the hydrogen storage cylinder. The effective length of the cylinder section is the axial length of the cylindrical section of the cylinder; the effective length of the head area is the arc length of the head meridian; the effective length of the transition section is the circumference length at the midpoint of the axial direction of the transition section; and the effective length of the Boss root is the circumference length at the interface between the Boss metal part and the composite material layer. When pre-embedding the sensing unit, a continuous all-fiber signal transmission link from the measuring point on the bottle to the external demodulation equipment is simultaneously constructed: The grating area and the pigtail of each sensing unit are a continuous structure of the same optical fiber. The reflected light signal carrying temperature and strain information sensed by the grating area is transmitted by total internal reflection through the pigtail core. The pigtails of each sensing unit are laid straight along the winding layer and converge at the end of the hydrogen storage tank. The pigtails are lightly pressed and fixed by fiber bundles to suppress optical signal loss caused by bending. The pigtails converged at the end of the hydrogen storage tank are fused and fixed to the rotor end of the optical fiber rotary connector. The stator end of the optical fiber rotary connector is connected to the external optical fiber signal demodulation unit through a fixed optical fiber. The optical coupling structure inside the optical fiber rotary connector enables continuous and uninterrupted transmission of optical signals under the condition of hydrogen storage tank rotation. Finally, the optical fiber signal demodulation unit converts the optical wavelength signal into a calculable digital signal for output.

[0012] A method for controlling the curing process of a hydrogen storage bottle involves pre-embedding an optical fiber sensor array on the hydrogen storage bottle using a fiber optic sensor array deployment method. Each sensing unit in the array includes at least one strain sensing grating and one temperature compensation grating. The method includes the following control steps: S1. After completing the pre-embedding of the fiber optic sensing array, under a constant temperature and no-load standard environment, the interlayer strain transfer correction coefficient is introduced to correct the grating sensitivity coefficient, and an improved multi-physics decoupling model is constructed. S2. After the hydrogen storage cylinder is put into the furnace and the curing program is started, the original wavelength signals of the temperature compensation grating and the strain sensing grating are collected at each sampling time according to the preset sampling frequency. The real-time absolute temperature of each measuring point is calculated based on the wavelength shift of the temperature compensation grating, and the total center wavelength shift of the strain sensing grating is obtained. At the same time, the real-time resin curing degree is calculated based on the real-time absolute temperature. S3. Input the real-time absolute temperature into the material curing kinetic constitutive model, which is corrected by both interlayer thermal resistance and winding angle. The real-time resin curing degree calculated at the previous sampling time is used as the initial state. Update the real-time resin curing degree at the current sampling time. Calculate the free chemical shrinkage strain based on the real-time resin curing degree. Then, input the total offset of the center wavelength of the strain sensing grating, the real-time temperature change, and the free chemical shrinkage strain into the multiphysics decoupled model to obtain the pure mechanical curing strain. Calculate the molding internal stress at each measuring point based on the real-time absolute temperature, the real-time resin curing degree, and the pure mechanical curing strain. S4. Based on the current process program segment of the curing oven, the rate of change of the oven temperature setpoint, and the real-time resin curing degree, jointly determine whether the current stage is heating, holding, or cooling. Compare the average resin curing degree, average molding internal stress, and maximum molding internal stress of each region with the current stage's curing degree target trajectory and allowable internal stress limit to obtain the measured curing degree deviation and measured internal stress deviation of each region. Perform equivalent correction and dimensionless processing on the deviations based on the physical occupancy of the sensing unit, and combine the regional risk weights of each region to obtain the overall molding state deviation. Match the multi-objective control strategy corresponding to the current stage, and under the constraints of the preset upper and lower limits of process parameters and the maximum rate of change, generate incremental adjustment commands for process parameters such as heating rate, holding temperature, holding time, or cooling rate, and send them to the curing oven for execution. S5. After adjusting the process parameters, continuously collect the real-time signal after regulation through the fiber optic sensor array and return to steps S2-S4 for cyclic verification. Use the deviation closed-loop iterative mechanism to incrementally fine-tune the process parameters until the entire curing process is completed.

[0013] As a further improvement to the above scheme: the improved multiphysics decoupling model and the material solidification dynamics constitutive model are solved in real time using a serial linkage iterative method. The solution process at each sampling time is as follows: First, perform temperature calculation and cure degree update: The total offset Δλ of the center wavelength of the temperature compensation grating, which is positioned at the same location as the strain sensing grating, is measured. B The real-time temperature change ΔT at the current measuring point is calculated. The real-time temperature change ΔT is added to the reference absolute temperature T0 of the constant temperature and no load standard environment to obtain the real-time absolute temperature T at the current measuring point, which is then input into the material curing kinetic constitutive model. In the material curing kinetic constitutive model, the real-time resin curing degree at the previous sampling time is used as the initial state. Based on the real-time absolute temperature T, the real-time curing reaction rate dα / dt at the current time is calculated. After time step integration, the real-time resin curing degree α at the current sampling time is obtained. The constitutive model of material solidification kinetics is expressed as follows: ; In the formula, A1 and A2 are the pre-exponential factors corresponding to the dual-reaction mechanism of the curing reaction, respectively; ΔE1 and ΔE2 are the activation energies corresponding to the dual-reaction mechanism of the curing reaction, respectively; R is the universal gas constant; m and n are the order of the curing reaction; f(θ) is the thermal conductivity correction function for the winding angle, and the independent variable θ is the fiber winding angle at the current measuring point; γ h is the interlayer thermal resistance correction factor; e is the natural constant; Then perform strain decoupling separation: Based on the updated real-time resin cure degree α, and combined with the chemical shrinkage coefficient of the material at full cure, the free chemical shrinkage strain component generated by the current resin crosslinking reaction is calculated. The total offset of the center wavelength Δλ B Temperature change ΔT, free chemical contraction strain component Substituting into the improved multiphysics decoupling model, and successively subtracting the temperature response component and the chemical shrinkage response component, we obtain the pure mechanical solidification strain caused solely by the constrained deformation of the structure. ; The improved multiphysics decoupling model is represented as follows: ; In the formula, Δλ B The total offset of the center wavelength of the strain sensing grating at the current measuring point is obtained in real time by the fiber optic signal demodulation unit; K T ΔT is the inherent temperature sensitivity coefficient of the strain sensing grating at the current measuring point; ΔT is the real-time temperature change at the current measuring point, which is calculated by the temperature compensation grating deployed at the same location. This is the actual grating sensitivity coefficient at the current measurement point after interlayer strain transfer correction; This represents the purely mechanical solidification strain at the current measuring point; This represents the free chemical shrinkage strain component generated by the resin crosslinking reaction at the current measuring point. As a further improvement to the above scheme: before calculating the overall forming state deviation, the single-point calculation results of each measuring point are first aggregated at the regional level to obtain the state characteristic quantities of each region. The specific process is as follows: For real-time absolute temperature and real-time resin curing degree, the average value of the region is calculated by the effective area weighting method of the measuring points, and the average absolute temperature and average resin curing degree of the corresponding region are obtained, which characterizes the overall thermal state and curing reaction process of the region. For the forming internal stress, both the regional average and the regional maximum are calculated: the regional average internal stress is calculated by weighting the effective area of ​​the measuring points, which is used to characterize the overall stress level of the region; the regional maximum internal stress is the maximum value of the forming internal stress of all measuring points in the corresponding region. The formula for calculating the internal stress during forming is as follows: ; In the formula, σ n (t) represents the molding internal stress accumulated at the nth measuring point from the start of curing to time t; G(·) is the modified relaxation modulus considering the coupling between curing degree and temperature. For the nth measurement point at Real-time resin curing degree at any given moment; For the nth measurement point at Real-time absolute temperature at any given moment; For the nth measurement point at Total strain at any given moment; For time variables The infinitesimal element; for The first derivative; The average resin curing degree and average internal stress of each region are subtracted from the target value of the current curing stage to obtain the measured curing degree deviation and measured internal stress deviation of each region, which are used as inputs for subsequent physical occupancy equivalent correction of the sensing unit.

[0014] As a further improvement to the above solution, the process for obtaining the overall molding state deviation is as follows: Calculate the volume fraction of the sensing unit in each region. The occupied volume fraction is the ratio of the equivalent volume of a single sensing unit to the effective volume of the corresponding composite material in the region. Combined with the fraction of occupied volume The sensor unit layout direction, real-time resin curing degree, gel point curing degree, real-time absolute temperature and glass transition temperature are used to construct the physical occupancy equivalent variables for each region: ; ; In the formula, Ω i (t) represents the physical occupancy equivalent variable of the i-th region at time t; η represents the volume fraction of the sensing unit within the i-th region; i Assign directional modulation coefficients to the sensing units in the i-th region to quantize the enhancement of occupancy disturbances by the fiber angle; θ i λ is the angle between the arrangement direction of the sensing units in the i-th region and the direction of the main supporting fiber in that region; o,i γ is the occupancy saturation coefficient for the i-th region; i α is the correction coefficient for the curing state of the i-th region; i (t) represents the real-time resin curing degree of the i-th region at time t; α g,i ρ represents the degree of gel point curing in the i-th region at time t; i ζ is the correction factor for the glass transition region of the i-th region; i(t) represents the normalized variable for the glass transition of the i-th region at time t; T i (t) represents the real-time absolute temperature of the i-th region at time t; T gi (t) represents the real-time glass transition temperature of the i-th region at time t; ΔT g is the half-width of the glass transition influence range; clip[·] is the cutoff function; exp[·] is the exponential function with the natural constant as the base; Based on the physical occupancy equivalent variable and the pre-calibrated occupancy deviation sensitivity coefficient, the measured curing degree deviation and measured internal stress deviation of each region are converted into the true state deviation without sensor unit interference: ; ; In the formula, Δα represents the actual degree of curing deviation of the i-th region at time t. i (t) represents the measured curing degree deviation of the i-th region at time t; c α,i The occupancy deviation sensitivity coefficient corresponds to the curing degree deviation of the i-th region. Δσ represents the actual internal stress deviation of the i-th region at time t. i (t) represents the measured internal stress deviation of the i-th region at time t; c σ,i This is the occupancy deviation sensitivity coefficient corresponding to the stress deviation in the i-th region; based on and Calculate the relative deviation corresponding to the state deviation: ; ; In the formula, e α,i (t) represents the relative curing degree deviation of the i-th region at time t; δ α,i e is the allowable deviation threshold for the degree of cure of region i in the current curing stage; σ,i (t) represents the relative internal stress deviation of the i-th region at time t; δ σ,i Let be the allowable deviation threshold of internal stress in region i during the current curing stage; [·] + = max(x,0) is the positive part truncation function, max is the maximum value operation, and x is the input value; The actual curing degree deviation and actual internal stress deviation of each region are weighted and summed using regional risk weights to obtain the overall molding state deviation, which includes both curing degree deviation and internal stress deviation. ; In the formula, D(t) represents the overall forming state deviation of the hydrogen storage cylinder at time t; λ αλ is the priority weighting coefficient for curing degree deviation. σ This is the priority weighting coefficient for internal stress deviation.

[0015] As a further improvement to the above scheme: Match the multi-objective control strategy corresponding to the current solidification stage, with differentiated target priorities, deviation weight ratios, and core control objects for each stage, specifically implemented as follows: During the heating stage, the core control objectives are temperature field uniformity and preventing excessive accumulation of internal stress during molding. The priority weighting coefficient λ for curing degree deviation is... α The priority weighting coefficient λ for internal stress deviation is 0.4~0.6. σ The value is 0.4~0.6, and the core control object is the heating rate. When the deviation of the overall molding state exceeds the set threshold, the heating rate is reduced. When the deviation is too large, the intermediate heat preservation platform at the corresponding temperature is automatically inserted. After the temperature field is homogenized and the internal stress growth rate drops back to the safe range, the preset heating process is restored. During the heat preservation stage, the dual core control objectives are uniformity of curing degree and relaxation of internal stress during molding. The priority weighting coefficient λ for curing degree deviation is... α The priority weighting coefficient λ for internal stress deviation is 0.2 to 0.4. σ The value is 0.6~0.8, and the core control objects are the heat preservation time and the heat preservation temperature. Taking the reduction of internal stress as the core, the heat preservation time increment adjustment command is generated. The high elastic viscoelasticity of the resin is used to promote the release of internal stress. At the same time, the degree of curing is met as the bottom line condition, while taking into account the sufficiency of the curing reaction. The cooling stage aims to suppress the superposition of internal stress during molding and freezing, with the priority weighting coefficient λ for curing degree deviation. α The priority weighting coefficient λ for internal stress deviation is 0.1 to 0.3. σ The value is set to 0.7~0.9, and the core control object is the cooling rate. When the temperature drops to the glass transition temperature range, a slow cooling platform is inserted in combination with the current internal stress level to prevent the internal stress from being frozen into irreversible residual stress during the glass transition process.

[0016] A hydrogen storage cylinder curing process control system is used to execute a hydrogen storage cylinder curing process control method, including a pre-embedded fiber optic sensor array, a fiber optic rotary connector, a fiber optic signal demodulation unit, an online analysis and decision-making unit, a curing oven execution control unit, and a process data iteration unit. The pre-embedded fiber optic sensing array adopts the fiber optic sensing array deployment method for hydrogen storage bottle curing, and is pre-embedded in the winding layers of the cylinder section, transition section, end cap area and Boss root of the hydrogen storage bottle; each sensing unit includes at least one strain sensing grating and one temperature compensation grating, used to collect temperature and strain light signals during the curing process; the pigtails of each sensing unit are laid straight along the winding layer and converge to the end of the hydrogen storage bottle for lead-out. The rotor end of the fiber optic rotary connector is fused and fixed to the lead-out pigtail of the pre-embedded fiber optic sensing array, and rotates synchronously with the hydrogen storage tank; the stator end of the fiber optic rotary connector is connected to the fiber optic signal demodulation unit through armored fixed fiber optic cable, which is used to maintain continuous optical path alignment under the condition of hydrogen storage tank rotation and solidification, so as to realize uninterrupted transmission of optical signal. The fiber optic signal demodulation unit is used to perform wavelength demodulation and signal preprocessing on the received optical signal, and outputs the raw wavelength data of each measurement point to the online analysis and decision unit. The online analysis and decision-making unit incorporates an improved multi-physics decoupling model, a material curing dynamics constitutive model corrected by interlayer thermal resistance and winding angle, a regional feature aggregation module, a sensor unit physical occupancy equivalent correction module, and a phased multi-objective control strategy library. It is used to calculate the real-time absolute temperature, pure mechanical curing strain, real-time resin curing degree, and cumulative internal stress of each measuring point through a serial linkage iterative method. After regional feature aggregation and physical occupancy equivalent correction, it calculates the overall molding state deviation, matches the control strategy corresponding to the current curing stage, and generates incremental adjustment instructions for curing process parameters. The curing oven execution control unit is connected to the command output terminal of the online analysis and decision unit to receive incremental adjustment commands and adjust the heating rate, holding temperature, holding time and cooling rate of the curing oven in real time. The process data iteration unit and the online analysis and decision-making unit interact bidirectionally to store monitoring data, process parameters and finished product performance data throughout the solidification process. The constitutive model parameters and control thresholds are iteratively corrected through a recursive parameter identification algorithm to continuously optimize the control accuracy of subsequent batches.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. In the fiber optic sensor array deployment method for hydrogen storage cylinder curing, the hydrogen storage cylinder is first divided into four regions along the structure: cylinder body, transition section, end cap region, and Boss root region. The structural stress concentration coefficient of each region is extracted through thermo-mechanical coupling simulation. The curing risk coefficient is calculated by combining the defect incidence rate of historical production samples. The two are coupled to obtain a normalized regional risk weight. On this basis, through a nonlinear deployment response function that includes local over-density suppression and medium-to-high risk curvature enhancement mechanisms, the sensor unit quota and measurement point spacing of each region are allocated according to the risk weight ratio. Finally, a differentiated deployment pattern is formed, with denser measurement points in high-risk regions such as the end cap and Boss root region, and sparser distribution in low-risk regions such as the cylinder body region as needed. This method departs from the traditional uniform layout logic of evenly spaced monitoring points. Under the premise of limited total number of monitoring points, it allocates more monitoring resources to areas with high stress concentration, which helps to improve the monitoring coverage density and data sampling precision of high-risk areas. It can better capture the state change characteristics of stress concentration areas and reduce the possibility of missing key feature data. It can provide more referential monitoring data for process adjustment during the curing process, assisting in timely adjustment of process parameters at the defect initiation stage. This helps to reduce the occurrence of molding defects such as interlayer debonding and interface cracking in key areas, thus helping to control production rework costs and improve the quality stability of mass production.

