Liquid carbon dioxide foamed polyurethane foam stepwise decompression control device and method

Abstract

This invention discloses a step-by-step decompression control device for liquid carbon dioxide foamed polyurethane foam, comprising: an emulsifier with a filter cavity extending along its axis inside; a multi-layer filter unit disposed within the filter cavity, including at least five layers of filter screens with different pore sizes; the inner diameter of the emulsifier being larger than the inner diameter of the inlet, forming a cross-sectional abrupt change zone; an inlet installed at one end of the emulsifier and communicating with the filter cavity, equipped with a one-way valve and a flow control valve connected in series; an outlet installed at the other end of the emulsifier and communicating with the filter cavity, equipped with a gas pressure gauge; and a controller communicatively connected to the one-way valve, the flow control valve, and the gas pressure gauge. This invention achieves gradient decompression through the synergistic effect of the multi-layer filter unit and the cross-sectional abrupt change zone, combined with an adaptive PID control algorithm to dynamically adjust the valve opening, effectively solving the problems of uneven bubble distribution and bubble merging and rupture caused by pressure abrupt changes, thus improving product consistency and production efficiency.

Description

Technical Field

[0001] This application relates to the technical field of polyurethane foam production equipment, and more specifically, to a device and method for controlling the step-by-step decompression of liquid carbon dioxide foamed polyurethane foam. Background Technology

[0002] Carbon dioxide foaming technology, as an environmentally friendly foaming process, utilizes supercritical carbon dioxide to replace traditional physical foaming agents, demonstrating significant advantages in polyurethane foam production. In existing technologies, the production of polyurethane block foam typically employs a chemical reaction system of polyols and isocyanates, combined with the physical foaming agent dichloromethane and the chemical foaming agent water. While this process is mature, it faces several technical bottlenecks in actual production.

[0003] First, when the mixture is sheared through a traditional filter, flow rates exceeding a critical value can lead to uneven foam distribution, directly affecting the uniformity of the final product's cells. Second, sudden pressure changes during decompression can easily cause bubble coalescence or rupture, making it difficult to form an ideal microporous structure. Furthermore, existing control systems lack the ability to dynamically adapt to fluid viscosity and gas content, and cannot automatically optimize the filtration gradient based on raw material characteristics. More significantly, conventional PID control exhibits lag in response to nonlinear pressure changes, particularly during formulation changes or fluctuations in production parameters, making precise, step-by-step decompression control difficult. These problems severely restrict the production efficiency and product consistency of high-quality polyurethane foam.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] The purpose of this invention is to provide a step-by-step pressure reduction control device and method for liquid carbon dioxide foamed polyurethane foam, which has the advantages of improving cell uniformity, preventing bubble merging and rupture, and realizing dynamic pressure regulation.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution:

[0007] A liquid carbon dioxide foamed polyurethane foam stepwise decompression control device includes:

[0008] The emulsifier has a filter cavity extending along its axis inside.

[0009] A multi-layer filtration unit is disposed within a filtration cavity and includes at least five layers of filter screens with different pore sizes, with each layer of filter screens arranged axially along the filtration cavity; the inner diameter of the emulsifier is larger than the inner diameter of the feed inlet, forming a cross-sectional abrupt change zone to accelerate pressure release.

[0010] The feed inlet is installed at one end of the emulsifier and communicates with the filter cavity. It is equipped with a one-way straight valve and a flow control valve connected in series.

[0011] The discharge port is installed at the other end of the emulsifier and communicates with the filter cavity, and is equipped with a gas pressure gauge;

[0012] The controller communicates with the one-way straight-through valve, flow control valve, and gas pressure gauge.

[0013] Furthermore, the pore size of the multi-layer filter unit satisfies:

[0014]

[0015] in, d0 is the pore size of the k-th filter layer, k is the reference pore size, k is the sequence number increasing along the fluid direction, and α is the pore size expansion coefficient. When α < 1, the pore size growth slows down, which is suitable for high viscosity formulations; when α > 1, the pore size growth accelerates, which is suitable for high gas content formulations.

