Design method for multi-objective optimization of vibration isolation system of offshore floating hydrogen production platform equipment
By adopting the parallel configuration of carbon fiber reinforced composite materials and magnetorheological liquid dampers on the offshore floating hydrogen production platform, combined with improved genetic algorithms, the multi-source vibration problem is solved, wide-band vibration isolation and efficient energy dissipation are achieved, and the safety and reliability of the platform are improved.
Patent Information
- Application Number
- CN202510735613.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-04
- Publication Date
- 2025-08-29
AI Technical Summary
The existing offshore floating hydrogen production platform faces multi-source vibration problems in complex marine environments. Traditional materials and structural designs are difficult to achieve efficient vibration energy dissipation, lack flexibility and multi-objective optimization, and existing algorithms are difficult to adapt to different sea conditions.
The acoustic black hole structure is constructed using carbon fiber reinforced composite materials, combined with pre-compressed negative stiffness mechanism and magnetorheological fluid damper, parallel configuration, combined with improved non-dominant sorting genetic algorithm for multi-objective optimization, and a multi-source heterogeneous environment perception system and energy self-energy leveling mechanism are constructed to achieve wide-band vibration isolation and real-time vibration suppression.
0.01Hz-800Hz wide-band vibration isolation is achieved, band gap attenuation exceeds 40dB, vibration transmission rate is ≤0.15, energy consumption is ≤200W, equipment life attenuation rate is less than 0.01%/h, external energy dependence is <10%, greatly improving the platform's safety and reliability.
Smart Images

Figure CN120562301A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical fields of marine engineering and mechanical engineering, and in particular relates to a multi-objective optimization design method for a vibration isolation system of an offshore floating hydrogen production platform. Background Art
[0002] With the surge in global demand for clean energy, offshore floating hydrogen production technology, a highly promising emerging energy solution, is booming. Leveraging abundant ocean resources and offshore wind power, it converts seawater into hydrogen, offering new hope for achieving large-scale, sustainable green energy supply. Offshore floating hydrogen production platforms have become a research hotspot and development direction in the energy sector, with numerous countries and companies investing resources to promote the research, development, and application of related technologies.
[0003] However, offshore floating hydrogen production platforms face severe vibration issues during operation. On the one hand, the complex marine environment, such as violent wave fluctuations, turbulent currents, and unstable wind turbulence, can continuously subject the platform to irregular external force impacts, causing significant vibration. This can not only cause equipment connections to loosen and increase wear, reducing equipment life, but can also cause deformation of the internal structure of the electrolyzer, affecting hydrogen production efficiency and quality. On the other hand, the vibration generated by the operation of the platform's own equipment, such as motors and pumps, couples with external environmental vibrations, further exacerbating the vibration situation and even causing resonance, posing a serious threat to the overall structural safety of the platform and increasing the risk of safety accidents.
[0004] To solve these problems, a lot of research has been carried out at home and abroad. For example, the patent for the "Floating Power Generation Platform Anti-roll Heave Plate Device" applied for by China National Nuclear Corporation Offshore Nuclear Power Development Co., Ltd. arranges a liftable heave plate truss structure on both sides of the platform, and uses the added mass and damping effect of the heave plate to suppress the heave motion and reduce the roll. There is also literature support that it is necessary to explore the coupling mechanism between the floating wind turbine blades and the wind rotor wake, establish an aerodynamic load prediction method, suppress the second-order motion of the semi-submersible floating wind turbine, and avoid the high-frequency resonance of the tension leg floating wind turbine. However, existing research mostly focuses on a single vibration source or local structural optimization, lacks comprehensive consideration of multi-source vibration and system-level vibration isolation design, and it is difficult to meet the vibration isolation needs of offshore floating hydrogen production platforms under complex working conditions.
[0005] At present, the research in this field at home and abroad has the following major defects:
[0006] 1) In terms of material application, traditional vibration isolation materials such as rubber and metal springs are mostly used, and there is a lack of innovative application of new materials, making it difficult to achieve efficient vibration energy dissipation.
[0007] 2) From the perspective of structural design, most studies adopt conventional fixed configurations, which cannot flexibly adapt to the complex and changeable offshore vibration environment.