[0018] 2. In the method for controlling the curing process of hydrogen storage cylinders, the internal interlayer signals collected by an embedded fiber optic sensor array are used to separate real-time temperature and pure mechanical strain through a multi-physics decoupling model. Combined with a modified material curing kinetic constitutive model, the actual state of curing degree and internal stress inside the structure is analyzed online, which improves the information limitation of traditional solutions that only rely on surface temperature feedback. At the same time, different control strategies are matched for the different internal stress evolution mechanisms in each stage of heating, holding, and cooling: in the heating stage, the heating rate is dynamically reduced with temperature difference and strain growth rate as dual driving forces to alleviate the rapid accumulation of interlayer thermal stress caused by excessive internal and external temperature difference; in the holding stage, the holding time is dynamically extended by combining the curing degree gap and stress overshoot, and the high elastic viscoelasticity of the resin is used to promote the relaxation of internal stress; in the cooling stage, the cooling rate is actively reduced in the glass transition range and a slow cooling platform is inserted to reduce the risk of internal stress being permanently frozen due to a sudden increase in matrix stiffness. This method upgrades the fixed-parameter open-loop curing to dynamic closed-loop control based on the actual internal state, which can be specifically adapted to the internal stress evolution characteristics of each stage. It helps to reduce the stress concentration level in key areas, reduce the probability of molding defects such as interlayer debonding, and reduce the batch dispersion of product mechanical properties. It has a positive effect on improving the finished product qualification rate and batch quality consistency. Attached Figure Description

[0019] Figure 1 This is a flowchart of the implementation process of the present invention.

[0020] Figure 2This is a schematic diagram of the fiber optic grating sensor arrangement points according to an embodiment of the present invention.

[0021] Figure 3 A schematic diagram showing the regional division of the composite material hydrogen storage bottle and the axial cross-sectional layout of the sensor array.

[0022] Figure 4 This is a schematic diagram showing the layout of circumferential sensing points in the left and right transition sections of a hydrogen storage cylinder.

[0023] Figure 5 A schematic diagram showing the layout of meridian sensing points in the left and right end cap areas of a hydrogen storage cylinder.

[0024] Figure 6 This is a schematic diagram showing the layout of circumferential sensing points at the base of the left and right bosses of the hydrogen storage tank.

[0025] Figure 7 A schematic diagram comparing the baseline solidification system and the phased control process curve.

[0026] Figure 8 This is a schematic diagram comparing the maximum Mises stress in different regions at each stage of the entire control process.

[0027] Figure 9 This is a schematic diagram comparing the maximum residual stress in each region at the end of cooling.

[0028] Figure 10 This is a schematic diagram of the staged curing online control process curve according to an embodiment of the present invention.

[0029] Figure 11 This is a comparison diagram of the stress control effect in key areas of the gas cylinder according to an embodiment of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] like Figure 1 As shown, this embodiment takes the I-IV type carbon fiber wound composite hydrogen storage cylinder as the specific implementation object, and combines the deployment method, control method and system architecture of the present invention to explain in detail the complete implementation process and core principles of the technical solution.

[0032] I. Hydrogen Storage Cylinder Solidification Process Control System

[0033] The hydrogen storage bottle curing process control system used in this embodiment is used to implement the curing process control method in Part III below. It includes a pre-embedded fiber optic sensor array, a fiber optic rotary connector, a fiber optic signal demodulation unit, an online analysis and decision-making unit, a curing oven execution control unit, and a process data iteration unit.

[0034] 1. Embedded fiber optic sensor array

[0035] The fiber optic sensor array deployment method described in Part II below is adopted, with the spiral and circumferential winding layers or corresponding adjacent winding layers pre-embedded in the interface of the cylindrical section, transition section, end cap area, and root of the hydrogen storage cylinder. Each sensing unit includes at least one strain sensing grating and one temperature compensation grating. The strain sensing grating retains its original coating to ensure strain transmission performance, while the temperature compensation grating is stress-free encapsulated using a quartz inert capillary sleeve and the same batch of resin. Both are deployed in the same location to sense the strain and temperature optical signals during the curing process, respectively. The pigtails of each sensing unit are laid straight along the winding layer, lightly pressed and fixed with fiber bundles, and finally converged and led out at the end of the hydrogen storage cylinder.

[0036] 2. Fiber Optic Rotary Connector

[0037] The rotor end of the fiber optic rotary connector is fused and fixed to the pigtail of the pre-embedded fiber optic sensor array using a fiber optic fusion splicer, rotating synchronously with the hydrogen storage tank in the curing oven. The stator end is led out of the curing oven via a fixed fiber and connected to the fiber optic signal demodulation unit. The fiber optic rotary connector employs a precision optical coupling structure, maintaining an insertion loss fluctuation of less than 0.5dB under continuous rotor rotation, achieving continuous and uninterrupted optical signal transmission and solving the industry pain point of the inability to extract fiber optic signals in real time during the rotation curing process.

[0038] Another approach employs a dual-ended fiber optic rotary connector architecture. One fiber optic rotary connector is configured at each end of the hydrogen storage cylinder, corresponding to the sensor arrays of the left and right halves of the cylinder, respectively. The sensor pigtails from the left half (left Boss root, left end cap, left transition section) converge at the left end of the hydrogen storage cylinder and are fused to the rotor end of the left-side fiber optic rotary connector. Similarly, the sensor pigtails from the right half (right Boss root, right end cap, right transition section, cylinder body) converge at the right end of the hydrogen storage cylinder and are fused to the rotor end of the right-side fiber optic rotary connector. The stator ends of the two fiber optic rotary connectors are led out of the curing oven via armored fixed optical fibers. After converging, they are connected to the corresponding channel of the fiber optic signal demodulation unit, enabling continuous and uninterrupted transmission of optical signals across all cylinder measurement points during rotation.

[0039] 3. Fiber optic signal demodulation unit

[0040] This embodiment uses a fiber optic grating demodulator with a wavelength resolution of 1 pm and an adjustable sampling frequency of 1~5 Hz. It has a built-in wavelength demodulation algorithm and a preprocessing module for sliding filtering and outlier removal. It can simultaneously analyze the center wavelength data of multiple grating channels, convert the optical signal into a digital signal, and transmit it to the online analysis and decision-making unit in real time.

[0041] 4. Online Analysis and Decision-Making Unit

[0042] This is an industrial-grade embedded computing unit that incorporates an improved multiphysics decoupling model, a material curing kinetic constitutive model corrected by both interlayer thermal resistance and winding angle, a region feature aggregation module, a sensor unit physical occupancy equivalent correction module, and a phased multi-objective control strategy library. The unit can perform the entire process of temperature-strain decoupling, online analysis of curing degree and internal stress, region state aggregation, occupancy deviation correction, overall deviation calculation, and control command generation. The command output is connected to the curing oven's execution control unit.

[0043] 5. Curing oven control unit

[0044] The extended control module of the original PLC temperature control system of the curing oven can receive incremental adjustment commands issued by the online analysis and decision-making unit. Under the constraints of the upper and lower limits of preset process parameters and the maximum rate of change, it can adjust the oven temperature setpoint, heating rate, holding time and cooling rate in real time to achieve dynamic correction of the process curve.

[0045] 6. Process Data Iteration Unit

[0046] It serves as a cloud or local database computing unit, enabling bidirectional data interaction with online analysis and decision-making units. It stores monitoring data, process parameter curves, and finished product performance test data for the entire solidification process of each batch of gas cylinders. It has a built-in recursive parameter identification algorithm that can iteratively correct the parameters and control thresholds of the material solidification kinetic constitutive model based on multiple batches of production data, continuously optimizing the control accuracy of subsequent batches.

[0047] II. Deployment of Pre-embedded Fiber Optic Sensor Array

[0048] This section describes the planning and pre-installation of the full-bottle sensor array for hydrogen storage bottles, following the fiber optic sensor array deployment method for hydrogen storage bottle solidification.

[0049] (I) Regional Division and Risk Weight Calculation

[0050] 1. Regional division

[0051] The hydrogen storage cylinders to be deployed are divided into four regions along the structure: the cylinder body, the transition section, the end cap region, and the root of the boss, denoted as Region 1 to Region 4. Each region has different structural characteristics and failure risks: the cylinder body has regular geometry and uniform stress distribution, making it a low-risk baseline region; the transition section and end cap region exhibit curvature changes, resulting in significant stress concentration effects; the root of the boss region, being a metal-composite material interface, is a high-risk area for molding defects and stress concentration.

[0052] 2. Calculate the stress concentration factor of the structure.

[0053] A three-dimensional finite element model of the hydrogen storage cylinder to be deployed was established, and a preset curing process curve was imported to perform a thermo-mechanical coupling simulation of the entire curing process. The maximum Mises equivalent stress σ in each region was extracted. max,i The average Mises equivalent stress σ of the cylinder section avg,ref Based on the benchmark, the structural stress concentration factor K of the i-th region is... σ,i =σ max,i / σ avg,ref Structural stress concentration factor K σ,i It quantifies the degree of local stress amplification caused by geometric structure; the higher the value, the higher the risk of stress concentration in the structure.

[0054] 3. Calculate the curing and molding risk factor.

[0055] Historical production samples with the same resin matrix, fiber reinforcement material, winding layup structure, and curing process as the hydrogen storage cylinders to be deployed were selected. Molding defects (interlayer debonding, porosity, interface cracking, etc.) in each region were statistically analyzed. The regional defect incidence rate P of the i-th region was calculated. i n is the number of samples with molding defects in this area. i N% of the total valid statistical samples total Ratio: P i =n i / N total .

[0056] The defect determination adopts the length threshold criterion: when there is interlayer debonding or interface cracking with a length greater than or equal to the set length (e.g., 5 mm) or concentrated pores with a diameter greater than or equal to the set diameter threshold (e.g., 1 mm) in the area, it is recorded as a forming defect in the area; multiple defects in the same area of ​​a single sample are counted as only one defect sample to ensure the engineering rationality of the risk coefficient statistics.

[0057] The regional defect incidence rate P of the cylinder section ref Using the baseline value, the curing and molding risk coefficient R of the i-th region is... i =P i / P ref This curing and molding risk coefficient quantifies the risk of regional molding quality fluctuations caused by process factors.

[0058] 4. Calculation and Normalization of Regional Risk Weights

[0059] The structural stress concentration factor K of the corresponding region σ,i With curing and molding risk coefficient R i Multiply by each other to obtain the regional risk weight w for the i-th region. i =K σ,i ·R i .

[0060] The regional risk weights of all four regions are normalized to obtain normalized regional risk weights. w j Let i be the regional risk weight for the j-th region, where i,j∈[1,4].

[0061] After normalization, the sum of all weights is 1, which is used for the subsequent calculation of the proportion of measurement point allocation.

[0062] (II) Constructing the Response Function and Assigning Measurement Points

[0063] To achieve the goal of optimizing the deployment by increasing the density of monitoring points with higher risks, while avoiding excessive density in high-risk areas and considering the monitoring sensitivity in medium- and high-risk areas, this invention employs a nonlinear deployment response function to allocate regional monitoring point quotas. This function is constructed through a three-layer mechanism: basic risk response, suppression of excessive local density, and enhancement of curvature in medium- and high-risk areas. The specific derivation process is as follows.

[0064] 1. Step-by-step construction and derivation of the response function

[0065] The deployment response value F of the i-th region i The construction process consists of four steps: (1) Basic risk response items To ensure that zero-risk weight corresponds to zero-deployment response, and to adjust for the response difference between low-risk and high-risk areas, a basic risk response term in the form of a power function is first constructed: ; In the formula, B i Let be the base risk value for the i-th region. p is the risk response index, where p > 0. The larger the value of p, the weaker the response of low-risk regions relative to high-risk regions, and the more significant the concentration trend of measurement quotas towards high-risk regions; the smaller the value of p, the more gradual the difference in response between high- and low-risk regions.

[0066] The basic risk response item meets the requirements. At that time, B i =0, which meets the physical logic that requires no additional deployment if the risk is zero.

[0067] (2) Local over-dense suppression response term

[0068] If only the power function basic term is used, high-risk areas may receive an excessively large number of sensing units, resulting in overly dense local measurement points, wasted costs, and the introduction of additional structural disturbances. Therefore, a denominator suppression term that increases with the risk weight is introduced to impose a nonlinear saturation constraint on the response growth in high-risk areas: ; In the formula, S i λ represents the local overdensity suppression response term for the i-th region; b q is the local over-density inhibition intensity coefficient, which controls the overall amplitude of saturation inhibition; q is the inhibition effect weight index, which controls the weight range of the inhibition effect intervention, q>0; A is the baseline constant.

[0069] The value of the local overdensity suppression term varies with... Increasing or decreasing the size of the deployment can effectively suppress overly dense deployments in high-risk areas caused by excessively strong responses. At that time, S i =1 / A, the local over-density suppression term only scales the overall response and does not change the original risk ranking of each region, but only adjusts the growth rate of the response.

[0070] Its monotonicity can be proven by taking the first derivative of the local over-density suppression term: ; Because of 0 <q≤p<1、0<λ b <1, The derivative is always greater than 0, indicating that the local over-density suppression term only adjusts the response growth rate and does not change the original risk ranking of each region.