[0016] Furthermore, the controller has an adaptive PID control algorithm:

[0017]

[0018] Where, Δ u ( t ) represents the valve opening adjustment amount; e(t) = ΔP(t), which represents the real-time pressure deviation; Kp′ is the dynamic proportional gain parameter, which controls the response intensity to the current pressure deviation; Ki′ is the dynamic integral gain parameter, which is used to eliminate historical accumulated errors and improve steady-state accuracy; Kd′ is the dynamic differential gain parameter, which is used to suppress pressure fluctuation advance control; Δt is the differential time window, which is used for the sampling interval of pressure deviation differential calculation.

[0019] Furthermore, the parameter update rules for Kp′, Ki′, and Kd′ are as follows:

[0020]

[0021] Where Kp0, Ki0, and Kd0 are the initial values ​​of the PID parameters; , and The learning rate coefficient is obtained through training with historical data from the database. is the curvature sensitivity coefficient, which is adaptively selected according to the viscosity of the raw material; tanh() is the hyperbolic tangent function, used to adjust the range of Kp′, Ki′ and Kd′ within (-1,1);

[0022] Dynamic proportional gain parameter Kp′: When a drastic pressure fluctuation is detected. When the value is large, the proportional term is automatically enhanced to suppress oscillations; during the steady phase, Kp′ is reduced to avoid overshoot;

[0023] Dynamic integral gain parameter Ki′: enhances the integral effect for continuous unidirectional deviation, and automatically weakens the integral to prevent saturation when the deviation changes rapidly;

[0024] Dynamic differential gain parameter Kd′: In the initial stage of pressure change, Larger and At the same time, the differential term is enhanced to achieve proactive adjustment.

[0025] A method for controlling the stepwise decompression of liquid carbon dioxide-blown polyurethane foam includes the following steps:

[0026] Step S1: Obtain the standard pressure release value;

[0027]

[0028] Among them, P in P is the emulsifier inlet pressure. out (t) represents the outlet pressure at time t, V0 represents the volume of the calibration mixture, and T0 represents the calibration duration;

[0029] Step S2: Calculate the pressure deviation in real time;

[0030]

[0031] Where ΔP(t) is the pressure deviation value, Vt is the real-time processing volume, and τ is the width of the sliding time window.

[0032] Step S3: Generate a pre-tuned signal based on the LSTM prediction model;

[0033]

[0034] Where δ represents the prediction time domain, λ represents the historical time-series data vector, and λ is the physical constraint weighting coefficient. For constraint functions;

[0035] Step S4: When At that time, the opening of the drive flow control valve is adjusted.

[0036]

[0037] in, To normalize the pressure deviation, The rate of change of pressure, It is a logarithmic flow.

[0038] Furthermore, this application includes batch optimization steps:

[0039] Step S5: After the production batch is completed, perform multi-objective optimization;

[0040]

[0041] Where X is the optimization variable vector, , The average value of the absolute value of the pressure deviation. The maximum absolute value of the pressure deviation is given by Evalve, where Evalve is the valve regulation energy consumption, and η is the penalty factor coefficient. This is the density constraint penalty function.

[0042] In summary, the present invention has the following beneficial effects:

[0043] In existing technologies, single-layer filtration structures are commonly used for material handling during polyurethane foam production. As the flow rate of the mixture increases, the fluid shear force on the filter screen rises sharply, leading to uneven pressure distribution. Existing equipment lacks an effective pressure-grading release mechanism, and turbulence easily occurs as the material passes through the filter screen, causing bubble coalescence or rupture. Especially in liquid carbon dioxide foaming systems, the rapid phase change characteristics of supercritical fluids exacerbate pressure fluctuations, expanding the range of foam particle size distribution and ultimately affecting the uniformity of the internal cell structure of the foam.