[0008] 3) Existing research rarely uses optimization algorithms, which makes it difficult to adapt to different sea conditions and achieve efficient multi-objective optimization solutions. Summary of the Invention
[0009] The present invention is made in view of the problems existing in the prior art. In terms of material application, carbon fiber reinforced composite materials are used to construct the acoustic black hole structure. According to the vibration energy propagation theory, the vibration energy absorption efficiency of the structure is related to parameters such as the thickness change rate, which can be approximated by the energy dissipation formula E dissipated =∫ A The structure is described by σ∈ηdA (where σ is stress, ∈ is strain, η is the material loss factor, and A is the structural area). Its gradient thickness design (from 10mm at the center to 0.2mm at the edge), combined with a curvature radius ranging from 0.5m to 5m, allows vibration energy to converge along specific paths within the material, dissipating through mechanisms such as internal friction. Theoretical calculations and experimental verification have demonstrated a vibration energy dissipation efficiency of at least 90%, effectively broadening the vibration isolation band and addressing the limitations of traditional materials in broadband vibration isolation.
[0010] From the perspective of structural design, the innovative composite negative stiffness-magnetorheological vibration isolation unit adopts a parallel configuration of a pre-compression negative stiffness mechanism and a magnetorheological fluid damper. According to the principle of mechanical balance, the restoring force F generated by the negative stiffness mechanism is ns =-k ns x(where k ns is the negative stiffness coefficient, x is the displacement), and under external vibration excitation, it generates a force opposite to the excitation direction to offset part of the vibration energy; the magnetorheological fluid damper is based on the magnetorheological effect, and changes the damping force by applying a magnetic field. Its damping force F d =cx˙ (c is the damping coefficient, x˙ is the velocity). The two work together to achieve a dynamic damping force range of 0.1kN to 50kN, with a response time of no more than 20ms. This allows for rapid adaptation to the complex and changing vibration environment at sea, overcoming the inflexibility of existing vibration isolation structures.
[0011] At the algorithm level, an improved non-dominated sorting genetic algorithm is used to solve the Pareto optimal solution set. The algorithm adopts an adaptive crossover mutation strategy. According to the evolutionary principle of the genetic algorithm, the crossover probability P c and mutation probability P m Dynamic adjustment. A higher crossover probability in the early stage can expand the exploration range of the solution space, while reducing the crossover probability and increasing the mutation probability in the later stage can enhance the local search ability. c The change of the linear function P c =P c0 -(P c0 -P cf )n / N max (where P c0is the initial crossover probability, P cf is the final crossover probability, N max The maximum number of iterations, n the current number of iterations), and the mutation probability are similar. This strategy can more efficiently find the optimal solution among multiple objectives (vibration transmissibility no greater than 0.15, system energy consumption no more than 200W, and equipment life decay rate less than 0.01% per hour), resolving the problem of traditional algorithms easily falling into local optimality.
[0012] To achieve the above objectives, the technical solution of the present invention is a multi-objective optimization method for designing a vibration isolation system for an offshore floating hydrogen production platform, which is characterized by comprising the following steps:
[0013] Step 1: Build a multi-source heterogeneous environmental perception system. This system uses a distributed piezoelectric fiber sensor array with a spatial resolution of no more than 0.5m and a sampling frequency of no less than 100Hz, along with a high-precision inertial navigation unit with an attitude angle accuracy of ±0.005 degrees and an IP69K salt spray resistance rating, to synchronously collect wave energy spectra, ocean current shear force, wind turbulence intensity, and the platform's six-degree-of-freedom motion parameters in real time. It also integrates an electrolyzer acoustic emission monitoring module with a frequency response range of 1mHz to 1MHz to analyze the catalyst microcrack growth rate and membrane electrode stress concentration factor in real time.
[0014] Step 2: Design a composite negative stiffness-magnetorheological vibration isolation unit, using a pre-compression negative stiffness mechanism and a magnetorheological fluid damper in parallel configuration. The damping force dynamic range is 0.1kN to 50kN, and the response time does not exceed 20ms. The acoustic black hole structure is used to achieve broadband vibration isolation from 0.01Hz to 800Hz, with a bandgap attenuation of no less than 40dB.