[0071] (3) Construct curvature enhancement terms for medium and high risk areas

[0072] Normalized risk weights ,function The curvature smoothly and monotonically increases from 0 to 1 within the interval [0,1], exhibiting a more significant curvature change in the medium-to-high weighted interval. This characteristic is used to construct an exponential-triangular curvature adjustment term to enhance the relative response sensitivity in the medium-to-high risk region: ; Among them, C i β is the curvature enhancement term for the high-risk area in the construction of the i-th region; b β is the curvature enhancement coefficient, 0 < β b <1; exp[·] is an exponential function with the natural constant as its base.

[0073] When β b When =0, C i =1, this enhancement term has no effect; with β bThe increase in size enhances the relative response in medium- and high-risk areas, thereby improving the sensitivity of measurement point allocation in the head and root region of the boss, and better capturing the distribution characteristics of stress concentration and forming defects.

[0074] (4) The final deployment response function is obtained by combining the components.

[0075] Multiplying the base response term after local over-density suppression with the curvature enhancement term and combining them, we obtain the final regional layout response value: ; Using the multiplication composition method can simultaneously guarantee Time F i =0、 When F increases i The basic characteristic of monotonically increasing, and the intensity of local over-density suppression and the degree of curvature enhancement in medium- and high-risk areas are respectively determined by λ. b β b Independent adjustment, clear physical meaning of parameters, and easy calibration and debugging.

[0076] 2. Setting up function parameter calibration

[0077] Configure the function parameters p, q, A, λ b β b The calibration was obtained by combining simulation fitting and generalization verification. The specific steps are as follows: (1) Establish a thermo-mechanical coupling simulation benchmark model of the curing process with the same structure, material system and layup process as the hydrogen storage cylinder to be deployed, and calculate the true Mises equivalent stress distribution of the entire curing process in each region of the cylinder, which serves as the benchmark true value for evaluating the deployment effect. (2) The total number of sensing units N that can be deployed per bottle tot Multiple different regional measurement point allocation schemes are set up, with each scheme corresponding to a different allocation ratio of the number of sensing units in four regions: the cylinder section, the transition section, the end cap region, and the root of the boss. (3) For each group of regional measurement point allocation schemes, the stress value of each regional measurement point is used as the sampling point for interpolation fitting to obtain the monitoring reconstruction stress distribution; the average relative error between the reconstruction stress of the head area and the root of the Boss and the reference true value is calculated as the evaluation index of the current regional measurement point allocation scheme. (4) Taking the minimum deviation between the regional measurement point allocation result output by the deployment response function and the optimal regional measurement point allocation scheme of the evaluation index as the fitting objective, the nonlinear least squares fitting algorithm is used to solve for the deployment function parameters p, q, A, and λ. b β b The initial optimal value; (5) Select multiple independent samples with different total number of sensing units for generalization verification. If the average relative error between the head region and the root of Boss of all samples does not exceed the preset error threshold, then the set of layout function parameters is determined as the final calibration value. If the error exceeds the threshold, supplement the sample of the area measurement point allocation scheme and refit until the accuracy requirements are met.

[0078] 3. Calculation of the number and spacing of measuring points in the area.

[0079] Based on the deployment response value F of each region i Based on the total measurement point quota and the effective length of the area, the number of sensor units deployed and the spacing between measurement points in each area are calculated: ; In the formula, N tot N represents the total number of sensor units that can be deployed per bottle. min The minimum number of sensor units required to ensure observability of a single area; round (·) is the rounding function; max{·} is the maximum value operation to ensure that low-risk areas also have basic monitoring coverage capabilities.

[0080] Spacing between adjacent sensor units: ; In the formula, L i The effective length of the i-th region along the layout direction is: the effective length of the cylinder section is the axial length of the cylindrical section of the cylinder, the effective length of the end cap region is the arc length of the end cap meridian, the effective length of the transition section is the circumference length at the midpoint of the axial direction of the transition section, and the effective length of the Boss root is the circumference length at the interface between the Boss metal part and the composite material layer.

[0081] (III) Deployment of sensor units and establishment of transmission links

[0082] 1. Layout direction and layer

[0083] like Figure 2The diagram shows the zoning and sensor array layout of a composite material hydrogen storage cylinder. The cylinder has a horizontal cylindrical structure with end caps and Boss metal interfaces at both ends. The cylinder body is divided along the axial direction into several structural regions: the Boss root, the end cap area, the transition section, and the cylinder body. Each region has different geometric characteristics and molding risk levels. The red dots in the diagram represent fiber optic sensing points, arranged with differentiated density: higher density and smaller spacing of sensing points in high-risk areas such as the end caps and Boss root, while relatively sparser sensing points are found in the lower-risk central cylinder body. This visually demonstrates the optimized layout logic of "higher risk, denser sensing points," improving monitoring accuracy in key areas with limited sensing point quotas. Simultaneously, surface temperature gratings are attached to the corresponding regions on the outer surface of the cylinder, paired with internal embedded sensing points, serving as a surface temperature reference for calculating the internal and external temperature difference during the heating phase, providing dual-end temperature data for heating rate control. Within each sensing unit, strain gratings are configured with different grating axis directions according to monitoring requirements. The cylinder section and end cap area are equipped with axial and circumferential strain gratings, while the transition section and the root of the boss are equipped with circumferential main direction strain gratings, enabling synchronous sensing of deformation in multiple directions.

[0084] 2. Construct an all-fiber continuous signal transmission link

[0085] The grating area and the pigtail of each sensing unit are a continuous structure of the same optical fiber. The optical signal sensed by the grating area is transmitted through the pigtail core by total internal reflection. During deployment, the pigtail is laid straight along the winding layer and fixed by light pressure of the fiber bundle to suppress optical signal loss caused by bending.

[0086] After the pigtails converge at the end of the hydrogen storage cylinder, they are fused and fixed to the rotor end of the optical fiber rotary connector, and the rotor end rotates synchronously with the gas cylinder. The stator end of the optical fiber rotary connector is connected to the external optical fiber signal demodulation unit through armored fixed optical fiber. The internal optical coupling structure enables continuous and uninterrupted transmission of optical signals under rotation conditions, and finally the demodulation unit converts it into a digital signal output.

[0087] (iv) Numerical Examples of Sensor Array Deployment

[0088] This embodiment uses a Type IV carbon fiber wound hydrogen storage cylinder with a nominal working pressure of 55 MPa and a design pressure of 82.5 MPa as an example. The metal inner liner material is 4130X, the nominal total length is 5470 mm, the outer diameter of the cylindrical section is 671 mm, the minimum design wall thickness is 9.0 mm, and the length of the straight section is 4559 mm. The composite material has a total of 21 layers, the final outer diameter of the middle part of the cylinder is 742.024 mm, and the maximum outer diameter is 743.836 mm.

[0089] like Figure 3The diagram shows a Cartesian coordinate system O-XY, established with the left end face of the steel inner liner as the origin of the axial coordinate system and extending to the right along the bottle's axis. Based on structural type, the system is divided into four regions: the body section (C), the transition section (T), the end cap section (H), and the Boss root section (B). These are further subdivided into seven physical sub-regions based on left-right symmetry: Left Boss root section BL (81.853–133.777 mm), Left end cap section HL (133.777–371.083 mm), Left transition section TL (371.083–455.500 mm), Body section C (455.500–5014.500 mm), Right transition section TR (5014.500–5098.917 mm), Right end cap section HR (5098.917–5336.223 mm), and Right Boss root section BR (5336.223–5388.147 mm). The same type of parameters are used on both the left and right ends, and the number and coordinates of measuring points are calculated independently.

[0090] 1. Calculation of regional risk weights

[0091] Based on the baseline results of the thermo-mechanical coupling simulation of the entire curing process, the integral point Mises equivalent stress of the composite material layer was extracted at the control stress time t=19847.35s. Using the average stress of the cylinder section as the benchmark, the stress concentration factors of the cylinder section C, transition section T, end cap section H, and Boss root B were calculated as follows: .

[0092] For the initial application scenario of a new product without historical production data, a simulation perturbation method was used to calculate the curing and molding risk coefficient. 200 sets of process perturbation samples were constructed, with perturbation amounts including furnace temperature boundary ±2℃, convective heat transfer coefficient ±10%, layup thickness ±5%, and kinetic pre-exponential factor ±5%. The maximum temperature difference in a region and the curing degree non-uniformity exceeding the threshold were considered risk events, and the statistically obtained risk occurrence rates for each region were 0.08, 0.16, 0.12, and 0.20, respectively. Based on the cylinder section, the curing and molding risk coefficients for cylinder section C, transition section T, end cap section H, and Boss root section B were calculated as follows: .

[0093] For the v-th physical sub-region, first multiply the structural stress concentration coefficient of the corresponding structural type of the sub-region by the curing and molding risk coefficient to obtain the original regional risk weight of the sub-region; then normalize the original risk weights of all 7 physical sub-regions to calculate the normalized regional risk weight of each sub-region. The sum of the normalized weights of all sub-regions is 1, which is used for the proportional allocation of subsequent measurement point quotas.

[0094] The normalized regional risk weights of each sub-region are substituted into the nonlinear deployment response function to calculate the deployment response value for each sub-region. This deployment response function integrates three mechanisms: basic risk response, local over-density suppression, and curvature enhancement in medium- and high-risk areas. All function parameters use pre-calibrated typical engineering values: risk response index p = 0.72, suppression index q = 0.38, baseline constant A = 0.28, and local over-density suppression intensity coefficient λ... b Take 0.35, curvature enhancement coefficient β b Take 0.45.

[0095] 2. Deployment of response function and allocation of measurement points

[0096] Set the total number N of sensor units that can be deployed per bottle. tot =32. Each physical sub-region is first configured with at least 2 sensing units to ensure basic observability. The remaining 18 quotas are allocated proportionally to the response values ​​using the maximum remainder method. The calculation results are shown in Table 1. Table 1 Physical sub-region distribution parameters

[0097] Open straight paths use equal interval center points, while closed circular paths use equal angle distribution. The maximum remainder distribution method can ensure that the total number of sensing units in each area is exactly equal to the set total number, avoiding the total deviation caused by rounding to each area.

[0098] 3. Measurement point coordinates and stratigraphic implementation

[0099] All sensing units are located on the outer surface of the third composite material layup. The radius of this interface in the cylinder section is 340.904 mm, and the specific coordinates are as follows: like Figure 3 As shown, the X coordinates of the cylinder sections C1 to C4 are 1025.375, 2165.125, 3304.875, and 4444.625 mm, respectively. like Figure 3 and Figure 4 As shown, the left and right transition sections TL and TR are located at x=413.292mm and x=5056.709mm respectively, with a path radius r of 335.994mm and azimuth angles of the measuring points of 0°, 72°, 144°, 216° and 288° respectively. like Figure 3 and Figure 5 As shown, the (x,r) coordinates of the left end caps HL1 to HL4 are (332.002, 315.539), (259.664, 277.114), (198.362, 222.782), and (151.563, 155.563) mm, respectively; the right end cap is symmetrical about the mid-plane of the bottle with x = 2735 mm. like Figure 3 and Figure 6 As shown, the circular paths of the left and right Boss roots BL and BR are located at x=133.777mm and x=5336.223mm respectively, with a path radius of 118.566mm. The azimuth angles of the measuring points are 0°, 72°, 144°, 216°, and 288°.

[0100] The three-dimensional coordinates of the circumferential measuring points can be obtained through trigonometric function conversion. Each embedded sensing unit contains one main direction strain grating, one vertical direction strain grating, and one strain-isolated temperature compensation grating. The 32 embedded sensing units contain a total of 96 embedded gratings. An additional 32 surface temperature gratings are attached to the corresponding positions on the outer surface of the bottle to calculate the internal and external temperature difference, totaling 128 gratings. These are networked using 8 demodulation channels, with each channel multiplexing 16 gratings. During interlayer embedding, after the sensing unit is fixed to the outer surface of the third layup, the cylinder section is covered by the fourth circumferential winding layer to form the interlayer interface. The transition section, end cap section, and Boss root are covered by the fifth spiral winding layer, forming the corresponding interlayer interface. In other words, the sensing unit is fixed to the outer surface of the predetermined layup and covered by the next fiber winding layer that actually covers that area, ensuring that all measuring points are at the interlayer interface and are not damaged by subsequent processes.

[0101] III. Methods for Controlling the Solidification Process of Hydrogen Storage Bottles

[0102] This curing process control method implements online monitoring and closed-loop control of the entire curing process, and the overall process is divided into the following five core steps.

[0103] (I) Initial parameter calibration and model construction

[0104] After the hydrogen storage cylinder with the fiber optic sensor array pre-embedded was hoisted to the curing oven station and the all-fiber optic signal transmission link was connected, initial parameter calibration was performed in a 25°C constant temperature no-load standard environment: The initial center wavelength of all strain sensing gratings is collected as the zero-point reference for subsequent wavelength offset calculations. The inherent temperature sensitivity coefficient K of each temperature compensation grating was calibrated through a stepped isothermal test (30℃, 50℃, 70℃, 90℃, 110℃, with each temperature held for 30 minutes). T ; The interlaminar strain transfer coefficient was calibrated through tensile tests on standard laminates with the same material system and ply structure: Standard laminate specimens with the same ply angle as the hydrogen storage cylinder were prepared, and strain-sensing gratings of the same specifications were embedded in the interlaminar layers. Resistance strain gauges were attached to the corresponding positions on the specimen surface as a reference. Tensile loads were applied in stages within the elastic range, and the wavelength shift of the strain-sensing grating and the strain gauge readings were simultaneously acquired. Using the true strain measured by the strain gauges as a reference, the ratio of the strain calculated by the strain-sensing grating to the true strain was calculated, which is the interlaminar strain transfer coefficient at that winding angle. The corrected actual strain sensitivity coefficient is: , This represents the inherent strain sensitivity of the bare strain sensing grating.

[0105] The reference absolute temperature T0 = 298.15K corresponding to the calibration environment was determined, and the improved multiphysics decoupling model and the material curing kinetic constitutive model were entered to construct a unified calculation reference.

[0106] (II) Raw signal acquisition and temperature calculation

[0107] Start the curing oven rotation mechanism, set the gas cylinder rotation speed to 2 r / min, and start the curing program after confirming the optical path is stable. Throughout the curing process, continuously acquire the raw wavelength signals of the temperature-compensated grating and strain-sensing grating at a sampling frequency of 2 Hz. Based on the wavelength offset of the temperature compensation grating, the real-time temperature change ΔT at each measuring point is calculated, and the real-time absolute temperature T = T0 + ΔT is obtained. The total center wavelength offset Δλ of the strain sensing grating at each measuring point is acquired synchronously. B This is used for subsequent strain decoupling.

[0108] (III) Online analysis of molding status

[0109] A serial, iterative approach is used to sequentially update the degree of cure, decouple the strain, and calculate the internal stress. The real-time absolute temperature is input into the material curing kinetic constitutive model, which is corrected by both interlayer thermal resistance and winding angle. The degree of curing at the previous sampling time is used as the initial state, and the real-time resin curing degree α at the current time is updated by integration. The free chemical shrinkage strain corresponding to the resin crosslinking reaction is calculated based on the real-time resin curing degree α. ; Substituting the total center wavelength shift, temperature change, and chemical shrinkage strain into the multiphysics decoupling model, and subtracting the temperature response and chemical shrinkage response components, the pure mechanical curing strain is obtained. ; By combining the temperature-curing degree coupling modified relaxation modulus, the cumulative molding internal stress σ at each measuring point from the start of curing is calculated using the Boltzmann superposition principle. n (t).