[0044] This invention achieves gradient pressure reduction through the synergistic effect of multi-layer filter units and cross-sectional abrupt change zones. Combined with an adaptive PID control algorithm to dynamically adjust the valve opening, it effectively solves the problems of uneven bubble distribution and bubble merging and rupture caused by pressure abrupt changes, and has the advantages of improving product consistency and production efficiency. Attached Figure Description

[0045] Figure 1 This is a structural diagram of the liquid carbon dioxide foamed polyurethane foam stepwise decompression control device described in this invention.

[0046] Figure 2 This is a schematic diagram of the liquid carbon dioxide foamed polyurethane foam stepwise decompression control device described in this invention.

[0047] The following are the labels in the diagram: Emulsifier 10, Filter cavity 101, Section change zone 102, Multi-layer filter unit 20, Filter screen 201, Feed port 30, One-way straight valve 301, Flow control valve 302, Discharge port 40, Gas pressure gauge 401. Detailed Implementation

[0048] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below with reference to the figures and specific embodiments.

[0049] like Figure 1 and Figure 2As shown, the present invention proposes a step-by-step decompression control device for liquid carbon dioxide foamed polyurethane foam, comprising: an emulsifier 10, which has a filter cavity 101 extending along the axis inside; a multi-layer filter unit 20 disposed in the filter cavity 101, including at least five layers of filter screens 201 with different pore sizes, each layer of filter screens 201 being arranged axially along the filter cavity 101; the inner diameter of the emulsifier 10 is larger than the inner diameter of the inlet 30, forming a cross-sectional abrupt change zone 102; the inlet 30 is installed at one end of the emulsifier 10 and communicates with the filter cavity 101, and is provided with a one-way valve 301 and a flow control valve 302 connected in series; the outlet 40 is installed at the other end of the emulsifier 10 and communicates with the filter cavity 101, and is provided with a gas pressure gauge 401; and a controller, which is communicatively connected to the one-way valve 301, the flow control valve 302 and the gas pressure gauge 401.

[0050] The emulsifier 10 refers to a fluid processing container with specific geometric features, which can be made of stainless steel. Its internal filtration cavity 101 has an axially extending structure that prolongs the fluid path. The multi-layer filtration unit 20 is a composite structure composed of filter elements with different pore sizes, which can be made of sintered metal mesh or polymer mesh, with the pore size of each layer increasing or decreasing along the flow direction. The abrupt change zone 102 is the transition area formed by the difference in diameter between the inlet 30 and the emulsifier 10, which can be achieved through a stepped diameter expansion structure, utilizing the change in cross-sectional area to generate a local flow velocity increase. The one-way straight-through valve 301 is a valve assembly that allows only unidirectional flow, which can be implemented using a spring-loaded check valve to prevent backflow from interfering with system stability. The gas pressure gauge 401 is a sensing device that monitors fluid pressure in real time, which can be implemented using a piezoresistive sensor, converting the pressure signal into an electrical signal output.

[0051] Specifically, when the high-pressure mixture enters the inlet 30, the one-way valve 301 ensures unidirectional fluid flow, and the flow control valve 302 adjusts its opening according to the controller's instructions. As the material passes through the abrupt change zone 102, the flow velocity increases, and some pressure energy is converted into kinetic energy. Subsequently, the material passes through multiple layers of filter screens 201. Each layer of filter screen 201 creates a pressure gradient due to differences in pore size, causing the fluid pressure to decrease progressively. The gas pressure gauge 401 continuously monitors the pressure at the outlet 40 and feeds the data back to the controller. The controller dynamically adjusts the opening of the flow control valve 302 by comparing the measured pressure with the set value, maintaining the pressure distribution within the filter cavity 101 within a predetermined range. The pressure difference between each layer of filter screen 201 is controlled below a critical value to prevent severe shearing that could cause bubble rupture.

[0052] Compared to existing technologies, traditional devices using a single-layer filtration structure are prone to turbulence due to a sudden pressure drop after the fluid passes through the filter screen 201, resulting in uneven bubble distribution. This solution uses multiple filtration units 20 to create a gradual pressure gradient, allowing the fluid's kinetic energy to be released in stages. The abrupt change in cross-section 102 alters the traditional straight-through flow channel structure, utilizing the Venturi effect to accelerate the initial pressure release. The closed-loop control system overcomes the limitation of fixed valve openings in adapting to flow fluctuations, ensuring that each filter layer operates within its optimal pressure range through real-time adjustment.