[0015] Step 3: Establish a self-powered leveling mechanism for the hydrogen-oxygen cycle. Utilize oxygen, a byproduct of electrolysis, to drive an array of piston-type pneumatic actuators. A single actuator must have an output force of no less than 5kN and a stroke accuracy of 0.1mm. Combined with a hydrogen pressure energy recovery turbine unit, the turbine isentropic efficiency must be no less than 92%, creating a self-closed-loop energy leveling system that reduces external energy dependency to less than 10%.
[0016] Step 4: Using a multi-objective collaborative optimization algorithm, with the optimization objectives of vibration transmissibility no greater than 0.15, system energy consumption no more than 200W, and equipment life attenuation rate less than 0.01% per hour, an improved non-dominated sorting genetic algorithm is used to solve the Pareto optimal solution set, with a population size of no less than 500 and a number of iterations of no less than 1000 generations.
[0017] Step 5: Implement real-time vibration suppression driven by edge computing by deploying FPGA-based distributed edge computing nodes with a computational delay of no more than 100 μs. Build a digital twin control architecture and synchronize the entire platform state through a time-sensitive network with a synchronization time error of less than 1 μs. Use an incremental federated learning framework to update the control strategy, with a model convergence time of no more than 10 minutes and local training data retention of no more than 24 hours.
[0018] Step 6: Build a dynamic risk assessment system: Define a vibration isolation failure probability threshold of 0.1, a life loss index threshold of 0.05, and an energy efficiency deviation threshold of 0.15; generate a three-dimensional risk cloud map using a random forest-Markov chain fusion algorithm, with a risk prediction confidence level of no less than 95%; integrate a Bayesian update mechanism to automatically refresh risk assessment parameters every 30 minutes;
[0019] Step 7: Design a multi-level response protection mechanism: Establish a three-level response threshold system: Level 1 response (80 ≥ comprehensive score ≥ 70) activates the negative stiffness mechanism preload adjustment, with a response time of no more than 50ms; Level 2 response (70 > comprehensive score ≥ 60) initiates multi-level magnetic field control of the magnetorheological fluid, with a field intensity adjustment range of 0.1T to 1.5T; Level 3 response (comprehensive score < 60) triggers the safe isolation of the electrolyzer and the release and solidification of hydrogen and oxygen, with an execution delay of no more than 200ms;
[0020] Step 8: System evolution based on physical constraint data enhancement: Generate extreme operating condition damage data sets every quarter through a long short-term memory network embedded with the vortex transport equation. The number of hidden layer nodes in the network is not less than 128, and the prediction mean square error does not exceed 0.01; drive the vibration isolation parameter self-update algorithm, the number of parameter optimization iterations is not less than 500, and the error between the generated data and the measured data does not exceed 5%.
[0021] On the basis of the above technical solution, it is further defined that the acoustic black hole structure described in step 2 adopts a gradient variable thickness design: it is composed of a laminate of carbon fiber reinforced composite materials, the thickness gradually changes from 10 mm in the center to 0.2 mm at the edge, the curvature radius varies from 0.5 m to 5 m, and the vibration energy dissipation efficiency is not less than 90%.
[0022] On the basis of the above technical solution, it is further defined that the turbine unit in step 3 includes a three-stage radial impeller, a single-stage pressure ratio of 1.5 to 2.0, a speed range of 5000 rpm to 20000 rpm, and a hydrogen flow adaptation range of 5 kg / h to 50 kg / h.
[0023] On the basis of the above technical solution, it is further defined that the improved non-dominated sorting genetic algorithm described in step 4 adopts an adaptive crossover mutation strategy: the crossover probability decreases linearly from an initial value of 0.9 to 0.6, the mutation probability increases from 0.1 to 0.4, and the elite retention rate is 15%.
[0024] Based on the above technical solution, it is further specified that the incremental federated learning described in step 5 adopts AES-256 encrypted transmission, the key update cycle does not exceed 10 minutes, and the model differential compression rate is not less than 70%.
[0025] On the basis of the above technical solution, it is further defined that the parameter self-update algorithm in step 8 adopts the trust region reflective optimization method, and the convergence tolerance is set to 1×10 –6 , each update iteration takes no more than 120s.
[0026] The present invention is beneficial in that:
[0027] 1) Innovative materials and structural design achieve 0.01Hz-800Hz wide-band vibration isolation, with a bandgap attenuation exceeding 40dB and excellent vibration isolation effect.