[0110] (iv) Decision-making and issuance of instructions

[0111] 1. Stage Determination

[0112] By combining the current process program segment of the curing oven, the rate of change of the oven temperature setpoint, and the real-time degree of resin curing (three-level joint judgment criteria for curing stage), it is determined whether the current stage is the heating stage, the heat preservation stage, or the cooling stage.

[0113] The specific details of the three-level joint judgment criteria for the curing stage are as follows: (1) Level 1: Basic determination of process procedure section The current operating segment of the curing oven's preset process curve is used as the initial determination basis: the heating segment during program operation is initially classified as the heating stage, the isothermal segment as the heat preservation stage, and the cooling segment as the cooling stage. This level is used to match the production line's conventional process logic, ensuring a basic connection between control and the preset program.

[0114] (2) Second level: Auxiliary verification of furnace temperature change rate

[0115] Based on furnace temperature setpoint T set real-time rate of change dT set / dt performs quantization verification to correct the fuzzy range during the program switching transition period: When dT set When / dt>0.3℃ / min, it is determined to be in the heating phase; When dT set When / dt≤0.1℃ / min, it is determined to be in the heat preservation stage; When dT set When / dt<-0.3℃ / min, it is determined to be in the cooling stage.

[0116] The above thresholds (0.3℃ / min, 0.1℃ / min, -0.3℃ / min) are typical engineering values ​​and can be adjusted according to the temperature control accuracy and process characteristics of different curing ovens. When the real-time rate of change dT... set When / dt is in the range of 0.1~0.3℃ / min or -0.3~0.1℃ / min, the stage determination result of the previous moment shall be used to keep the stage attribute unchanged and avoid frequent stage jumps caused by program fine-tuning.

[0117] (3) Third level: Real-time curing degree final calibration

[0118] The real-time resin curing degree obtained from internal measured data analysis is used as the final calibration basis to resolve the discrepancy between "program operation and actual material reaction". When the real-time resin curing degree α < α g (α) g When the gel point curing degree is reached, it is finally determined to be the heating stage, and the heating stage control strategy is implemented. When α g <α<α target (α) target When the target curing degree is reached, it is finally determined to be the heat preservation stage, and the heat preservation stage control strategy is implemented. When α≥α targetFurthermore, when the process enters the cooling zone, it is ultimately determined to be in the cooling stage, and the cooling stage control strategy is implemented.

[0119] If the process enters the cooling stage but the real-time curing degree has not reached the target value, it is still judged as the heat preservation stage, and the heat preservation stage control strategy continues to be implemented until the curing degree reaches the target and then it is switched to the cooling stage; if the real-time curing degree has reached the target but the process is still in the heat preservation stage, the heat preservation stage judgment is maintained, and the stage attribute is switched synchronously after the program switches to the cooling stage.

[0120] (4) Conflict resolution rules

[0121] When the conclusions of the three-level criteria are inconsistent, the judgment result of the real-time resin curing degree shall be the final effective basis to ensure that the control always matches the actual curing reaction process inside the material and avoid the control failure due to the disconnect between the program and the actual reaction.

[0122] 2. Regional Aggregation

[0123] The single-point calculation results of each measuring point are aggregated at the regional level to obtain the average absolute temperature, average resin curing degree, average internal stress and maximum internal stress of each region.

[0124] 3. Deviation Calculation

[0125] The difference between the regional state quantity and the current stage's curing degree target trajectory and allowable internal stress limit is used to obtain the measured curing degree deviation and measured internal stress deviation. After the physical occupancy equivalent correction of the sensing unit, the overall molding state deviation is obtained by combining the regional risk weight.

[0126] 4. Instruction Generation

[0127] Matching the multi-objective control strategy of the current stage, and under the constraints of the upper and lower limits of preset process parameters and the maximum rate of change, incremental adjustment commands for heating rate, holding temperature, holding time or cooling rate are generated and sent to the curing oven execution control unit for execution.

[0128] (v) Closed-loop feedback and iterative optimization

[0129] After the process parameters are adjusted, the sensor signals are continuously collected and returned to the (ii)-(iv) cycle verification in this method. The deviation closed-loop iteration mechanism is used for incremental fine-tuning. The adjustment range of the single heating and cooling rate does not exceed 0.2℃ / min, so as to avoid defects caused by drastic process fluctuations.

[0130] After solidification, the monitoring data, process parameters and finished product performance data of the whole process are stored in the process data iteration unit. The model parameters and control thresholds are updated by the recursive parameter identification algorithm for the control and optimization of the next batch.

[0131] IV. Detailed Solution Principles and Derivation of the Core Model

[0132] (I) Constitutive Model of Material Solidification Kinetics

[0133] 1. Basic Form

[0134] The curing reaction rate is described using a dual-mechanism autocatalytic kinetic model, incorporating corrections for winding angle thermal conductivity and interlayer thermal resistance. The calculation formula is as follows: ; The physical meanings of each parameter in the formula are as follows: dα / dt: Resin curing reaction rate, characterizing the change in degree of curing per unit time; A1, A2: Pre-exponential factors corresponding to the dual reaction mechanism, which correspond to the autocatalytic reaction mechanism before gelation and the diffusion-controlled reaction mechanism after gelation, respectively. ΔE1, ΔE2: Activation energies of the two-reaction mechanism, in J / mol; R: Universal gas constant, valued at 8.314 J / (mol) K); T: Real-time absolute temperature at the measuring point, in K, calculated from T=T0+ΔT; α: Real-time resin curing degree at the current moment, with a value range of [0,1]; m, n: order of the curing reaction, describing the order characteristics of the autocatalytic reaction; f(θ): The winding angle thermal conductivity correction function, where the independent variable θ is the fiber winding angle at the measuring point. It is used to correct for local thermal conductivity differences caused by fiber anisotropy. The larger the winding angle, the weaker the continuity of the fiber along the heat conduction direction, and the smaller the thermal conductivity correction coefficient. The specific function form is: f(θ) = 1 - k θ ·sin 2 θ, k θ The thermal conductivity attenuation coefficient ranges from 0.08 to 0.15. A larger winding angle weakens the continuity of the fiber along the heat conduction direction, resulting in a smaller conversion factor for the local temperature relative to the furnace temperature. The corrected local actual temperature is substituted into the kinetic equation to calculate the reaction rate, ensuring that the degree of curing calculation accurately reflects the true heat transfer characteristics of the thick-walled wound structure.

[0135] γ h Interlayer thermal resistance correction factor, with a value of 0.85~0.95, is used to correct the overall heat transfer attenuation caused by interlayer interface thermal resistance in thick-walled wound structures.

[0136] 2. Solution Method

[0137] The constitutive model of material solidification kinetics is solved using a recursive integral method: at the previous sampling time t k-1 Curing degree α k-1 As the initial condition, based on the real-time absolute temperature T at the current time k.k Calculate the real-time curing rate of the reaction (dα / dt) k The current solidification degree is updated by first-order forward difference: ; In the formula, Δt is the sampling time step.

[0138] During the calculation process, the gel point is identified simultaneously (when the degree of curing reaches α). g The glass transition point and the glass transition point are key thresholds for dividing the curing stage.

[0139] (II) Improved Multiphysics Decoupling Model

[0140] The wavelength shift of the strain sensing grating is simultaneously affected by the interaction of temperature, mechanical strain, and chemical shrinkage strain. To accurately separate the purely mechanical curing strain involved in the formation of internal stress, this invention, based on the independent response characteristics of a temperature compensation grating and a strain sensing grating arranged at the same location, constructs a multi-physics decoupling model through step-by-step decoupling. The complete derivation process is as follows: 1. Temperature response of temperature-compensated grating The temperature-compensated grating employs strain-isolated encapsulation and responds only to temperature changes; its normalized center wavelength shift is linearly related to the temperature change. ; The physical meanings of each parameter in the formula are as follows: Δλ T The center wavelength offset of the temperature compensation grating, in pm, is acquired in real time by the fiber optic signal demodulation unit. λ T0 The initial center wavelength of the temperature compensation grating under constant temperature and no-load standard environment, in pm, is entered into the system during the initial parameter calibration stage. K (T,T) The inherent temperature sensitivity coefficient of the temperature-compensated grating, in pm / (pm (℃), obtained by step temperature test calibration; ΔT: Real-time temperature change at the measuring point, in °C.

[0141] The real-time temperature change at the measuring point can be calculated from the above formula: ; Adding the temperature change to the reference absolute temperature T0 yields the real-time absolute temperature T at the measuring point, which is then input into the material curing kinetic constitutive model for degree of curing calculation.

[0142] 2. Multi-field coupling response of strain sensing grating

[0143] The strain-sensing grating simultaneously senses temperature changes and the total strain along the grating axis. The total strain includes two parts: the purely mechanical curing strain that generates internal stress, and the free chemical shrinkage strain that does not generate internal stress. Its normalized total center wavelength offset satisfies: ; The physical meanings of each parameter in the formula are as follows: Δλ B : Total offset of the center wavelength of the strain sensing grating, in pm, acquired in real time by the demodulation unit; λ B0 : The initial center wavelength of the strain sensing grating under constant temperature and no load standard environment, in pm, is entered during the initial parameter calibration stage; K (TB) The inherent temperature sensitivity coefficient of the strain sensing grating, in pm / (pm (℃), obtained by step temperature test calibration; p e The effective elastic-optic coefficient of optical fiber is dimensionless and is an inherent material parameter of silica optical fiber. It is used to characterize the modulation effect of the elastic-optic effect on wavelength shift. Interlaminar strain transfer coefficient, dimensionless, obtained by standard laminate tensile test, used to correct the strain transfer loss between fiber and resin under interlaminar embedding conditions, and to ensure the authenticity of strain calculation. Pure mechanical solidification strain is dimensionless; it is the effective strain generated by the constrained deformation of the structure and participates in the formation of internal stress, and is the core input for internal stress calculation. Free chemical shrinkage strain is dimensionless and is generated by the volume shrinkage of the resin cross-linking reaction. It can occur freely, does not form internal stress, and must be removed from the total strain.

[0144] 3. Pure mechanical solidification strain decoupling derivation

[0145] Substituting the formula for the temperature change ΔT into the strain sensing grating response formula, and rearranging the terms, we obtain the pure mechanical curing strain: ; Define the actual strain sensitivity coefficient after interlaminar strain transfer correction: ; Simplifying the above formula, we obtain the following standard form: ; The physical meanings of each parameter in the formula are as follows: K T The inherent temperature sensitivity coefficient of the grating is calibrated by a stepped isothermal test. : Actual strain sensitivity coefficient after interlayer strain transfer correction.

[0146] The above derivation process achieves the separation of multiple physical quantities such as temperature, chemical shrinkage, and mechanical strain, ensuring the accuracy of subsequent internal stress calculations.

[0147] (iii) Calculate the cumulative value of forming internal stress

[0148] The evolution of internal stress during the curing process of composite materials conforms to the thermoviscoelastic constitutive law. The core principle is based on the Boltzmann superposition principle, which accumulates the residual contribution of the historical strain increment throughout the curing process after stress relaxation. Before calculation, the effective constrained strain that truly participates in the formation of internal stress needs to be separated from the strain observed by the fiber grating, and then the integral solution is completed by combining the nonlinear relaxation modulus coupled with temperature and degree of curing.

[0149] 1. Separation of effectively constrained strain

[0150] The strain directly measured by the strain sensing grating includes not only constrained deformation that generates internal stress, but also freely occurring thermal deformation and resin chemical shrinkage deformation. Directly substituting these into stress calculations will lead to overestimation of the results. Therefore, the observed strain at the nth measuring point at time t... It can be decomposed into: ; In the formula, Let be the elastic strain at the nth measuring point at time t. Let be the viscoelastic strain at the nth measuring point at time t. Together, these two constitute the effective constrained strain that participates in the formation of internal stress. Let be the free thermal strain at the nth measuring point at time t. The free chemical shrinkage strain at time t for the nth measuring point is based on the resin crosslinking. Both are free deformations that do not generate internal stress.

[0151] The free thermal strain is expressed as: ; In the formula, T n (t) represents the absolute temperature of the nth measuring point at time t; α n (t) represents the degree of resin curing at the nth measuring point at time t. Let α be the equivalent thermal expansion coefficient at the nth measuring point. This is a bivariate function, where T is the instantaneous temperature, indicating that the thermal expansion coefficient changes with temperature itself; the independent variable α... n (t) represents the real-time degree of resin curing, indicating that the coefficient of thermal expansion also varies with the degree of resin cross-linking (the higher the degree of curing, the tighter the cross-linking of the matrix, and the lower the coefficient of thermal expansion); dT represents the temperature element.

[0152] Chemical shrinkage strain is expressed as: ; In the formula, β ch,n γ represents the fully cured chemical shrinkage coefficient corresponding to the nth measurement point, i.e., the maximum chemical shrinkage strain when the resin reaches a degree of curing of 1 (fully cross-linked). It is an inherent material parameter of the resin matrix, obtained through material shrinkage rate testing. n γ is the nonlinear exponent of chemical shrinkage at the nth measurement point, dimensionless, used to describe the nonlinear law of chemical shrinkage development with degree of cure: if γ n >1, the shrinkage effect is mainly concentrated in the later stage of curing; if γ n <1 indicates faster shrinkage development in the early curing stage; this parameter is obtained by differential scanning calorimetry (DSC) or volume shrinkage rate test of the resin system.

[0153] Therefore, the total effective strain at time t of the nth measuring point involved in the internal stress calculation for: ; By separating the strains as described above, we can avoid miscalculating free thermal expansion and free chemical contraction as constrained strain, thus ensuring the physical accuracy of internal stress calculation.

[0154] 2. Discretization Derivation of Internal Stress Integral

[0155] Divide the time from the start of curing (0) to the current time (t) into multiple time intervals of unequal length, with the start time of the kth time interval being... The time interval length of the kth time period is The effective strain increment at the nth measuring point within this time period for: ; In the formula, For the nth measurement point at Total effective strain at any given moment; For the nth measurement point at Real-time, effective response at every moment.

[0156] When the time step When the time interval is sufficiently small, the effective strain increment can be approximated by the strain change rate at the beginning of the time interval as follows: ; In the formula, For the nth measurement point at Total strain at time The first derivative of represents the amount of strain change per unit time.

[0157] This method employs a strategy of dividing time steps unequally: shortening the time step in the gel point, glass transition region, and regions with large curvature of the strain curve to accurately capture the inflection point and arc-shaped characteristics of stress evolution; and increasing the time step in regions with gentle state changes to improve computational efficiency.