[0053] Through the above technical solutions, this application effectively solves the problem of insufficient filtration shear efficiency under high flow conditions. The multi-layer filtration structure disperses fluid kinetic energy, avoiding damage to the bubble structure caused by sudden local pressure changes. Pressure gradient control enables uniform nucleation of carbon dioxide seed cells during the gradual decompression process, significantly improving the cell uniformity of polyurethane foam. The closed-loop regulation mechanism ensures stable operation of the system under different flow conditions, improving the control accuracy of the production process.

[0054] This application further proposes that the pore size of the multilayer filter unit 20 satisfies:

[0055]

[0056] in, d0 is the pore size of the k-th filter layer 201, k is the reference pore size, k is the sequence number increasing along the fluid direction, and α is the pore size expansion coefficient. When α < 1, the pore size growth slows down, which is suitable for high viscosity formulations; when α > 1, the pore size growth accelerates, which is suitable for high gas content formulations.

[0057] in, dk refers to the pore size of the k-th filter layer 201 along the fluid flow direction. This can be achieved using laser drilling, with the pore size accuracy controlled by adjusting the laser power and focusing parameters. These parameters directly affect the fluid shear strength and gas release rate. d0 refers to the reference pore size of the initial filter layer, which can be determined through raw material viscosity testing. The pore size is matched after measuring the rheological properties of the mixture using a viscometer. This reference value serves as the design starting point for the entire filtration system. k refers to the sequential numbering of the 201 filter layers along the fluid direction, using incremental integers starting from the first layer at the inlet. This numbering establishes the spatial relationship between layers. α refers to the pore size expansion ratio coefficient between adjacent filter layers, obtained through training with historical formulation data from a material database. The optimal coefficient range is determined through regression analysis, and this coefficient determines the rate of change of the pore size gradient.

[0058] Specifically, when processing high-viscosity formulations, setting α causes the pore size to increase exponentially with decreasing velocity. As the fluid flows, it encounters progressively smaller pores, enhancing shear and preventing the formation of large bubbles due to insufficient shear in high-viscosity fluids. When processing formulations with high gas content, setting α causes the pore size to expand exponentially, rapidly increasing the flow cross-sectional area, reducing flow resistance, promoting orderly gas release, and preventing sudden pressure drops that could lead to bubble structure collapse. The reference pore size d0 is dynamically adjusted based on the initial viscosity of the raw material, ensuring that the first-layer filter 201 effectively breaks up large particles without causing excessive clogging.

[0059] Compared to existing technologies, traditional fixed-pore-size filter structures cannot adapt to raw materials with different physical properties, resulting in insufficient shearing for high-viscosity formulations or excessively rapid decompression for high-gas-content formulations. The linear pore-size variation method used in existing technologies exhibits adjustment lag when dealing with sudden pressure differences, while the exponential pore-size gradient achieves a smooth transition of the pressure field through nonlinear changes, providing strong shearing at the inlet and controllable decompression at the outlet.

[0060] Through the above technical solution, this application solves the problem of excessively large bubble size and uneven distribution caused by insufficient shear in high-viscosity formulations, while avoiding bubble collapse defects caused by uncontrolled pressure release in high-gas-content formulations. For high-viscosity systems, a progressively decreasing pore size gradient continuously applies shear force to refine bubble size; for high-gas-content systems, a progressively increasing pore size gradient establishes a gradual pressure release channel to maintain the stability of the bubble structure.