[0028] 2) Innovative multi-objective collaborative optimization algorithm provides precise solutions, with vibration transmissibility ≤ 0.15, energy consumption ≤ 200W, and extended equipment life.
[0029] 3) Build an energy self-closed loop and intelligent response system, with external energy dependence less than 10%, greatly improving the security and reliability of the platform. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:
[0031] Figure 1 This diagram illustrates the overall architecture of the vibration isolation system for multi-objective optimization of offshore floating hydrogen production platform equipment. The figures are numeraled as follows: 101 - Multi-source heterogeneous environmental perception system; 102 - Composite negative stiffness-magnetorheological vibration isolation system; 103 - Oxygen-driven piston pneumatic actuator array; 104 - Magnetorheological fluid multi-stage magnetic field control system; 105 - Hydrogen-oxygen cycle self-powered leveling module.
[0032] Figure 2 Schematic diagram of the structure of the composite negative stiffness-magnetorheological vibration isolation unit. The reference numerals are as follows:
[0033] 201-Pre-compression negative stiffness mechanism; 202-Magnetorheological fluid damper; 203-Acoustic black hole structure; 204-Vibration damping base.
[0034] Figure 3 This is a schematic diagram of the hydrogen-oxygen cycle self-energy leveling mechanism. The accompanying figures are as follows:
[0035] 301 - oxygen piston pneumatic actuator array; 302 - hydrogen piston pneumatic actuator array; 303 - hydrogen pipeline; 304 - oxygen pipeline; 305 - pressure energy recovery turbine.
[0036] Figure 4 This is the flow chart of the multi-objective collaborative optimization algorithm.
[0037] Figure 5 Architecture diagram of edge computing-driven real-time vibration suppression and risk assessment. DETAILED DESCRIPTION
[0038] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.
[0039] As an embodiment of the present invention, a multi-objective optimization method for designing a vibration isolation system for an offshore floating hydrogen production platform is provided. The overall architecture of the system is shown in the attached diagram. Figure 1 As shown, the specific steps include:
[0040] Step 1: Build a multi-source heterogeneous environmental perception system. This system uses a distributed piezoelectric fiber sensor array with a spatial resolution of no more than 0.5m and a sampling frequency of no less than 100Hz, along with a high-precision inertial navigation unit with an attitude angle accuracy of ±0.005 degrees and an IP69K salt spray resistance rating, to synchronously collect wave energy spectra, ocean current shear force, wind turbulence intensity, and the platform's six-degree-of-freedom motion parameters in real time. It also integrates an electrolyzer acoustic emission monitoring module with a frequency response range of 1mHz to 1MHz to analyze the catalyst microcrack growth rate and membrane electrode stress concentration factor in real time.
[0041] Step 2: Design the composite negative stiffness-magnetorheological vibration isolation unit (see attached Figure 2 As shown), a pre-compression negative stiffness mechanism and a magnetorheological fluid damper are connected in parallel, the dynamic range of the damping force is 0.1kN to 50kN, and the response time does not exceed 20ms; the acoustic black hole structure is used to achieve 0.01Hz to 800Hz broadband vibration isolation, and the band gap attenuation is not less than 40dB; wherein, the acoustic black hole structure adopts a gradient variable thickness design: it is composed of a laminate of carbon fiber reinforced composite materials, the thickness gradually changes from 10mm in the center to 0.2mm at the edge, the curvature radius varies from 0.5m to 5m, and the vibration energy dissipation efficiency is not less than 90%.
[0042] Step 3: Establish a self-powered leveling mechanism for hydrogen and oxygen cycle (see attached Figure 3As shown), the electrolysis by-product oxygen is used to drive a piston pneumatic actuator array, with the output force of a single actuator not less than 5kN and the stroke accuracy reaching 0.1mm; combined with the hydrogen pressure energy recovery turbine unit, the turbine isentropic efficiency is not less than 92%, and an energy self-closed loop leveling system is constructed, reducing the dependence on external energy to less than 10%; wherein, the turbine unit includes a three-stage radial impeller, a single-stage pressure ratio of 1.5 to 2.0, a speed range of 5000rpm to 20000rpm, and a hydrogen flow adaptation range of 5kg / h to 50kg / h.