[0158] exist The strain increment generated at time t will undergo stress relaxation during the curing process, and the remaining stress contribution at the current time t is: ; In the formula, Let be the residual stress at the nth measuring point at time t; For the nth measurement point at The degree of resin curing at any given time; For the nth measurement point at The absolute temperature at any given moment.

[0159] G(·) is the temperature-curing degree coupled modified relaxation modulus. Its physical properties are as follows: strain increments that occur earlier have longer relaxation times and smaller current residual contributions; strain increments that occur later have shorter relaxation times and larger retained stress contributions.

[0160] The current cumulative value of forming internal stress is equal to the sum of the residual stresses of all historical strain increments: ; When the maximum time step approaches zero, the above discrete summation converges to a continuous integral form: ; The physical essence of this continuous integral form is to accumulate the portion of each infinitesimal strain increment since the start of curing that has not yet relaxed at the current moment, rather than using a simple linear correspondence between the current strain and the current stress. This is more in line with the thermoviscoelastic mechanical laws of the composite material curing process.

[0161] 3. Temperature-Cure Degree Coupled Nonlinear Relaxation Modulus Construction

[0162] Conventional methods often simplify the relaxation modulus to a piecewise linear form, which fails to accurately describe the nonlinear transition characteristics of the gel region and glass transition region. This invention employs a continuous function to construct a temperature-curing degree coupled nonlinear relaxation modulus, the specific process of which is as follows: First, based on the time-temperature equivalence principle and the degree of cure equivalence principle, a temperature-curing-degree coupling reduction time is constructed: ; In the formula, For the nth measurement point from time... The reduced time to the current time t; Let be the degree of resin curing at the nth measuring point at time s; Let be the absolute temperature of the nth measuring point at time s. The temperature-curing degree shift factor at the nth measuring point. for The reciprocal of.

[0163] Based on the reduced time, a nonlinear relaxation modulus of the following form is constructed: ; In the formula, for The unit strain increment generated at time t, after After a relaxation period, the remaining stress contribution at the current time t.

[0164] The long-term equilibrium modulus (rubber state modulus) at the nth measurement point is the modulus value that the material ultimately retains when the relaxation time approaches infinity. It corresponds to the high-elasticity equilibrium modulus after the resin is fully cross-linked and is the lower limit of the modulus during the relaxation process.

[0165] The instantaneous modulus (glass modulus) at the nth measuring point is the initial modulus when the relaxation time approaches 0. It corresponds to the material stiffness at the instant the strain is applied and is the upper limit of the modulus during the relaxation process. Its value is much greater than the long-term equilibrium modulus.

[0166] This represents the total magnitude of modulus decay at the nth measurement point, which is the modulus component that gradually decreases over time during the relaxation process.

[0167] a n b n The shoulder shape parameters at the nth measurement point control the growth rates of the numerator and denominator, which together determine the position and width of the shoulder platform.

[0168] p n q n The shoulder power parameter at the nth measurement point controls the curvature of the shoulder curve; it typically satisfies 0. n ≤p n This ensures that the overall modulus exhibits a monotonic decreasing trend.

[0169] The reduced time (equivalent relaxation time) at the nth measurement point is converted from the real physical time to the equivalent time under a unified reference state using a temperature-curing degree shift factor. This incorporates the effects of temperature and curing degree on the relaxation rate into the time scale, allowing the modulus to be measured using only the time scale. It is expressed as an independent variable.

[0170] ​The characteristic relaxation time of the nth measurement point represents the time scale of the main relaxation process.

[0171] The modulation amplitude coefficient at the nth measurement point is much smaller than 1. It only makes a small arc correction to the modulus curve and does not change the overall attenuation trend.

[0172] This is the frequency parameter of the logarithmic sine term at the nth measurement point, controlling the density of the arc-shaped oscillation. Here... This is a shape modulation parameter, which shares the same symbol as the degree of curing α(t) but has a completely different physical meaning. It can be distinguished by whether or not it includes a time independent variable.

[0173] The phase parameter of the nth measurement point controls the position of the arc inflection point on the reduced time axis, which is used to match the corresponding time of the gel point and the glass transition point.

[0174] The relaxation modulus constructed in this invention uses a single continuous function to simultaneously express the shoulder, the main attenuation region, and the local arc-shaped transition region, without having to divide the data into several straight line segments, which better reflects the actual stress relaxation law of composite material curing.

[0175] 4. Curvature-weighted identification method for model parameters

[0176] To address the structural characteristics of abrupt curvature changes in the end cap of hydrogen storage cylinders, the experimental data at the arc-shaped abrupt change points along the outer edge of the composite material layer are prone to distortion, and the effective sample ratio for key inflection points such as the gel region and glass transition region is low. This invention employs a curvature-weighted nonlinear fitting method to identify model parameters for the relaxation modulus, avoiding the overwhelming of key arc-shaped feature points by a large amount of data in the smooth region, thus improving the overall accuracy of parameter identification throughout the entire process. The specific details are as follows: The optimal parameter set of the nth measuring point Solve by minimizing the sum of the weighted sum of squared residuals and the regularization term: ; In the formula, The vector of relaxation modulus parameters to be identified at the nth measurement point includes all parameters to be calibrated, such as instantaneous modulus, long-term modulus, and shape factor. The measured relaxation modulus value of the x-th test data point at the n-th measurement point; The reduction time for the x-th data point at the nth measurement point; These are the model calculation values ​​corresponding to the reduced time. λ is the curvature fitting weight for the x-th data point; r This is the regularization coefficient, used to suppress parameter overfitting; for L2 regularization term for the intermediate parameter; N dThis represents the total number of test data points.

[0177] The fitting weight for each data point is determined by the curvature of the curve at that point, and the formula for calculating the curvature weight is: ; In the formula, κ is the curvature modulation coefficient, which is used to control the magnitude of curvature enhancement on the weight. , These are the measured relaxation modulus values ​​at the (x-1)th and (x+1)th experimental data points of the nth measurement point, respectively. Numerator term The absolute value of the second difference of the relaxation modulus test curve represents the curvature of the curve at the corresponding data point; a very small positive number ε is added to the denominator to avoid division by zero when the modulus value is close to zero.

[0178] The core physical logic of curvature weight calculation is as follows: the larger the absolute value of the second-order difference, the closer the test point is to the arc transition point or local inflection point, the more critical the corresponding data feature, and the higher the fitting weight. This weighting mechanism can effectively increase the contribution ratio of key inflection points such as the gel region and glass transition region in the fitting objective function, preventing a large amount of data in the smooth change range from "drowning" the features of a small number of key arc points, and ensuring that the identified parameters can accurately reproduce the core features of the stress relaxation curve, such as the shoulder and inflection point.

[0179] During the parameter fitting process, the following physical constraints are applied simultaneously to ensure that the identification results conform to the thermoviscoelastic mechanical properties of cured composite materials: (1) Modulus order of magnitude constraint: The instantaneous relaxation modulus is greater than the long-term equilibrium modulus, and both are positive, i.e. .

[0180] (2) Positive constraints on shape parameters: All shape parameters and feature time parameters are positive values, i.e. .

[0181] (3) Modulation amplitude constraint: The amplitude coefficient of the arc-shaped modulation term is a positive number less than 1, ensuring that the modulation effect is only a local correction and does not change the overall modulus attenuation trend, that is .

[0182] This curvature-weighted identification method can effectively improve the fitting accuracy of key inflection points such as the gel region and glass transition region, preventing a large amount of flat region data from "submerging" a small number of key arc points. At the same time, physical constraints are applied to the parameters during the fitting process (such as non-negative modulus, monotonically decaying, etc.) to ensure that the identification results conform to the laws of mechanics.

[0183] (iv) Aggregation of regional feature values

[0184] Before calculating the overall deviation, the results of individual point calculations are aggregated into regional feature quantities: 1. Regional average of temperature and degree of cure The average value of the region is calculated using the effective area weighting method of measuring points, which is suitable for layout schemes with uneven measuring point density.

[0185] 2. Polymerization with dual eigenvalues ​​of molding internal stress

[0186] Simultaneously calculate the regional average and regional maximum values: Regional average internal stress: Calculated using effective area weighting, it characterizes the overall stress level of the region and is used in the overall deviation calculation; Maximum internal stress in the region: The maximum value of all measuring points in the region is taken as the criterion for the risk of local stress concentration and is used to trigger the control threshold.

[0187] (v) Equivalent correction of physical space occupied by sensing unit

[0188] Pre-embedded sensing units disrupt local fiber continuity and alter local stiffness, amplifying the measured strain and stress deviations. By quantifying the degree of disturbance using an equivalent physical occupancy variable, the additional contribution of the occupancy effect is subtracted from the measured deviation to obtain the true deviation without sensing interference. The specific correction process is as follows: 1. Fraction of occupied volume Calculate the volume fraction of the sensing unit in each region. The volume fraction is the ratio of the equivalent volume of a single sensing unit to the effective volume of the corresponding composite material in the region.

[0189] 2. Physical occupancy equivalent variables

[0190] Combined with the fraction of occupied volume The sensor unit layout direction, real-time resin curing degree, gel point curing degree, real-time absolute temperature and glass transition temperature are used to construct the physical occupancy equivalent variables for each region: ; In the formula, Ω i (t) represents the physical occupancy equivalent variable of the i-th region at time t; η represents the volume fraction of the sensing unit within the i-th region; i A direction modulation coefficient is assigned to the sensing unit in the i-th region to quantify the enhancement of the occupancy disturbance by the fiber angle. This coefficient can be obtained by establishing a finite element model of a laminate containing sensing units with different angles, comparing the stress deviation with and without sensing units, and then performing nonlinear fitting. i λ is the angle between the arrangement direction of the sensing units in the i-th region and the direction of the main supporting fiber in that region; o,i γ represents the saturation coefficient of the i-th region, which can be obtained by setting multiple sets of simulation samples with different saturation volume fractions and fitting the saturation trend of the disturbance intensity; iThe correction coefficient for the curing state of the i-th region can be obtained by comparing the deviation of the occupancy disturbance before and after the gel point and fitting the correlation between the degree of curing and the disturbance intensity; α i (t) represents the real-time resin curing degree of the i-th region at time t; α g,i ρ represents the degree of gel point curing in the i-th region at time t; i The correction coefficient for the glass transition region of the i-th region can be obtained by comparing the occupancy perturbation deviation inside and outside the glass transition region and fitting the perturbation enhancement magnitude of the transition region.

[0191] Geometric direction terms: This describes the fundamental influence of the angle between the occupied volume and the fiber direction on the disturbance intensity. It reflects the rule that the disturbance is strongest when the sensing unit is perpendicular to the fiber and weakest when it is parallel; the saturation term in the denominator describes the nonlinear saturation characteristics of the occupancy effect.

[0192] Curing process item: This describes the phenomenon that the matrix stiffness increases after the gel point, and the occupancy disturbance effect strengthens with increasing degree of curing.

[0193] Glass transition modulation term: It describes the characteristics of drastic changes in matrix stiffness during the glass transition range, where the occupancy disturbance reaches its peak.

[0194] The normalized variable for glass transition is: ; In the formula, ζ i (t) represents the normalized variable for the glass transition of the i-th region at time t; T i (t) represents the real-time absolute temperature of the i-th region at time t; T gi (t) represents the real-time glass transition temperature of the i-th region at time t; ΔT g is the half-width of the glass transition influence range; clip[·] is the cutoff function that restricts the variable to the interval [0,1]; exp[·] is the exponential function with the natural constant as the base.

[0195] 3. Correction for true deviation

[0196] Based on the physical occupancy equivalent variable and the pre-calibrated occupancy deviation sensitivity coefficient, the measured curing degree deviation and measured internal stress deviation of each region are converted into the true state deviation without sensor unit interference: ; ; In the formula, Δα represents the actual degree of curing deviation of the i-th region at time t. i(t) represents the measured curing degree deviation of the i-th region at time t; c α,i The sensitivity coefficient of the occupancy deviation corresponding to the curing degree deviation of the i-th region can be obtained by establishing a regional finite element model containing sensing units, using the physical occupancy equivalent variable as the independent variable, and fitting the slope of the curing degree deviation difference. Δσ represents the actual internal stress deviation of the i-th region at time t. i (t) represents the measured internal stress deviation of the i-th region at time t; c σ,i The occupancy deviation sensitivity coefficient corresponding to the stress deviation in the i-th region can be obtained by establishing a regional finite element model containing sensing elements, using the physical occupancy equivalent variable as the independent variable, and fitting the slope of the internal stress deviation difference.

[0197] The occupancy deviation sensitivity coefficient was pre-calibrated using the finite element benchmarking method: two sets of finite element models were established, one with sensing elements and one without. Under the same temperature load and curing boundary conditions, the curing degree deviation and internal stress deviation of the two sets of models were calculated; the physical occupancy equivalent variable was used. Using the deviation difference as the dependent variable and performing a linear fit, the slope of the fit is the sensitivity coefficient c of the curing degree occupancy deviation in the corresponding region. α,i Sensitivity coefficient c of internal stress occupancy deviation σ,i .

[0198] (vi) Calculation of deviation in overall molding state

[0199] The actual curing degree deviation and actual internal stress deviation of each region are weighted and summed using regional risk weights to obtain the overall molding state deviation, which includes both curing degree deviation and internal stress deviation. ; ; ; In the formula, D(t) represents the overall forming state deviation of the hydrogen storage cylinder at time t; λ α λ is the priority weighting coefficient for curing degree deviation. σ e is the priority weighting coefficient for internal stress deviation; α,i (t) represents the relative curing degree deviation of the i-th region at time t; δ α,i e is the allowable deviation threshold for the degree of cure of region i in the current curing stage; σ,i (t) represents the relative internal stress deviation of the i-th region at time t; δ σ,i Let be the allowable deviation threshold of internal stress in region i during the current curing stage; [·] += max(x,0) is the positive part cutoff function. max is the maximum value operation. The deviation is only included when the positive internal stress exceeds the safety limit. When the stress is below the threshold, this item is 0 to avoid the negative deviation from offsetting the curing degree risk.

[0200] The overall forming state deviation D(t) is used to quantify the overall deviation of the current forming state from the target trajectory, serving as the total gain coefficient for phased control: the larger the deviation value, the greater the adjustment range of the process parameters; when the deviation value is lower than the set threshold (e.g., 0.005), the state is judged to be stable, and active control is not triggered. This deviation is also connected to the closed-loop iteration module for effect evaluation and incremental fine-tuning after each round of control, ensuring that the control process is smooth and controllable.