[0061] This application further proposes a controller with an adaptive PID control algorithm:

[0062]

[0063] Where Δu(t) is the valve opening adjustment amount; e(t) = ΔP(t) represents the real-time pressure deviation; Kp′ is the dynamic proportional gain parameter, which controls the response intensity to the current pressure deviation; Ki′ is the dynamic integral gain parameter, used to eliminate historical accumulated errors; Kd′ is the dynamic differential gain parameter, used to suppress pressure fluctuation lead control; Δt is the differential time window, used for the sampling interval of pressure deviation differential calculation. The parameter update rules for Kp′, Ki′, and Kd′ are as follows: Kp0, Ki0, and Kd0 are the initial values ​​of PID parameters; the learning rate coefficient is obtained through training with historical data in the database; the curvature sensitivity coefficient is adaptively selected according to the raw material viscosity; the hyperbolic tangent function is used to limit the parameter adjustment range to the (-1,1) interval.

[0064] Among them, the dynamic proportional gain parameter Kp′ refers to the proportional coefficient that is adjusted in real time according to the pressure fluctuation amplitude. Specifically, it can be achieved by using the real-time pressure deviation signal collected by the pressure sensor and calculating the absolute value of the pressure deviation change rate. The dynamic integral gain parameter Ki′ refers to the integral coefficient that is adjusted according to the deviation duration and change rate. Specifically, it can be achieved by using a sliding time window to statistically calculate the accumulated deviation and combining it with a threshold judgment of the deviation change rate to achieve integral action adjustment. The dynamic differential gain parameter Kd′ refers to the differential coefficient that is adjusted according to the pressure change trend. Specifically, it can be achieved by calculating the pressure deviation difference within the differential time window and combining it with the response of the curvature sensitivity coefficient to the raw material viscosity to achieve anticipatory control. The hyperbolic tangent function is an activation function that limits the parameter adjustment amplitude to the nonlinear saturation range. Specifically, it can be achieved by using a mathematical operation module to perform nonlinear transformation on the parameter increment to prevent the parameter adjustment amplitude from being too large and causing system instability.

[0065] Specifically, this technical solution calculates the pressure deviation value by collecting outlet pressure data in real time, and inputs the deviation signal into an adaptive PID algorithm to generate valve opening adjustment commands. When a drastic pressure fluctuation is detected, the dynamic proportional gain parameter automatically increases the proportional action intensity to quickly suppress pressure oscillations; when the pressure deviation shows a continuous unidirectional shift, the dynamic integral gain parameter enhances the integral action to eliminate steady-state errors; when the pressure change rate exceeds a set threshold, the dynamic differential gain parameter captures the pressure change trend through a differential time window and adjusts the valve opening in advance to achieve proactive control. During parameter updates, the hyperbolic tangent function limits the adjustment range of each gain parameter to a reasonable range, the learning rate coefficient is dynamically optimized based on historical control effect data, and the curvature sensitivity coefficient is automatically matched based on the real-time detected raw material viscosity value, enabling the control algorithm to adapt to the material characteristics of different formulations.

[0066] Compared to existing technologies, traditional fixed-parameter PID controllers often exhibit lag or excessive oscillation when dealing with high-viscosity materials or sudden pressure fluctuations. Existing technologies are prone to saturation of the integral term, leading to control failure, and cannot automatically adjust control parameters according to material characteristics. This solution addresses the differentiated control response requirements of materials with varying viscosities through an adaptive adjustment mechanism for dynamic gain parameters. It effectively captures sudden pressure changes using sliding differential time windows and, combined with the nonlinear limiting function of the hyperbolic tangent function, avoids system instability caused by excessive parameter adjustment.

[0067] Through the above technical solution, this application achieves precise and stable control of the internal pressure of the emulsifier 10 during the polyurethane foaming process, effectively suppressing the problem of uneven distribution of carbon dioxide foam caused by pressure fluctuations. This solution can automatically adapt to viscosity changes of different raw material formulations, respond quickly and maintain pressure stability under sudden pressure changes, avoid adjustment failures caused by integral saturation in traditional control methods, and improve the stability of the foaming process and the uniformity of the cell structure.