[0043] Step 4: With the help of multi-objective collaborative optimization algorithm (such as Figure 4 As shown in the figure, with the optimization objectives of vibration transmissibility not exceeding 0.15, system energy consumption not exceeding 200W, and equipment life attenuation rate less than 0.01% per hour, an improved non-dominated sorting genetic algorithm is used to solve the Pareto optimal solution set, with a population size of not less than 500 and the number of iterations of not less than 1000 generations; wherein, the improved non-dominated sorting genetic algorithm adopts an adaptive crossover and mutation strategy: the crossover probability decreases linearly from the initial value of 0.9 to 0.6, the mutation probability increases from 0.1 to 0.4, and the elite retention rate is 15%.
[0044] Step 5: Implement real-time vibration suppression driven by edge computing, deploy FPGA-based distributed edge computing nodes, and have a computing delay of no more than 100μs; build a digital twin control architecture, achieve full-platform state synchronization through a time-sensitive network, and have a synchronization time error of less than 1μs; use an incremental federated learning framework to update the control strategy, with a model convergence time of no more than 10 minutes and a local training data retention time of no more than 24 hours; wherein, the incremental federated learning uses AES-256 encrypted transmission, the key update cycle does not exceed 10 minutes, and the model differential compression rate is not less than 70%.
[0045] Step 6: Build a dynamic risk assessment system: define the vibration isolation failure probability threshold of 0.1, the life loss index threshold of 0.05, and the energy efficiency deviation threshold of 0.15; generate a three-dimensional risk cloud map through the random forest-Markov chain fusion algorithm, and the risk prediction confidence level is not less than 95%; integrate the Bayesian update mechanism to automatically refresh the risk assessment parameters every 30 minutes; the specific steps involved in steps 5 and 6 are as follows Figure 5 As shown;
[0046] Step 7: Design a multi-level response protection mechanism: Establish a three-level response threshold system: Level 1 response (80 ≥ comprehensive score ≥ 70) activates the negative stiffness mechanism preload adjustment, with a response time of no more than 50ms; Level 2 response (70 > comprehensive score ≥ 60) initiates multi-level magnetic field control of the magnetorheological fluid, with a field intensity adjustment range of 0.1T to 1.5T; Level 3 response (comprehensive score < 60) triggers the safe isolation of the electrolyzer and the release and solidification of hydrogen and oxygen, with an execution delay of no more than 200ms;
[0047] Step 8: System evolution based on physical constraint data enhancement: generate extreme operating condition damage data sets every quarter through a long short-term memory network embedded in the vortex transport equation. The number of hidden layer nodes in the network is not less than 128, and the prediction mean square error does not exceed 0.01; drive the vibration isolation parameter self-update algorithm, the number of parameter optimization iterations is not less than 500, and the error between the generated data and the measured data does not exceed 5%. The parameter self-update algorithm adopts the trust region reflective optimization method, and the convergence tolerance is set to 1×10 –6 , each update iteration takes no more than 120s.
[0048] Example 1
[0049] A floating hydrogen production platform deployed in an area of the South China Sea uses a traditional design that utilizes a combination of rubber vibration isolation pads and spring dampers. Sensors are placed only in key locations on the platform, and the data sampling frequency is only 20Hz, making it unable to accurately capture high-frequency wave vibrations. During typhoon season testing, the vibration transmissibility reached as high as 0.52, the equipment bolt loosening rate exceeded 30%, the system energy consumption remained above 450W for a long time, and the internal structural condition of the electrolyzers could not be effectively monitored.
[0050] The present invention is gradually advanced according to the above steps and will not be repeated here. Among them, in step 1, 128 distributed piezoelectric fiber sensors with a spatial resolution of 0.3m are densely deployed in a grid around the platform deck, columns and electrolytic cells. These sensors are like sensitive "tactile nerves" and can accurately sense tiny vibration changes; at the same time, 4 sets of high-precision inertial navigation units with an attitude angle accuracy of ±0.003 degrees and a salt spray resistance level of IP69K are equipped to work with the sensors to collect wave energy spectrum, ocean current shear force, wind turbulence intensity and platform six-degree-of-freedom motion parameters in real time at a high frequency of 120Hz. In addition, the integrated electrolytic cell acoustic emission monitoring module with a frequency response range covering 1mHz to 1MHz can analyze the catalyst microcrack propagation rate and membrane electrode stress concentration coefficient in real time, providing comprehensive and accurate data support for subsequent vibration isolation design.