[0201] V. Implementation of a phased, multi-objective control strategy

[0202] Different control targets and parameters are matched for different curing stages, as detailed below: (I) Regulation during the warming phase Core objectives: To improve temperature field uniformity and suppress the rapid accumulation of internal stress during molding; Control logic: The heating rate adjustment formula is driven by both temperature difference and internal stress growth rate. ; In the formula, v T,new This represents the adjusted actual temperature rise rate, expressed in °C / min, and is the final temperature control parameter sent to the curing oven's control unit. T,0 The preset reference heating rate, i.e., the original heating rate set in the initial process curve, serves as the reference value for rate adjustment. T ΔT is the gain coefficient for temperature difference regulation, dimensionless, controlling the adjustment range of the heating rate by the temperature difference deviation; the larger the value, the more significant the rate reduction at the same temperature difference. dev This represents the measured internal and external temperature difference in the current critical area (head and root of the Boss), i.e., the temperature difference between the measuring points inside the structure and the surface measuring points, in °C. ΔT lim This is a preset safe threshold for the internal and external temperature difference, in °C. If this threshold is exceeded, the heating rate will be reduced. The gain coefficient for strain growth rate regulation is dimensionless and controls the adjustment range of strain growth deviation on the heating rate. The measured pure mechanical strain growth rate in the current key area The deviation indirectly characterizes the rate of increase of internal stress. This is a preset safety threshold for strain growth rate; exceeding this threshold will trigger a reduction in the heating rate.

[0203] When the calculated heating rate is lower than the set lower limit (0.5℃ / min), the intermediate insulation platform at the corresponding temperature is automatically inserted; after the internal and external temperature difference falls back to within the threshold and the internal stress growth rate falls back to the safe range, the preset heating process is resumed.

[0204] (II) Control during the heat preservation stage

[0205] Core objectives: uniformity of curing and relaxation of internal molding stress; Control logic: The insulation time adjustment formula is based on internal stress. ; In the formula, Δt hold This refers to the extended insulation time, which is the additional insulation time required beyond the preset insulation time, measured in minutes. α α is a dimensionless weighting coefficient for adjusting the degree of cure, controlling the contribution of the cure gap to the extended cooking time. target α represents the target degree of cure during the insulation stage, while α represents the minimum degree of cure threshold required by the process. current This is the real-time calculated value of the current overall average degree of curing. k represents the current curing reaction rate, i.e., the rate of change of the degree of curing over time, characterizing how fast the curing reaction progresses. σ The stress regulation weighting coefficient is dimensionless and controls the proportion of the contribution of internal stress overshoot to the duration of extension.

[0206] σ max This represents the maximum forming internal stress value in the current critical area (head, Boss root), in MPa.

[0207] σ safe This is a preset internal stress safety threshold, measured in MPa. If the stress is exceeded, the insulation period needs to be extended to promote stress relaxation.

[0208] t0 is the preset baseline insulation time, in minutes, which serves as the baseline scale for proportional adjustment.

[0209] When the maximum internal stress in the critical area (head and root of the boss) continues to be higher than the permissible threshold and shows no trend of decline, extend the heat preservation time and use the high elastic viscoelasticity of the resin to promote stress relaxation; when the curing degree reaches the standard and the internal stress falls back to the safe range, end the heat preservation stage and enter the cooling stage.

[0210] (III) Regulation during the cooling phase

[0211] Core objective: To suppress the freezing caused by the superposition of internal stress during molding; Control logic: The cooling rate adjustment formula using internal stress-temperature coupling is adopted. ; In the formula, vcool,new This represents the adjusted actual cooling rate, expressed in °C / min, and is the final temperature control execution parameter issued. cool,0 The preset baseline cooling rate, i.e., the original cooling rate set in the initial process curve, serves as the baseline value for rate adjustment. η T (T) is the temperature range correction coefficient, a piecewise function of the current temperature T; it takes a value less than 1 within the glass transition temperature range, with the coefficient decreasing and the cooling rate decreasing as the temperature approaches the midpoint of the transition region; it takes a value of 1 within the normal cooling range, without additional adjustment. η α (α) is the curing degree correction coefficient, which is a function of the current curing degree α. The higher the curing degree, the more fully the matrix crosslinks and the greater the stiffness. The internal stress is more easily frozen. The smaller the coefficient, the lower the cooling rate.

[0212] When the temperature drops to the glass transition zone and the internal stress is higher than the threshold, the slow cooling platform is automatically inserted. The higher the internal stress, the longer the slow cooling time. After the product has passed through the glass transition zone uniformly and the internal stress growth rate has dropped back to the safe range, the normal cooling rate is restored until it is taken out of the furnace.

[0213] VI. Experimental Verification

[0214] This embodiment takes a Type IV carbon fiber wound composite hydrogen storage cylinder with a rated working pressure of 70MPa as the implementation object. The aforementioned deployment and control methods are used to complete the online monitoring and intelligent control of the entire curing process. The specific implementation process is as follows.

[0215] (I) Offline playback verification of dynamics

[0216] To verify the logical effectiveness and parameter rationality of the phased control strategy, offline playback comparison tests were first conducted based on the material curing kinetic constitutive model to clarify the input, output and triggering mechanism of the control method.

[0217] 1. Benchmark process and kinetic model

[0218] The baseline curing regime used in the test was as follows: the temperature was increased from 20°C to 130°C at a rate of 9°C / min, with a heating time of approximately 733s; the temperature was held at 130°C for 18000s (5h); and then the temperature was reduced from 130°C to 20°C in 21600s, with an average cooling rate of approximately 0.3056°C / min; the total process time was approximately 40333s.

[0219] The resin curing reaction rate is described by a dual-mechanism autocatalytic kinetic model. The reaction rate is determined by the pre-exponential factor, the activation energy of the reaction, the current degree of curing, and the reaction order. The change in degree of curing per unit time is calculated in combination with the real-time Celsius temperature at the measuring point. The initial degree of curing is taken as 0.01%, and the time unit is seconds.

[0220] The temperature response of each region is simulated using a first-order thermal inertia model, which means that the rate of change of the region temperature is directly proportional to the difference between the furnace temperature setpoint and the current region temperature, and inversely proportional to the thermal time constant of that region; the thermal time constants of the cylinder section, transition section, head section, and Boss root are 60 seconds, 90 seconds, 100 seconds, and 120 seconds, respectively.

[0221] 2. Phased Regulation Triggers and Results

[0222] During playback, positive deviation control logic is used, which triggers adjustment only when the deviation exceeds the threshold, and does not adjust the process parameters when the deviation is below the threshold.

[0223] During the heating phase, a rate adjustment logic driven by both temperature difference and strain is adopted: First, the over-limit ratio of the internal and external temperature difference and the pure mechanical strain growth rate relative to the preset threshold is calculated separately, and only the positive over-limit part is retained. When the threshold is not exceeded, the corresponding item is set to zero. Then, based on the preset benchmark heating rate, the heating rate is adjusted down according to the weighted result of the two over-limit ratios, and the final rate is limited to the preset upper and lower limit range to avoid rate fluctuations exceeding the allowable range of the process.

[0224] The insulation stage adopts a time adjustment logic driven by two parameters: curing degree and stress. The time increment corresponding to the curing degree gap and the maximum internal stress exceedance is calculated separately. The total insulation extension is obtained by superimposing the two, and the extension is limited between zero and the preset maximum extension time to avoid the insulation time deviating too much from the baseline process.

[0225] like Figure 7 As shown, the adjusted process curve obtained from offline playback forms a clear comparison with the baseline process: Figure 7 This diagram illustrates the comparison of oven temperature curves between the baseline curing regime and the staged control process. The horizontal axis represents curing time in minutes, and the vertical axis represents temperature in degrees Celsius (°C). The solid blue line represents the baseline oven temperature setting curve, the dashed red line represents the offline playback control oven temperature setting curve, and the gray dotted line represents the internal temperature curve of the Boss region. Four key control nodes are marked in the diagram: ① At 90°C, the heating rate is reduced from 9°C / min to 6°C / min; ② At 115°C, a 15-minute temperature equalization plateau is inserted; ③ At 130°C, the total holding time is extended from 5 hours to 9 hours; ④ In the glass transition range of 110~90°C, the cooling rate is reduced to 0.25°C / min. The lower right corner of the diagram shows the final effect of the kinetic playback: the average final degree of cure in the four regions increased from 0.877 in the baseline process to 0.960.

[0226] The control trigger logic corresponds perfectly to the markings in the diagram: when the furnace temperature rises to 90℃, the temperature difference between the inside and outside of the Boss zone is about 17.5℃, and the pure mechanical strain growth rate is about 12.2με / min, both of which exceed the preset threshold, thus triggering the first rate reduction; when the temperature rises to 115℃, a uniform temperature platform is inserted to homogenize the internal temperature field; the average degree of curing in the region at the end of the original 5-hour heat preservation was about 0.828, which is lower than the target value of 0.95, so the heat preservation time is extended; when the temperature drops through the glass transition sensitive range of 110~90℃, the cooling rate is reduced to avoid freezing of internal stress.

[0227] 3. Stress control effect

[0228] like Figure 8 and Figure 9 As shown, staged regulation has a significant inhibitory effect on the stress level in each region: Figure 8 The chart compares the maximum stress in each region at every control point throughout the entire process. The solid blue squares represent the baseline process results, while the dashed red circles represent the offline playback control results. It can be seen that the maximum stress in all four regions was significantly reduced after control, with the peak stress in the transition section decreasing from 337.7 MPa to 228.4 MPa, indicating a significant suppression of stress concentration effects.

[0229] Figure 9 To compare the maximum residual stress in the region after cooling, the peak residual stress at the root of the Boss under the baseline process reached 107.1 MPa, which was reduced to 78.9 MPa after adjustment. The overall residual stress level decreased significantly, verifying the effect of staged control on suppressing internal stress accumulation and freezing.

[0230] 4. Review the conclusion

[0231] The offline kinetic playback results show that the average final degree of curing in the four types of regions was 0.877 at the end of the baseline process, while the average final degree of curing in this controlled process was increased to 0.960. At the same time, the peak stress and final residual stress in the key regions throughout the process were significantly reduced, which verifies the effectiveness of the staged control strategy in improving curing uniformity and suppressing internal stress accumulation. It can provide a reliable parameter benchmark and logical support for the control of solid bottle curing.

[0232] (ii) Simultaneous implementation of sensor array pre-embedding and winding

[0233] This step involves manual operation in conjunction with CNC fiber winding equipment, and the sensor array embedding is completed synchronously with the winding process. The specific process is as follows: 1. Sensor Unit Prefabrication: Standardized sensor units are prefabricated, and bare fiber optic gratings are cut to specified lengths. The original coating layer of the strain-sensing grating is retained to ensure strain transfer performance. The temperature-compensating grating is encapsulated using a high-purity quartz inert capillary tube with an air gap for stress-free encapsulation: the grating is inserted into a quartz capillary tube with an inner diameter larger than the outer diameter of the fiber. The grating area is located in the cavity in the middle section of the capillary tube, and both ends are sealed with high-temperature resistant ceramic adhesive. An air gap is maintained between the grating area and the tube wall, without resin filling. After encapsulation, the grating is only affected by ambient temperature and is not affected by curing shrinkage or structural deformation. Each grating has a 15cm pigtail at both ends, which is marked with a number. The center wavelength and light transmission performance are tested in advance.

[0234] 2. Layered Installation: Start the CNC fiber winding machine and perform fiber winding of the gas cylinder liner according to the established process; stop the machine when the third spiral winding layer is completed, and lay the sensing units according to the regional risk weight layout plan: 3 sets of sensing units are evenly spaced along the axial direction in the cylinder body section, 4 sets of sensing units are evenly arranged along the circumference in the transition section, 3 sets of sensing units are evenly arranged along the circumference at the root of the boss, and 2 sets of sensing units are arranged along the meridian in the end cap area. Each sensing unit includes 1 axial strain grating, 1 circumferential strain grating, and 1 temperature compensation grating. The grating area is aligned with the center of the measuring point, and a small amount of premixed resin is used for pre-fixation to ensure that no displacement occurs during the winding process.

[0235] 3. Fiber optic cable lead-out and protection: After all sensing units are deployed, the fiber optic cables of each grating are gathered along the winding layer to the end of the gas cylinder. The fiber optic cables are laid straight along the axial direction and lightly pressed and fixed with fiber bundles to avoid bending and damage during the winding process. The fiber optic cables are finally gathered at the flange at the end of the gas cylinder, with a 30cm length reserved for subsequent welding. The end is sealed with a protective sleeve to prevent the fiber optic cables from being crushed during the winding process.

[0236] 4. Winding and forming: Continue to complete the winding of the remaining winding layers until the gas cylinder is fully wound and formed; the status of the tail fiber is manually inspected throughout the process to ensure that the sensor array is embedded in the interface between the spiral winding layer and the circumferential winding layer at the same time as the winding process, so as to simultaneously acquire multi-directional deformation data in the axial, circumferential and interlayer normal directions.

[0237] (III) Sensor system connection and initial parameter calibration

[0238] This step combines manual operation and system debugging to complete the establishment of the all-fiber signal transmission link and initial parameter calibration. The specific process is as follows: 1. Optical Path Connectivity and Rotation Verification: The gas cylinder is hoisted to the rotating position of the curing oven. The coating layer of the pre-reserved pigtails at the ends is removed. Using a fiber optic fusion splicer, all the pigtails of the sensing gratings are point-to-point fused to the rotor end fiber of the fiber optic rotary connector. After splicing, the fiber is protected with heat-shrink tubing and fixed to the rotor flange, ensuring that the rotor end rotates synchronously with the gas cylinder without any pulling. Then, the stator end of the fiber optic rotary connector is led out of the curing oven through armored fiber and connected to the corresponding test channel of the fiber optic signal demodulation unit. The gas cylinder is manually rotated while observing the real-time wavelength signal of the demodulation unit to confirm that the optical signal of all channels is continuous and uninterrupted, verifying the optical path stability under rotation conditions.

[0239] 2. Initial Parameter Calibration: Stabilize the curing oven temperature at 25℃ and keep the gas cylinder unloaded for 30 minutes; trigger initial wavelength acquisition, record the initial center wavelength of all gratings and store it in the system database. Conduct stepped isothermal calibration: Sequentially set the oven temperature to 30℃, 50℃, 70℃, 90℃, and 110℃, hold at each temperature for 30 minutes, and then collect wavelength data. Fit and calculate the inherent temperature sensitivity coefficient of each grating. Simultaneously, perform a tensile test on a standard laminate to calibrate the interlaminar strain transfer coefficient at the corresponding winding angle, and calculate the actual strain sensitivity coefficient after embedding; enter all calibration parameters into the online analysis and decision unit as the calculation benchmark for subsequent signal decoupling.

[0240] (iv) Configuration and joint testing of control system parameters

[0241] This step involves manual parameter input and system function debugging to complete model configuration and end-to-end verification. The specific process is as follows: 1. Material constitutive parameter input: Input material constitutive parameters in the online analysis and decision unit, including the dual-mechanism pre-exponential factor, reaction activation energy, reaction order, as well as the winding angle thermal conductivity correction function, interlayer thermal resistance correction coefficient, etc., to complete the parameter configuration of the material curing kinetic constitutive model after dual correction of interlayer thermal resistance and winding angle; at the same time, input the chemical shrinkage coefficient of the resin after complete curing and the temperature-curing degree coupled relaxation modulus series parameters to ensure that the online analytical formula of the curing state can be calculated normally.