[0068] This application further proposes a stepwise decompression control method for liquid carbon dioxide foamed polyurethane foam, including the following steps:

[0069] A method for controlling the stepwise decompression of liquid carbon dioxide-blown polyurethane foam includes the following steps:

[0070] Step S1: Obtain the standard pressure release value;

[0071]

[0072] Among them, P in For the inlet pressure of emulsifier 10, P out (t) represents the outlet pressure at time t, V0 represents the volume of the calibration mixture, and T0 represents the calibration duration;

[0073] Step S2: Calculate the pressure deviation in real time;

[0074]

[0075] Where ΔP(t) is the pressure deviation value, Vt is the real-time processing volume, and τ is the width of the sliding time window.

[0076] Step S3: Generate a pre-tuned signal based on the LSTM prediction model;

[0077]

[0078] Where δ represents the prediction time domain, λ represents the historical time-series data vector, and λ is the physical constraint weighting coefficient. For constraint functions;

[0079] Step S4: When At that time, the opening adjustment amount of the drive flow control valve 302 is adjusted;

[0080]

[0081] in, To normalize the pressure deviation, The rate of change of pressure, It is a logarithmic flow.

[0082] The standard pressure release value refers to a dynamic benchmark parameter constructed by the ratio of the calibrated mixture volume to the calibration time. Specifically, it can be obtained by real-time acquisition of inlet and outlet pressure data using pressure sensors, combined with preset calibration parameters, to provide an adaptive benchmark for different production batches. Real-time pressure deviation calculation refers to dynamically comparing the real-time processed volume through a sliding time window. Specifically, it can use an exponential weighted algorithm to process historical pressure data, eliminating instantaneous noise interference and reflecting pressure change trends. The LSTM prediction model is a time-series prediction module built based on a long short-term memory neural network. Specifically, it can be trained using historical pressure data, valve status data, and flow data to predict future pressure deviation trends in the time domain. The pre-adjustment signal refers to the valve adjustment command generated by combining physical constraints. Specifically, it can use constraint functions to correct the prediction results to prevent exceeding the safe operating range of the equipment. The opening adjustment of the flow control valve 302 refers to the valve action executed based on the collaborative judgment result of the predicted deviation and the measured deviation. Specifically, it can use a servo motor to drive the valve stem displacement, achieving millisecond-level response speed.

[0083] Specifically, in the pressure control process, the standard pressure release value is first calculated by the dynamic ratio of the inlet pressure monitoring value to the real-time outlet pressure value. This benchmark value is automatically adjusted according to changes in the volume of the calibrated mixture and the production time, solving the problem that traditional fixed thresholds cannot adapt to batch differences in raw materials. Subsequently, a sliding time window is used to perform integral calculations on the real-time processing volume, and historical pressure data is weighted using an exponential decay function, which preserves the characteristics of current pressure abrupt changes while smoothing out random fluctuations. In the prediction stage, continuous time-series data is input into a pre-trained LSTM model, which outputs predicted pressure deviation values ​​for multiple future sampling periods. At the same time, a constraint function constructed from fluid dynamics equations is introduced to correct the boundary of the prediction results. When the absolute value of the predicted deviation exceeds the current measured deviation and meets the preset threshold condition, the valve opening compensation mechanism is immediately triggered. This dual judgment condition effectively avoids malfunctions caused by a single threshold trigger.

[0084] Compared with existing technologies, traditional pressure control methods, which employ fixed pressure thresholds and proportional-integral-derivative (PID) control, exhibit response lag and overshoot during sudden flow changes, leading to pressure fluctuations in the foaming process exceeding process requirements. This method, by establishing a dynamic pressure benchmark model and combining a coordinated control mechanism of time-series prediction and physical constraints, initiates compensation actions before pressure fluctuations reach the trigger threshold, achieving proactive regulation of the foaming pressure.

[0085] Through the above technical solution, this application effectively solves the problem of uneven foaming caused by the response lag when the flow rate changes suddenly in traditional control methods. By the synergistic effect of dynamic pressure benchmark and prediction model, the compensation mechanism is activated in the early stage of pressure deviation, so that the pressure fluctuation amplitude inside the emulsifier 10 is reduced to the allowable range of the process, thus ensuring the uniformity of the polyurethane foam cell structure.