[0051] In step 2, the key is to design a composite negative stiffness-magnetorheological vibration isolation unit. This innovative structure, which uses a pre-compressed negative stiffness mechanism in parallel with a magnetorheological fluid damper, gives the vibration isolation system powerful dynamic adjustment capabilities. The dynamic range of the damping force of the magnetorheological fluid damper can reach 0.1kN to 50kN, with a response time of no more than 18ms, enabling rapid response to external vibration changes. To achieve broadband vibration isolation, an acoustic black hole structure is applied to the electrolytic cell base. It is constructed from a laminate of carbon fiber reinforced composite materials, with a thickness that gradually changes from 10mm in the center to 0.2mm at the edge, and a fixed curvature radius of 3m. Based on the theory of vibration energy propagation, this gradient thickness design enables vibration energy to converge and dissipate along a specific path in the material. Testing has shown that the structure achieves broadband vibration isolation from 0.01Hz to 800Hz, with a bandgap attenuation of up to 45dB.
[0052] Step 3 establishes a self-powered leveling mechanism for the hydrogen-oxygen cycle. This mechanism utilizes oxygen, a byproduct of electrolysis, to drive an array of piston-type pneumatic actuators. Each actuator achieves an output force of no less than 5kN and a stroke accuracy of 0.1mm. Combined with a hydrogen pressure energy recovery turbine, the turbine achieves an isentropic efficiency of 92%, creating a self-enclosed energy leveling system that significantly reduces reliance on external energy. Step 4 leverages a multi-objective collaborative optimization algorithm, employing an improved non-dominated sorting genetic algorithm to find the Pareto optimal solution set, with the optimization objectives of a vibration transmissibility no greater than 0.15, system energy consumption no more than 200W, and an equipment life decay rate less than 0.01% per hour. The algorithm's efficient optimization is ensured by appropriately setting the population size and number of iterations. Steps 5-8 implement edge computing-driven real-time vibration suppression, construct a dynamic risk assessment system, design a multi-level response protection mechanism, and implement system evolution enhanced with physical constraint data. Each step is closely coordinated.
[0053] Measurements under typhoon conditions have shown that the implementation of this invention significantly reduced the platform's vibration transmissibility to 0.12, the equipment lifespan attenuation rate to only 0.008% / h, and system energy consumption to 180W. Compared to traditional designs, this design not only achieves a qualitative leap in vibration isolation performance but also significantly improves the platform's energy efficiency and equipment reliability through a self-powered mechanism and intelligent algorithm optimization, fully demonstrating the significant advantages of this invention in the design of vibration isolation systems for offshore floating hydrogen production platforms.
[0054] Example 2
[0055] On a platform located 80 kilometers offshore in the Bohai Sea, a traditional design relied on diesel generators for power, failing to utilize the energy from hydrogen production byproducts. The vibration isolation system had a response time of up to 150ms, making it unable to adjust promptly to sudden changes in ocean currents. During one strong current, the platform tilted by more than 8°, causing localized stress concentrations in the membrane electrodes within the electrolyzers.
[0056] In step 3 of this embodiment, targeting a platform with an average daily oxygen production of 2,000 cubic meters and a hydrogen output of 500 kg, a four-stage radial turbine unit was used, with a single-stage pressure ratio of 2.2 and a rotational speed maintained at 18,000 rpm. This achieved a high efficiency recovery of 94% of the hydrogen pressure energy. Combined with dynamic preload adjustment of the negative stiffness mechanism, the platform's tilt angle was controlled within 3° under the same current conditions, reducing external energy dependence from 100% to 8%, and energy consumption was further reduced by 10% compared to Example 1.
[0057] Example 3
[0058] On a platform in a deep-sea area of the Yellow Sea, a traditional vibration isolation system using a fixed-parameter PID control algorithm significantly degraded when subjected to the combined effects of complex surges (frequency 0.1-0.5 Hz) and high-frequency vibrations of the equipment (200-500 Hz). Measurements showed that the vibration transmissibility rose to 0.48 under these combined conditions.