[0242] 2. Regulation strategy parameter configuration: Manually set the staged regulation thresholds and weight parameters: set the internal and external temperature difference threshold of 15℃ and the internal stress growth rate threshold of 10με / min for the heating stage; set the target curing degree of 98% and the stress safety threshold of 55MPa for the heat preservation stage; set the glass transition temperature range of 120~135℃ for the cooling stage; and configure the priority weight of the regulation targets for each stage to complete the parameter configuration of the staged multi-objective regulation strategy.

[0243] 3. System full-link integration and debugging: Import the preset initial curing process curve (room temperature → 2℃ / min heating to 120℃ → holding for 60min → 1.5℃ / min heating to 180℃ → holding for 240min → 1℃ / min cooling to 60℃ and exiting the oven), trigger the system simulation operation, verify that the signal acquisition, decoupling calculation, status parsing, and command generation functions are normal, confirm that the curing oven execution control unit can normally receive and execute temperature adjustment commands, and complete the system full-link integration and debugging.

[0244] (v) Implementation of online monitoring and closed-loop control throughout the entire process of solidification

[0245] This step primarily relies on automated system operation, supplemented by manual inspection, to implement closed-loop control of the entire curing process. The sampling frequency is set to 2Hz, and the cylinder rotation speed is 2r / min. The specific process is as follows: 1. Heating Stage Control: After the curing process enters the first heating stage, the system collects the grating wavelength signal in real time. After sliding filtering and outlier removal, the internal temperature and pure mechanical strain of each measuring point are calculated using an improved multiphysics decoupling model. Simultaneously, the degree of curing is calculated using a material curing kinetic constitutive model, and the curing stage is jointly determined. When the furnace temperature reaches 112℃, the system monitors that the internal and external temperature difference at the transition section measuring point reaches 17℃ and the internal stress growth rate reaches 13.5με / min, both exceeding the preset threshold. The adjusted heating rate of 0.8℃ / min is then calculated using the heating rate increment adjustment formula. The system automatically sends a command to the curing furnace execution control unit to reduce the heating rate and inserts a 115℃ intermediate heat preservation platform. During the heat preservation process, the equipment status is manually inspected every 30 minutes to confirm that the optical path signal and temperature control system are operating normally. After 90 minutes of heat preservation, the internal temperature field becomes homogenized, the internal stress growth rate drops back to 8με / min, and the system automatically resumes the preset heating process.

[0246] 2. Temperature Holding Stage Control: When the temperature rises to 180℃, the system enters the high-temperature holding stage. The system uses both curing uniformity and stress relaxation as dual control objectives, calculating the curing degree and cumulative internal stress at each measuring point in real time. After 220 minutes of holding, the overall average curing degree reaches 93%, and the maximum internal stress in the critical area exceeds the safety threshold. The system calculates the holding time extension by 90 minutes using a dual-parameter driven formula for curing degree and stress, automatically extending the 180℃ holding time. During the extended holding process, the resin undergoes viscoelastic stress relaxation in a highly elastic state, and the internal stress continues to decrease. When the overall curing degree reaches 98.5% and the stress in the critical area falls back to a safe range, the system determines the end of the holding stage and automatically enters the cooling stage.

[0247] 3. Cooling Phase Control: Upon entering the cooling phase, the system prioritizes suppressing the superposition of internal stress during molding and freezing. The cooling rate is dynamically adjusted using a phased-adjustment cooling rate formula. When the temperature drops to 132℃ (entering the glass transition temperature range), the system automatically reduces the cooling rate from 1℃ / min to 0.4℃ / min and inserts a 128℃ slow cooling platform for 35 minutes. Once the structure has uniformly passed through the glass transition range and the internal stress growth rate has returned to a safe range, the system resumes the normal cooling rate until the temperature drops to 60℃, at which point the curing process ends.

[0248] 4. Closed-loop iterative fine-tuning: During the control process, the system completes an evaluation of the control effect every 30 seconds, and adopts a deviation closed-loop iterative mechanism for incremental parameter fine-tuning. The adjustment range of the single heating and cooling rate does not exceed 0.2℃ / min, so as to avoid molding defects caused by drastic fluctuations in process parameters. The operator is on duty throughout the process and records the temperature, strain and internal stress data of key measuring points every hour to confirm the stable operation of the system, forming a full closed-loop control of "perception-decoupling-analysis-decision-execution-feedback".

[0249] (vi) Post-curing treatment and process iteration optimization

[0250] After curing, the rotating mechanism and temperature control system are shut down, the gas cylinder is lifted out of the curing oven, the fiber optic rotating connector is removed, and the exposed pigtail is cut off, completing the subsequent finishing process for the gas cylinder. Residual stress testing is performed on key areas such as the transition section and the root of the Boss (the gas cylinder's base) to obtain finished product performance data.

[0251] The system synchronously imports the monitoring data, process parameter curves, and finished product performance test results into the process data iteration unit. The system uses a built-in recursive parameter identification algorithm to iteratively correct the parameters of the material curing kinetic constitutive model and the control threshold. After the correction results are manually confirmed, the system parameter library is updated for the control optimization of the next batch of gas cylinders.

[0252] (vii) Comparison of Implementation Results

[0253] The comparison between this embodiment and the traditional ring-opening curing process is shown in the accompanying drawings of the instruction manual: Comparison of process curves ( Figure 10 Traditional open-loop processes use a three-segment curve with a fixed slope; the real-time control process curve of this invention can automatically insert the intermediate heat preservation platform and the slow cooling platform of the glass transition zone according to the internal molding state, so as to realize the dynamic adaptive adjustment of the process curve.

[0254] Stress distribution comparison ( Figure 11Before real-time control, the maximum equivalent stress in key areas such as the head and the root of the boss reached 338 MPa, with a significant stress concentration effect. After closed-loop control using this solution, the maximum equivalent stress in key areas was reduced to about 190 MPa, the maximum residual stress was reduced by about 28%, the incidence of interlayer interface defects decreased by more than 40%, and the consistency of molding quality between batches was significantly improved.

[0255] VII. Industrialized Mass Production Applications

[0256] The above-mentioned deployment and control methods can be directly applied to the industrial-scale mass curing production line of composite material hydrogen storage cylinders. By deeply integrating with existing curing oven temperature control systems and production management systems, it can achieve full-process coverage from single-cylinder trial production to stable mass production. The specific application methods are as follows: (a) Production line deployment and parameter preset The system hardware is adapted to the existing curing oven station configuration: the fiber optic rotary connector is installed at the end of the curing oven's rotating spindle, compatible with the original rotor tooling; the fiber optic signal demodulation unit is connected to the production line's industrial control network, and the online analysis and decision-making unit establishes bidirectional communication with the original PLC control unit of the curing oven, retaining the original manual / automatic control mode without changing the original operating logic of the production line.

[0257] For the gas cylinder models that are designed for production, the pre-calibrated material constitutive parameters, layout response function parameters, and phased control thresholds and weight parameters are imported in batches into the online analysis and decision-making unit to form a standard process parameter package for the corresponding model. Before production, operators only need to select the product model on the system interface, and the system will automatically load the corresponding parameters. There is no need to debug each cylinder individually, and the pre-production preparation can be completed quickly.

[0258] (II) First Article Process Validation and Baseline Consolidation

[0259] Before mass production, a first-piece trial is performed: the first-piece curing is run using the standard parameter package of the corresponding model, and the temperature, degree of curing, internal stress evolution data and process adjustment trajectory of each area are recorded throughout the process; after curing, the parameter package is calibrated and corrected for the first time by combining the results of residual stress detection, water pressure burst test and interlayer defect detection of the finished product, so as to eliminate the systematic deviation caused by material batch fluctuations and individual equipment differences.

[0260] After the first piece passes verification, the calibrated parameter package is solidified as the production baseline process for that model, serving as the default benchmark for subsequent mass production. This ensures that the initial process benchmark is consistent across all batches, reducing quality fluctuations between batches from the source.

[0261] (III) Routine closed-loop operation of mass production

[0262] In routine production, the system automatically calls upon the baseline process parameters for the corresponding model. Each gas cylinder undergoes a fully closed-loop control process throughout the curing process, encompassing "signal acquisition - status analysis - control decision - command execution - feedback iteration," requiring no real-time manual monitoring. Only in extreme conditions such as optical path anomalies, stress exceeding safety thresholds in critical areas, or temperature deviations exceeding ranges are audible and visual alarms triggered and pushed to the operator's terminal for manual intervention. During production, the system automatically associates each gas cylinder with a unique product identifier, synchronously stores full-process monitoring data, process adjustment records, and the final molding status, automatically generates a single-cylinder molding quality file, and supports integration with the factory's MES system, enabling quality traceability and conformity assessment throughout the product's entire lifecycle.

[0263] (iv) Dynamic iterative optimization between batches

[0264] After each production batch is completed, the process data iteration unit automatically extracts the monitoring data and finished product inspection data of all gas cylinders in that batch. Through the built-in recursive parameter identification algorithm, it incrementally corrects the parameters of the material curing kinetic constitutive model, the control gain coefficient, and the safety threshold, generates an optimized parameter package, and archives it.

[0265] When the next batch of production starts, the system automatically loads the latest optimized parameter package to achieve continuous iteration of "producing one batch and optimizing another". As production batches accumulate, the model calculation accuracy and control adaptability continue to improve, and the fluctuation of molding quality between batches gradually narrows, eliminating the need for repeated manual adjustment of process parameters.

[0266] (v) Rapid adaptation of multiple product models

[0267] For newly developed gas cylinder models (different volumes, different pressure levels, and different layup structures), initial parameter schemes can be quickly generated based on existing parameter packages of similar specifications by scaling up structural dimensions, correcting layup angles, and recalculating risk weights. Parameter finalization can be completed in just 1-2 rounds of trial production, significantly shortening the process development cycle of new products and reducing trial and error costs and R&D cycle.

[0268] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for deploying an optical fiber sensor array for solidification in a hydrogen storage bottle, characterized in that, The deployment steps include the following: The hydrogen storage cylinders to be deployed are divided into four regions: the cylinder body, the transition section, the end cap region, and the root of the boss. The structural stress concentration factor K for each region is obtained. σ,i With curing and molding risk coefficient R i ; i is the region number; K in the corresponding region σ,i With R i Multiply by each other to obtain the regional risk weight w. i Normalization process yields the corresponding normalized value. ; Through the corresponding Calculate the deployment response value for each area. p, q, A, λ b β b exp[·] represents the pre-calibrated layout function parameters; exp[·] represents the natural exponent. According to F in each region i Combined with the total number of sensor units N that can be deployed tot The effective length L of each area along the corresponding deployment direction i The minimum number of sensing units N in a single area min The number of sensor units deployed in each region was calculated. Spacing between adjacent sensor units `max{·}` is the maximum value operation, and `round[·]` is the rounding function; F j This represents the deployment response value for the j-th region. Based on N i With d i Sensing units are pre-embedded at the interface between adjacent winding layers in the corresponding area along the current area's layout direction until the pre-embedded layout of the entire bottle's sensing array is completed.

2. The fiber optic sensor array deployment method for hydrogen storage bottle solidification according to claim 1, characterized in that, Structural stress concentration factor K σ,i The acquisition process is as follows: By establishing a three-dimensional finite element model of the hydrogen storage cylinder to be deployed and importing the preset curing process curve, a thermo-mechanical coupling simulation of the entire curing process was carried out, and the ratio of the maximum Mises equivalent stress in the corresponding region to the average Mises equivalent stress in the reference region of the cylinder was extracted. Curing and molding risk factor R i The acquisition process is as follows: Historical production samples with the same resin matrix, fiber reinforcement material, winding lay-up structure and curing process as the hydrogen storage cylinders to be deployed are selected. Molding defects in each region are counted, and the regional defect incidence rate of the i-th region is calculated. The regional defect incidence rate is the proportion of the number of samples with molding defects in that region to the total number of valid statistical samples. Using the regional defect incidence rate of the cylinder section as the benchmark value, the ratio of the regional defect incidence rate of each region to the benchmark value is determined as the curing and molding risk coefficient of the corresponding region.

3. The fiber optic sensor array deployment method for hydrogen storage bottle solidification according to claim 1, characterized in that, Configure the function parameters p, q, A, λ b β b Satisfying 0 < q ≤ p < 1, 0 < A < 1, 0 < λ b <1, 0<β b <1, and obtained through the following calibration steps: Establish a thermo-mechanical coupling simulation benchmark model of the curing process with the same structural type, material system and layup process as the hydrogen storage cylinder to be deployed. Calculate the true Mises equivalent stress distribution of each region of the entire cylinder during the curing process, and use it as the benchmark true value for evaluating the deployment effect. The total number of sensor units N that can be deployed per fixed bottle tot Multiple different regional measurement point allocation schemes are set up, with each scheme corresponding to a different allocation ratio of the number of sensing units in four regions: the cylinder section, the transition section, the end cap region, and the root of the boss. For each group of regional measurement point allocation schemes, the stress value of each regional measurement point is used as the sampling point for interpolation fitting to obtain the monitoring and reconstruction stress distribution; the average relative error between the reconstruction stress of the head region and the root of the Boss and the reference true value is calculated as the evaluation index of the current regional measurement point allocation scheme. With the objective of minimizing the deviation between the area measurement point allocation result output by the deployment response function and the area measurement point allocation scheme with the optimal evaluation index, a nonlinear least squares fitting algorithm is used to solve for the deployment function parameters p, q, A, and λ. b β b The initial optimal value; Multiple independent samples with different total number of sensing units are selected to verify the generalization of the initial optimal value. If the average relative error between the head region and the root of the Boss of all samples does not exceed the preset error threshold, the initial optimal value is determined to be the final calibration value. If the error exceeds the threshold, supplement the sample of the area measurement point allocation scheme and refit until the accuracy requirements are met.

4. The fiber optic sensor array deployment method for hydrogen storage bottle solidification according to claim 1, characterized in that, The specific deployment directions for each area are as follows: The sensing units in the cylinder body section are arranged along the axial direction of the hydrogen storage cylinder; the sensing units in the head area are arranged along the meridian direction of the head; and the sensing units in the transition section and the root of the Boss are arranged along the circumference of the hydrogen storage cylinder. The effective length of the cylinder section is the axial length of the cylindrical section of the cylinder; the effective length of the head area is the arc length of the head meridian; the effective length of the transition section is the circumference length at the midpoint of the axial direction of the transition section; and the effective length of the Boss root is the circumference length at the interface between the Boss metal part and the composite material layer. When pre-embedding the sensing unit, a continuous all-fiber signal transmission link from the measuring point on the bottle to the external demodulation equipment is simultaneously constructed: The grating area and the pigtail of each sensing unit are a continuous structure of the same optical fiber. The reflected light signal carrying temperature and strain information sensed by the grating area is transmitted by total internal reflection through the pigtail core. The pigtails of each sensing unit are laid straight along the winding layer and converge at the end of the hydrogen storage tank. The pigtails are lightly pressed and fixed by fiber bundles to suppress optical signal loss caused by bending. The pigtails converged at the end of the hydrogen storage tank are fused and fixed to the rotor end of the optical fiber rotary connector. The stator end of the optical fiber rotary connector is connected to the external optical fiber signal demodulation unit through a fixed optical fiber. The optical coupling structure inside the optical fiber rotary connector enables continuous and uninterrupted transmission of optical signals under the condition of hydrogen storage tank rotation. Finally, the optical fiber signal demodulation unit converts the optical wavelength signal into a calculable digital signal for output.