[0086] This application further proposes a step of performing multi-objective optimization after the production batch is completed.

[0087] Step S5: After the production batch is completed, perform multi-objective optimization;

[0088]

[0089] Where X is the optimization variable vector, Evalve represents the energy consumption for valve regulation, and η is the penalty factor coefficient. This is the density constraint penalty function.

[0090] This application integrates valve regulation energy consumption and foam density constraints into an optimization objective function through mathematical modeling. The optimization variable vector includes the opening degree of flow control valve 302, the switching frequency of the filter unit, and the pressure setpoint. The penalty factor coefficient is dynamically adjusted according to the density detection results. The density constraint penalty function uses a piecewise quadratic function to quantify the foam density deviation.

[0091] Among them, the optimized variable vector refers to the set of adjustable process parameters, which can be achieved by combining the opening degree of the flow control valve 302, the switching frequency of the filter unit, and the pressure set value. By adjusting these parameters, the fluid dynamics conditions in the production process can be changed.

[0092] Valve regulation energy consumption refers to the energy consumed by the control valve's operation. Specifically, it can be calculated by integrating the solenoid valve drive current, reflecting the energy loss caused by frequent valve operation.

[0093] The penalty factor coefficient is a weighted parameter for the degree of density constraint violation. Specifically, it can be implemented using an adaptive algorithm based on the historical batch density pass rate, which is used to balance energy consumption targets and quality requirements during the optimization process.

[0094] Among them, the density constraint penalty function refers to the mathematical model that quantifies the deviation of foam density from the standard. Specifically, it can be implemented by constructing a piecewise quadratic function, and a nonlinear penalty is applied when the density detection value exceeds the tolerance range.

[0095] Specifically, after completing a single production batch, the system automatically collects data on the number of valve actions, foam density distribution, and pressure fluctuation curves for that batch. The adjustment range of the flow control valve 302 opening, the switching frequency of the filter unit, and the pressure setpoint offset are input into the multi-objective model as optimization variables. During optimization, the valve adjustment energy consumption term constrains unnecessary frequent valve actions, and the density constraint penalty function applies quality constraints by calculating the variance of the density distribution across the foam cross-section. The penalty factor coefficient dynamically adjusts the penalty intensity based on the density pass rates of the previous three batches. The model solves for the Pareto optimal solution set to select the parameter combination that reduces energy consumption while meeting density tolerance requirements, which serves as the initial setpoint for the next production batch.

[0096] Compared to existing technologies, current polyurethane foam production systems only use fixed parameters for batch production and lack an optimization model linking energy consumption and quality indicators, resulting in parameter adjustments lagging behind changes in raw material properties. This solution, by constructing a multi-objective optimization function, simultaneously optimizes valve operation economy and foam structure uniformity during batch-to-batch iterations, resolving the energy consumption and quality contradictions caused by traditional single-objective control.

[0097] Through the above technical solution, this application achieves closed-loop optimization and iteration of production parameters, reducing the adjustment energy consumption of the flow control valve 302 by approximately 18%-22% while ensuring that the foam density distribution meets the standards. At the same time, the optimized parameter combination controls the difference in cell size between different batches within ±5μm, effectively reducing the uneven cell structure caused by parameter solidification.

[0098] In this document, the terms "upper," "lower," "front," "back," "left," "right," "top," "bottom," "inner," "outer," "vertical," and "horizontal," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only used for the clarity of expressing the technical solution and for the convenience of description, and therefore should not be construed as limiting the present invention.