[0059] This example enhances the algorithm in step 4 by expanding the population size of the improved non-dominated sorting genetic algorithm to 800, dynamically adjusting the crossover probability from 0.95 to 0.65, and incorporating the real-time adjustment function of the acoustic black hole structure's curvature radius from 0.8 to 4 meters in step 2. In a typical composite vibration test, the system automatically adjusted the curvature radius to 1.2 meters, increasing the bandgap attenuation to 50 dB and reducing the vibration transmissibility to 0.11, achieving a 20% faster convergence speed than in Example 1.
[0060] Example 4
[0061] On a 2,200-ton platform in a typhoon-prone area in the East China Sea, the traditional centralized control system experienced damage to three electrolyzer sensors due to delayed vibration isolation parameter adjustments caused by data transmission delays (over 500ms) during a typhoon.
[0062] This example upgrades the edge computing nodes in step 5, using the latest Xilinx FPGA chips, to reduce computational latency to 80μs. In step 6, a random forest-Markov chain model is trained based on three years of historical platform data, adding 12 risk factors, including typhoon paths and wave spectra. In a simulated typhoon test, the system issued a red alert 40 minutes in advance and, through rapid field strength adjustment of magnetorheological dampers, reduced peak vibration of key equipment by 65%.
[0063] Example 5
[0064] On a 1,000-ton platform in a shallow sea area of the Beibu Gulf (50 meters deep), the traditional single-threshold protection mechanism frequently triggers shutdowns during daily tidal vibrations, with 18 false triggers per year, seriously affecting hydrogen production efficiency.
[0065] This example optimizes the three-level response system in step 7, increasing the vibration severity weight to 45% for eight key equipment types, including platform electrolyzers and compressors. During an abnormal tidal vibration, the system triggered a first-level response based on a comprehensive score of 72. By increasing the preload of the negative stiffness mechanism by 15% and pre-magnetizing the magnetorheological fluid, the response time was shortened to 40ms, preventing unintended equipment shutdowns while keeping the vibration transmissibility below 0.13.
[0066] Example 6
[0067] On a platform in the South China Sea, after two years of operation, the traditional vibration isolation system experienced a mismatch in isolation parameters due to changes in the marine environment, and the vibration transmissibility increased from an initial 0.35 to 0.42.
[0068] This embodiment strengthens the data enhancement module in step 8. Based on the data of 200 long-term monitoring sensors deployed on the platform, an LSTM network with 150 hidden layer nodes is trained to generate 500 sets of extreme working condition simulation data every month. The parameter self-update algorithm sets the convergence tolerance to 5×10 –7 After one year of operation, the system automatically completed 12 parameter iteration optimizations, the vibration transmission rate stabilized at 0.10, and the equipment failure rate was reduced by 82% compared with the traditional system.
[0069] Comparison items
[0070]
[0071] The specific embodiments described above are only used to specifically illustrate the spirit of the present invention, and the scope of protection of the present invention is not limited thereto. For those skilled in the art, it is of course possible to easily make other embodiments by changing, replacing or modifying the technical contents disclosed in this specification, and these other embodiments should all be included in the scope of protection of the present invention.