5. A method for controlling the solidification process of a hydrogen storage bottle, characterized in that, The fiber optic sensor array deployment method for hydrogen storage bottle solidification according to any one of claims 1-4 is used to pre-embed a fiber optic sensor array on the hydrogen storage bottle, wherein each sensing unit includes at least one strain sensing grating and one temperature compensation grating, and includes the following control steps: S1. After completing the pre-embedding of the fiber optic sensing array, under a constant temperature and no-load standard environment, the interlayer strain transfer correction coefficient is introduced to correct the grating sensitivity coefficient, and an improved multi-physics decoupling model is constructed. S2. After the hydrogen storage cylinder is put into the furnace and the curing program is started, the original wavelength signals of the temperature compensation grating and the strain sensing grating are collected at each sampling time according to the preset sampling frequency. The real-time absolute temperature of each measuring point is calculated based on the wavelength shift of the temperature compensation grating, and the total center wavelength shift of the strain sensing grating is obtained. At the same time, the real-time resin curing degree is calculated based on the real-time absolute temperature. S3. Input the real-time absolute temperature into the material curing kinetic constitutive model, which is corrected by both interlayer thermal resistance and winding angle. The real-time resin curing degree calculated at the previous sampling time is used as the initial state. Update the real-time resin curing degree at the current sampling time. Calculate the free chemical shrinkage strain based on the real-time resin curing degree. Then, input the total offset of the center wavelength of the strain sensing grating, the real-time temperature change, and the free chemical shrinkage strain into the multiphysics decoupled model to obtain the pure mechanical curing strain. Calculate the molding internal stress at each measuring point based on the real-time absolute temperature, the real-time resin curing degree, and the pure mechanical curing strain. S4. Based on the current process program segment of the curing oven, the rate of change of the oven temperature setpoint, and the real-time resin curing degree, jointly determine whether the current stage is heating, holding, or cooling. Compare the average resin curing degree, average molding internal stress, and maximum molding internal stress of each region with the current stage's curing degree target trajectory and allowable internal stress limit to obtain the measured curing degree deviation and measured internal stress deviation of each region. Perform equivalent correction and dimensionless processing on the deviations based on the physical occupancy of the sensing unit, and combine the regional risk weights of each region to obtain the overall molding state deviation. Match the multi-objective control strategy corresponding to the current stage, and under the constraints of the preset upper and lower limits of process parameters and the maximum rate of change, generate incremental adjustment commands for process parameters such as heating rate, holding temperature, holding time, or cooling rate, and send them to the curing oven for execution. S5. After adjusting the process parameters, continuously collect the real-time signal after regulation through the fiber optic sensor array and return to steps S2-S4 for cyclic verification. Use the deviation closed-loop iterative mechanism to incrementally fine-tune the process parameters until the entire curing process is completed.

6. The method for controlling the solidification process of a hydrogen storage bottle according to claim 5, characterized in that, The improved multiphysics decoupling model and the material solidification kinetic constitutive model are solved in real time using a serial linkage iterative method. The solution process at each sampling time is as follows: First, perform temperature calculation and cure degree update: The total offset Δλ of the center wavelength of the temperature compensation grating, which is positioned at the same location as the strain sensing grating, is measured. B The real-time temperature change ΔT at the current measuring point is calculated. The real-time temperature change ΔT is added to the reference absolute temperature T0 of the constant temperature and no load standard environment to obtain the real-time absolute temperature T at the current measuring point, which is then input into the material curing kinetic constitutive model. In the material curing kinetic constitutive model, the real-time resin curing degree at the previous sampling time is used as the initial state. Based on the real-time absolute temperature T, the real-time curing reaction rate dα / dt at the current time is calculated. After time step integration, the real-time resin curing degree α at the current sampling time is obtained. The constitutive model of material solidification kinetics is expressed as follows: ； In the formula, A1 and A2 are the pre-exponential factors corresponding to the dual-reaction mechanism of the curing reaction, respectively; ΔE1 and ΔE2 are the activation energies corresponding to the dual-reaction mechanism of the curing reaction, respectively; R is the universal gas constant; m and n are the order of the curing reaction; f(θ) is the winding angle thermal conductivity correction function, where the independent variable θ is the fiber winding angle at the current measuring point, used to correct for the local reaction rate differences caused by the anisotropic thermal conductivity of the fiber; γ h is the interlayer thermal resistance correction factor, used to correct the overall reaction rate attenuation caused by interlayer interface thermal resistance in thick-walled wound structures; e is the natural constant. Then perform strain decoupling separation: Based on the updated real-time resin cure degree α, and combined with the chemical shrinkage coefficient of the material at full cure, the free chemical shrinkage strain component generated by the current resin crosslinking reaction is calculated. ; The total offset of the center wavelength Δλ B Temperature change ΔT, free chemical contraction strain component Substituting into the improved multiphysics decoupling model, and successively subtracting the temperature response component and the chemical shrinkage response component, we obtain the pure mechanical solidification strain caused solely by the constrained deformation of the structure. ; The improved multiphysics decoupling model is represented as follows: ； In the formula, Δλ B The total offset of the center wavelength of the strain sensing grating at the current measuring point is obtained in real time by the fiber optic signal demodulation unit; K T ΔT is the inherent temperature sensitivity coefficient of the strain sensing grating at the current measuring point; ΔT is the real-time temperature change at the current measuring point, which is calculated by the temperature compensation grating deployed at the same location. This is the actual grating sensitivity coefficient at the current measurement point after interlayer strain transfer correction; This represents the purely mechanical solidification strain at the current measuring point; This represents the free chemical shrinkage strain component generated by the resin crosslinking reaction at the current measuring point.

7. The method for controlling the solidification process of a hydrogen storage bottle according to claim 5, characterized in that, Before calculating the overall forming state deviation, the individual point calculation results of each measuring point are first aggregated at the regional level to obtain the state characteristic quantities of each region. The specific process is as follows: For real-time absolute temperature and real-time resin curing degree, the average value of the region is calculated by the effective area weighting method of the measuring points, and the average absolute temperature and average resin curing degree of the corresponding region are obtained, which characterizes the overall thermal state and curing reaction process of the region. For the forming internal stress, both the regional average and the regional maximum are calculated: the regional average internal stress is calculated by weighting the effective area of ​​the measuring points, which is used to characterize the overall stress level of the region; the regional maximum internal stress is the maximum value of the forming internal stress of all measuring points in the corresponding region. The formula for calculating the internal stress during forming is as follows: ； In the formula, σ n (t) represents the molding internal stress accumulated at the nth measuring point from the start of curing to time t; G(·) is the modified relaxation modulus considering the coupling between curing degree and temperature. For the nth measurement point at Real-time resin curing degree at any given moment; For the nth measurement point at Real-time absolute temperature at any given moment; For the nth measurement point at Total strain at any given moment; For time variables The infinitesimal element; for The first derivative; The average resin curing degree and average internal stress of each region are subtracted from the target value of the current curing stage to obtain the measured curing degree deviation and measured internal stress deviation of each region, which are used as inputs for subsequent physical occupancy equivalent correction of the sensing unit.

8. The method for controlling the solidification process of a hydrogen storage bottle according to claim 7, characterized in that, The process for obtaining the overall molding state deviation is as follows: Calculate the volume fraction of the sensing unit in each region. The occupied volume fraction is the ratio of the equivalent volume of a single sensing unit to the effective volume of the corresponding composite material in the region. Combined with the fraction of occupied volume The sensor unit layout direction, real-time resin curing degree, gel point curing degree, real-time absolute temperature and glass transition temperature are used to construct the physical occupancy equivalent variables for each region: ； ； In the formula, Ω i (t) represents the physical occupancy equivalent variable of the i-th region at time t; Let be the volume fraction of the sensing unit within the i-th region; η i Assign directional modulation coefficients to the sensing units in the i-th region to quantize the enhancement of the occupancy disturbance by the fiber angle; θ i λ is the angle between the arrangement direction of the sensing units in the i-th region and the direction of the main supporting fiber in that region; o,i The occupancy saturation coefficient for the i-th region; γ i α is the correction coefficient for the curing state of the i-th region; i (t) represents the real-time resin curing degree of the i-th region at time t; α g,i The degree of curing at the gel point of the i-th region at time t is an inherent material property of the resin matrix and does not change with curing time. ρ i This is the correction factor for the glass transition region of the i-th region; ζ i (t) represents the normalized variable for the glass transition of the i-th region at time t; T i (t) represents the real-time absolute temperature of the i-th region at time t; T gi (t) represents the real-time glass transition temperature of the i-th region at time t. Its value dynamically increases with the current degree of resin curing and is a material state parameter related to the degree of curing; ΔT g is the half-width of the region affected by the glass transition; clip[·] is the cutoff function; Based on the physical occupancy equivalent variable and the pre-calibrated occupancy deviation sensitivity coefficient, the measured curing degree deviation and measured internal stress deviation of each region are converted into the true state deviation without sensor unit interference: ； ； In the formula, Δα represents the actual degree of curing deviation of the i-th region at time t. i (t) represents the measured curing degree deviation of the i-th region at time t; c α,i The occupancy deviation sensitivity coefficient corresponds to the curing degree deviation of the i-th region. Δσ represents the actual internal stress deviation of the i-th region at time t. i (t) represents the measured internal stress deviation of the i-th region at time t; c σ,i This is the occupancy deviation sensitivity coefficient corresponding to the stress deviation in the i-th region; based on and Calculate the relative deviation corresponding to the state deviation: ； ； In the formula, e α,i (t) represents the relative curing degree deviation of the i-th region at time t; δ α,i e is the allowable deviation threshold for the degree of cure of region i in the current curing stage; σ,i (t) represents the relative internal stress deviation of the i-th region at time t; δ σ,i Let be the allowable deviation threshold of internal stress in region i during the current curing stage; [·] + = max(x,0) is the positive part truncation function, max is the maximum value operation, and x is the input value; The actual curing degree deviation and actual internal stress deviation of each region are weighted and summed using regional risk weights to obtain the overall molding state deviation, which includes both curing degree deviation and internal stress deviation. ； In the formula, D(t) represents the overall forming state deviation of the hydrogen storage cylinder at time t; λ α λ is the priority weighting coefficient for curing degree deviation. σ This is the priority weighting coefficient for internal stress deviation.

9. The method for controlling the solidification process of a hydrogen storage bottle according to claim 8, characterized in that, Match the multi-objective control strategy corresponding to the current solidification stage, with differentiated objective priorities, deviation weight ratios, and core control targets for each stage, and execute them in the following manner: During the heating stage, the core control objectives are temperature field uniformity and preventing excessive accumulation of internal stress during molding. The priority weighting coefficient λ for curing degree deviation is... α The priority weighting coefficient λ for internal stress deviation is 0.4~0.

6. σ The value is 0.4~0.6, and the core control object is the heating rate. When the deviation of the overall molding state exceeds the set threshold, the heating rate is reduced. When the deviation is too large, the intermediate heat preservation platform at the corresponding temperature is automatically inserted. After the temperature field is homogenized and the internal stress growth rate drops back to the safe range, the preset heating process is restored. During the heat preservation stage, the dual core control objectives are uniformity of curing degree and relaxation of internal stress during molding. The priority weighting coefficient λ for curing degree deviation is... α The priority weighting coefficient λ for internal stress deviation is 0.2 to 0.

4. σ The value ranges from 0.6 to 0.8, and the core control parameters are the insulation duration and insulation temperature. With internal stress reduction as the core, an instruction for adjusting the insulation time increment is generated. The high-elasticity viscoelasticity of the resin is used to promote the release of internal stress. At the same time, the curing degree is taken into account as the bottom line condition, while also taking into account the fullness of the curing reaction. The cooling stage aims to suppress the superposition of internal stress during molding and freezing, with the priority weighting coefficient λ for curing degree deviation. α The priority weighting coefficient λ for internal stress deviation is 0.1 to 0.

3. σ The value is set to 0.7~0.9, and the core control object is the cooling rate. When the temperature drops to the glass transition temperature range, a slow cooling platform is inserted in combination with the current internal stress level to prevent the internal stress from being frozen into irreversible residual stress during the glass transition process.

10. A hydrogen storage cylinder solidification process control system, characterized in that, The method for controlling the curing process of hydrogen storage bottles according to any one of claims 5-9 includes a pre-embedded fiber optic sensor array, a fiber optic rotary connector, a fiber optic signal demodulation unit, an online analysis and decision-making unit, a curing oven execution control unit, and a process data iteration unit. The pre-embedded fiber optic sensing array adopts the fiber optic sensing array deployment method for hydrogen storage bottle curing as described in any one of claims 1-4, and is pre-embedded in the winding layers of the cylinder section, transition section, end cap area and Boss root of the hydrogen storage bottle; each sensing unit includes at least one strain sensing grating and one temperature compensation grating, used to collect temperature and strain light signals during the curing process; the pigtails of each sensing unit are laid straight along the winding layer and converge to the end of the hydrogen storage bottle for lead-out. The rotor end of the fiber optic rotary connector is fused and fixed to the lead-out pigtail of the pre-embedded fiber optic sensing array, and rotates synchronously with the hydrogen storage tank; the stator end of the fiber optic rotary connector is connected to the fiber optic signal demodulation unit through armored fixed fiber optic cable, which is used to maintain continuous optical path alignment under the condition of hydrogen storage tank rotation and solidification, so as to realize uninterrupted transmission of optical signal. The fiber optic signal demodulation unit is used to perform wavelength demodulation and signal preprocessing on the received optical signal, and outputs the raw wavelength data of each measurement point to the online analysis and decision unit. The online analysis and decision-making unit incorporates an improved multi-physics decoupling model, a material curing dynamics constitutive model corrected by interlayer thermal resistance and winding angle, a regional feature aggregation module, a sensor unit physical occupancy equivalent correction module, and a phased multi-objective control strategy library. It is used to calculate the real-time absolute temperature, pure mechanical curing strain, real-time resin curing degree, and cumulative internal stress of each measuring point through a serial linkage iterative method. After regional feature aggregation and physical occupancy equivalent correction, it calculates the overall molding state deviation, matches the control strategy corresponding to the current curing stage, and generates incremental adjustment instructions for curing process parameters. The curing oven execution control unit is connected to the command output terminal of the online analysis and decision unit to receive incremental adjustment commands and adjust the heating rate, holding temperature, holding time and cooling rate of the curing oven in real time. The process data iteration unit and the online analysis and decision-making unit interact bidirectionally to store monitoring data, process parameters and finished product performance data throughout the solidification process. The constitutive model parameters and control thresholds are iteratively corrected through a recursive parameter identification algorithm to continuously optimize the control accuracy of subsequent batches.

Images