Inert gas generating system

Abstract

The invention provides an inert gas generation system, and relates to the technical field of industrial gas production. The inert gas generation system comprises the following modules: a gas pretreatment module, an intelligent separation module, a dynamic purification module, a pressure regulation and control module, a quality analysis module, a safety control module and an energy efficiency optimization module, and the gas pretreatment module is used for removing particulate matters through an electrostatic dust collection device by adopting a multi-stage gradient filtering algorithm based on ambient air input, adjusting humidity by combining a dew point control technology, forming a standardized gas source and generating a standard gas source. Through combination of a multi-stage gradient filtering algorithm and a dew point control technology, collaborative treatment of particulate matters and humidity is realized, stable gas source input is formed, the reliability of a subsequent processing link is improved, membrane separation parameters are dynamically predicted by applying a convolutional neural network model, and the permeation efficiency of a hollow fiber membrane stack is optimized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial gas production, in particular to an inert gas generation system. BACKGROUND

[0002] The technical field of industrial gas production mainly involves the industrial production system for preparing high-purity gas on a large scale through physical or chemical methods, and the core lies in the research and application of gas separation, purification and stable supply technology. This field covers key technologies such as cryogenic air separation, membrane separation, pressure swing adsorption, etc., and focuses on solving engineering problems such as accurate control of gas components, optimization of energy efficiency and safe and stable supply. Among them, the inert gas generation system refers to a complete set of equipment for continuously preparing inert gases such as nitrogen and argon through air separation technology, which is mainly used to create an oxygen-free or low-oxygen environment to prevent oxidation reaction and inhibit combustion and explosion, and plays a key role in the fields of petroleum storage and transportation, lithium battery manufacturing, aerospace, etc.

[0003] The traditional inert gas preparation system has limitations in dynamic working condition adaptability. The cryogenic air separation technology is subject to a fixed temperature curve and is difficult to quickly respond to load changes, resulting in significant energy consumption fluctuations. The membrane separation component relies on empirical parameter settings and lacks real-time permeation efficiency feedback mechanisms, which can easily lead to a decrease in separation efficiency when air composition fluctuates. In the pressure swing adsorption process, the molecular sieve regeneration period is fixed and cannot be dynamically adjusted according to the adsorption saturation level, resulting in energy waste and adsorbent loss. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides an inert gas generation system, which solves the problem of limitations of traditional inert gas preparation systems in dynamic working condition adaptability, and the cryogenic air separation technology is subject to a fixed temperature curve and is difficult to quickly respond to load changes, resulting in significant energy consumption fluctuations.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: an inert gas generation system, comprising the following modules: a gas pretreatment module, an intelligent separation module, a dynamic purification module, a pressure regulation module, a quality analysis module, a safety control module, and an energy efficiency optimization module; The gas pretreatment module is based on environmental air input and uses a multi-stage gradient filtration algorithm. It removes particulate matter through an electrostatic precipitation device and adjusts humidity using dew point control technology to form a standardized gas source and generate a standard gas source. The gas pretreatment module includes an air filtration sub-module, a temperature and humidity adjustment sub-module, and a pressure stabilization sub-module. The intelligent separation module is based on the standard gas source and uses a convolutional neural network permeation prediction model to achieve dynamic separation of oxygen / nitrogen through the selective permeation characteristics of the hollow fiber membrane component, generating primary inert gas. The intelligent separation module comprises a membrane stack array sub-module, a permeation efficiency monitoring sub-module, and a neural network prediction sub-module. The dynamic purification module is based on primary inert gas, uses a microwave-assisted adsorption method, utilizes the dynamic adsorption characteristics of molecular sieve materials, deeply removes residual oxygen through a variable temperature and pressure coupling technology, and generates high-purity inert gas. The dynamic purification module comprises a molecular sieve adsorption sub-module, a microwave regeneration sub-module, and a pressure oscillation control sub-module. The pressure regulation module is based on high-purity inert gas, implements a fuzzy PID control algorithm, realizes precise pressure stabilization of the output pressure through the synergistic effect of multi-stage centrifugal compressors and buffer tanks, and generates stable pressure inert gas flow. The pressure regulation module comprises a pressure sensing sub-module, a fuzzy control sub-module, and a gas flow buffering sub-module. The quality analysis module is based on stable pressure inert gas flow, uses laser spectroscopy detection technology, realizes real-time monitoring of gas component changes through multi-wavelength absorption characteristic analysis, and generates a purity analysis report. The quality analysis module comprises a laser detection sub-module, a spectral analysis sub-module, and a data calibration sub-module. The safety control module is based on the purity analysis report, uses a risk probability tree evaluation model, dynamically adjusts system operating parameters through a multi-parameter threshold interlocking mechanism, and generates safety control instructions. The safety control module comprises a risk evaluation sub-module, an emergency response sub-module, and an instruction issuing sub-module. The energy efficiency optimization module is based on safety control instructions, implements digital twin simulation optimization, dynamically adjusts device operating states through real-time mapping of energy consumption data and physical models, and generates a system energy efficiency map. The energy efficiency optimization module comprises a digital modeling sub-module, an energy consumption monitoring sub-module, and a dynamic optimization sub-module.