Claims
1. A multi-objective optimization method for designing a vibration isolation system for an offshore floating hydrogen production platform, characterized in that: The following steps are involved: Step 1: Build a multi-source heterogeneous environmental perception system. This system uses a distributed piezoelectric fiber sensor array with a spatial resolution of no more than 0.5m and a sampling frequency of no less than 100Hz, along with a high-precision inertial navigation unit with an attitude angle accuracy of ±0.005 degrees and an IP69K salt spray resistance rating, to synchronously collect wave energy spectra, ocean current shear force, wind turbulence intensity, and the platform's six-degree-of-freedom motion parameters in real time. It also integrates an electrolyzer acoustic emission monitoring module with a frequency response range of 1mHz to 1MHz to analyze the catalyst microcrack growth rate and membrane electrode stress concentration factor in real time. Step 2: Design a composite negative stiffness-magnetorheological vibration isolation unit, using a pre-compression negative stiffness mechanism and a magnetorheological fluid damper in parallel configuration. The damping force dynamic range is 0.1kN to 50kN, and the response time does not exceed 20ms. The acoustic black hole structure is used to achieve broadband vibration isolation from 0.01Hz to 800Hz, with a bandgap attenuation of no less than 40dB. Step 3: Establish a self-powered leveling mechanism for the hydrogen-oxygen cycle. Utilize oxygen, a byproduct of electrolysis, to drive an array of piston-type pneumatic actuators. A single actuator must have an output force of no less than 5kN and a stroke accuracy of 0.1mm. Combined with a hydrogen pressure energy recovery turbine unit, the turbine isentropic efficiency must be no less than 92%, creating a self-closed-loop energy leveling system that reduces external energy dependency to less than 10%. Step 4: Using a multi-objective collaborative optimization algorithm, with the optimization objectives of vibration transmissibility no greater than 0.15, system energy consumption no more than 200W, and equipment life attenuation rate less than 0.01% per hour, an improved non-dominated sorting genetic algorithm is used to solve the Pareto optimal solution set, with a population size of no less than 500 and a number of iterations of no less than 1000 generations. Step 5: Implement real-time vibration suppression driven by edge computing and deploy distributed edge computing nodes based on FPGA, with computing latency no more than 100 μs. Build a digital twin control architecture and achieve full platform state synchronization through time-sensitive networking, with a synchronization time error of less than 1μs; use an incremental federated learning framework to update the control strategy, with model convergence time of no more than 10 minutes and local training data retention time of no more than 24 hours; Step 6: Build a dynamic risk assessment system: Define a vibration isolation failure probability threshold of 0.1, a life loss index threshold of 0.05, and an energy efficiency deviation threshold of 0.15; generate a three-dimensional risk cloud map using a random forest-Markov chain fusion algorithm, with a risk prediction confidence level of no less than 95%; integrate a Bayesian update mechanism to automatically refresh risk assessment parameters every 30 minutes; Step 7: Design a multi-level response protection mechanism: Establish a three-level response threshold system: Level 1 response (80 ≥ comprehensive score ≥ 70) activates the negative stiffness mechanism preload adjustment, with a response time of no more than 50ms; Level 2 response (70 > comprehensive score ≥ 60) initiates multi-level magnetic field control of the magnetorheological fluid, with a field intensity adjustment range of 0.1T to 1.5T; Level 3 response (comprehensive score < 60) triggers the safe isolation of the electrolyzer and the release and solidification of hydrogen and oxygen, with an execution delay of no more than 200ms; Step 8: System evolution based on physical constraint data enhancement: Generate extreme operating condition damage data sets every quarter through a long short-term memory network embedded with the vortex transport equation. The number of hidden layer nodes in the network is not less than 128, and the prediction mean square error does not exceed 0.01; drive the vibration isolation parameter self-update algorithm, the number of parameter optimization iterations is not less than 500, and the error between the generated data and the measured data does not exceed 5%.
2. The method according to claim 1, characterized in that The acoustic black hole structure described in step 2 adopts a gradient thickness design: it is composed of a laminate of carbon fiber reinforced composite materials, with a thickness gradually varying from 10 mm in the center to 0.2 mm at the edge, a curvature radius ranging from 0.5 m to 5 m, and a vibration energy dissipation efficiency of not less than 90%.
3. The method according to claim 1, characterized in that The turbine unit in step 3 includes a three-stage radial impeller, a single-stage pressure ratio of 1.5 to 2.0, a rotation speed range of 5000 rpm to 20000 rpm, and a hydrogen flow rate adaptability range of 5 kg / h to 50 kg / h.
4. The method according to claim 1, wherein The improved non-dominated sorting genetic algorithm described in step 4 adopts an adaptive crossover and mutation strategy: the crossover probability decreases linearly from the initial value 0.9 to 0.6, the mutation probability increases from 0.1 to 0.4, and the elite retention rate is 15%.
5. The method according to claim 1, characterized in that The incremental federated learning described in step 5 uses AES-256 encrypted transmission, the key update cycle does not exceed 10 minutes, and the model differential compression rate is not less than 70%.
6. The method according to claim 1, characterized in that The parameter self-update algorithm described in step 8 uses the trust region reflective optimization method, and the convergence tolerance is set to 1×10 –6 , each update iteration takes no more than 120s.