[0099] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0100] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A step-by-step decompression control device for liquid carbon dioxide foamed polyurethane foam, characterized in that, include: Emulsifier (10), which has a filter cavity (101) extending along the axis inside; A multi-layer filter unit (20) is disposed in a filter cavity (101) and includes at least five layers of filter screens (201) with different pore sizes. Each layer of filter screen (201) is arranged along the axial direction of the filter cavity. The inner diameter of the emulsifier (10) is larger than the inner diameter of the feed inlet, forming a cross-sectional abrupt change zone (102) to accelerate pressure release. The feed inlet (30) is installed at one end of the emulsifier (10) and communicates with the filter cavity (101), and is equipped with a one-way straight valve (301) and a flow control valve (302) connected in series. The discharge port (40) is installed at the other end of the emulsifier (10) and communicates with the filter cavity (101), and is equipped with a gas pressure gauge (401); The controller is communicatively connected to the one-way straight-through valve (301), the flow control valve (302), and the gas pressure gauge (401); The controller has an adaptive PID control algorithm: Where, Δ u ( t ) represents the valve opening adjustment amount; e(t) = ΔP(t), which represents the real-time pressure deviation; Kp′ is the dynamic proportional gain parameter, which controls the response intensity to the current pressure deviation; Ki′ is the dynamic integral gain parameter, which is used to eliminate historical accumulated errors and improve steady-state accuracy; Kd′ is the dynamic differential gain parameter, which is used to suppress pressure fluctuation lead control; Δt is the differential time window, which is used for the sampling interval of pressure deviation differential calculation; The parameter update rules for Kp′, Ki′, and Kd′ are as follows: Where Kp0, Ki0, and Kd0 are the initial values ​​of the PID parameters; , and The learning rate coefficient is obtained through training with historical data from the database. is the curvature sensitivity coefficient, which is adaptively selected according to the viscosity of the raw material; tanh() is the hyperbolic tangent function, used to adjust the range of Kp′, Ki′ and Kd′ within (-1,1); This represents the rate of change of pressure. Dynamic proportional gain parameter Kp′: When a drastic pressure fluctuation is detected. When the value is large, the proportional term is automatically enhanced to suppress oscillations; during the steady phase, Kp′ is reduced to avoid overshoot; Dynamic integral gain parameter Ki′: enhances the integral effect for continuous unidirectional deviation, and automatically weakens the integral to prevent saturation when the deviation changes rapidly; Dynamic differential gain parameter Kd′: In the initial stage of pressure change, Larger and At the same time, the differential term is enhanced to achieve proactive adjustment.

2. The liquid carbon dioxide foamed polyurethane foam stepwise decompression control device according to claim 1, characterized in that, The pore size of the multi-layer filter unit (20) satisfies: in, d0 is the pore size of the k-th filter layer, k is the reference pore size, k is the sequence number increasing along the fluid direction, and α is the pore size expansion coefficient. When α < 1, the pore size growth slows down, which is suitable for high viscosity formulations; when α > 1, the pore size growth accelerates, which is suitable for high gas content formulations.

3. A method for controlling the stepwise decompression of liquid carbon dioxide-blown polyurethane foam, characterized in that, Includes the following steps: Step S1: Obtain the standard pressure release value; Among them, P in P is the emulsifier inlet pressure. out (t) represents the outlet pressure at time t, V0 represents the volume of the calibration mixture, and T0 represents the calibration duration; Step S2: Calculate the pressure deviation in real time; Where ΔP(t) is the pressure deviation value, Vt is the real-time processing volume, and τ is the width of the sliding time window. Step S3: Generate a pre-tuned signal based on the LSTM prediction model; Where δ represents the prediction time domain, λ represents the historical time-series data vector, and λ is the physical constraint weighting coefficient. For constraint functions; Step S4: When At that time, the opening of the flow control valve is adjusted. in, To normalize the pressure deviation, The rate of change of pressure, It is a logarithmic flow.

4. The method for controlling the stepwise decompression of liquid carbon dioxide foamed polyurethane foam according to claim 3, characterized in that, Further steps include batch optimization: Step S5: After the production batch is completed, perform multi-objective optimization; Where X is the optimization variable vector, , This represents the average of the absolute values ​​of the pressure deviation. The maximum absolute value of the pressure deviation is given by Evalve, where Evalve is the valve regulation energy consumption, and η is the penalty factor coefficient. This is the density constraint penalty function.

Images