[0006] Preferably, the air filtration sub-module is based on environmental air input, uses a multi-stage gradient filtration algorithm, realizes hierarchical particle interception through an electrostatic dust removal device, and generates preliminary purified gas. The temperature and humidity adjustment sub-module is based on preliminary purified gas, uses dew point control technology, adjusts gas humidity through condensation-reheating circulation, and generates humidity-stable gas source. The pressure stabilization sub-module is based on humidity-stable gas source, implements a pressure feedback regulation method, uses a proportional valve to realize dynamic balance of inlet gas pressure, and generates a standard gas source.

[0007] Preferably, the membrane stack array sub-module is based on the standard gas source, uses a hollow fiber membrane stacking technology, realizes gas component separation through osmotic pressure difference, and generates preliminary separation gas. The permeation efficiency monitoring submodule, based on the primary inert gas, uses an optical fiber sensing detection method to monitor the permeation rate change on the membrane surface in real time, and generates permeation efficiency data; The neural network prediction submodule, based on the permeation efficiency data, constructs a convolutional neural network model to predict the optimal membrane stack operating parameter combination, and generates the primary inert gas.

[0008] Preferably, the molecular sieve adsorption submodule, based on the primary inert gas, uses a temperature swing adsorption method to selectively adsorb residual oxygen by molecular sieves, and generates a preliminary purified gas; The microwave regeneration submodule, based on the preliminary purified gas, implements a microwave radiation regeneration technology to excite the dissociation of adsorbed substances on the surface of the molecular sieve, and generates regenerated molecular sieves; The pressure oscillation control submodule, based on the regenerated molecular sieves, uses a pressure pulse oscillation algorithm to enhance the contact efficiency of the gas and the adsorbent, and generates high-purity inert gas.

[0009] Preferably, the pressure sensing submodule, based on the high-purity inert gas, uses a distributed pressure sensing technology to monitor the pressure fluctuation characteristics of the pipeline in real time, and generates a pressure fluctuation spectrum; The fuzzy control submodule, based on the pressure fluctuation spectrum, constructs a fuzzy PID control model to calculate the compressor speed adjustment parameters, and generates pressure regulation instructions; The airflow buffer submodule, based on the pressure regulation instructions, implements a multi-stage volume compensation method to smooth pressure pulsations through elastic energy storage devices, and generates a stable pressure inert gas flow.

[0010] Preferably, the laser detection submodule, based on the stable pressure inert gas flow, uses tunable laser absorption spectroscopy technology to obtain gas molecular characteristic absorption peaks, and generates spectral feature data; The spectral analysis submodule, based on the spectral feature data, uses multivariate regression analysis methods to analyze the oxygen / nitrogen concentration ratio, and generates component analysis results; The data calibration submodule, based on the component analysis results, implements a dynamic baseline correction algorithm to eliminate detection deviations caused by environmental interference, and generates a purity analysis report.

[0011] Preferably, the risk assessment submodule, based on the purity analysis report, constructs a Bayesian risk probability tree model to calculate the system operation risk level, and generates a risk level assessment; The emergency response submodule, based on the risk level assessment, uses a multi-threshold interlocking control method to trigger a hierarchical protection mechanism, and generates an emergency response strategy; The instruction issuing submodule, based on the emergency response strategy, implements an industrial bus communication protocol to transmit control instructions to the actuator, and generates safety control instructions.

[0012] Preferably, the digital modeling sub-module constructs a digital twin simulation model based on the safety control instruction, establishes a virtual mirror of the physical system, and generates a virtual system mirror; The energy consumption monitoring sub-module generates an energy consumption feature matrix by using an energy consumption data flow analysis technology to collect device operation energy consumption in real time based on the virtual system mirror; The dynamic optimization sub-module generates a system energy efficiency atlas by using a genetic evolution optimization algorithm to iteratively solve an optimal operation parameter combination based on the energy consumption feature matrix.

[0013] The present application provides an inert gas generation system. Has the following beneficial effects: The present application realizes the collaborative management of particulate matter and humidity by using a multi-stage gradient filtering algorithm combined with dew point control technology, forms a stable gas source input, improves the reliability of the subsequent processing link, dynamically predicts the membrane separation parameters by using a convolutional neural network model, optimizes the permeation efficiency of the hollow fiber membrane stack, enhances the separation precision of oxygen and nitrogen, uses microwave radiation and variable temperature and pressure coupling, excites the active sites on the surface of the molecular sieve, improves the adsorption and removal efficiency of residual oxygen, at the same time, through pressure pulse oscillation, enhances the gas-solid contact effect, adjusts the multi-stage compressor set based on the fuzzy PID control algorithm, combines the dynamic compensation mechanism of the buffer container, realizes the precise and stable control of the output pressure, uses the multi-wavelength characteristic analysis technology of laser spectrum to construct a real-time monitoring system of gas composition, eliminates environmental interference errors through dynamic baseline correction, establishes a risk probability tree evaluation model, combines a multi-parameter interlocking mechanism, realizes intelligent diagnosis and rapid response of the system operation state, uses digital twin technology to construct an energy consumption optimization model, dynamically optimizes through a genetic algorithm, and forms a closed-loop control of energy efficiency improvement. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The system block diagram of the present application is shown in the figure; Figure 2 The gas pretreatment module schematic diagram of the present application is shown in the figure; Figure 3 The intelligent separation module schematic diagram of the present application is shown in the figure; Figure 4 The dynamic purification module schematic diagram of the present application is shown in the figure; Figure 5 The pressure regulation module schematic diagram of the present application is shown in the figure; Figure 6 The quality analysis module schematic diagram of the present application is shown in the figure; Figure 7 The safety control module schematic diagram of the present application is shown in the figure; Figure 8 The energy efficiency optimization module schematic diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0015] 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.

[0016] Example: like Figures 1-8 As shown, this embodiment of the invention provides an inert gas generation system, including the following modules: a gas pretreatment module, an intelligent separation module, a dynamic purification module, a pressure regulation module, a quality analysis module, a safety control module, and an energy efficiency optimization module; The gas pretreatment module, based on ambient air input, employs a multi-level gradient filtration algorithm, removes particulate matter through an electrostatic dust removal device, and adjusts humidity using dew point control technology to form a standardized gas source. The gas pretreatment module includes an air filtration submodule, a temperature and humidity control submodule, and a pressure stabilization submodule; The intelligent separation module, based on a standard gas source, uses a convolutional neural network permeation prediction model and the selective permeation characteristics of the hollow fiber membrane module to achieve dynamic separation of oxygen and nitrogen, generating primary inert gas. The intelligent separation module includes a membrane stack array submodule, a permeation efficiency monitoring submodule, and a neural network prediction submodule; The dynamic purification module, based on primary inert gas, employs a microwave-assisted adsorption method. It utilizes the dynamic adsorption characteristics of molecular sieve materials and deeply removes residual oxygen through temperature and pressure coupling technology to generate high-purity inert gas. The dynamic purification module includes a molecular sieve adsorption submodule, a microwave regeneration submodule, and a pressure oscillation control submodule; The pressure regulation module, based on high-purity inert gas, implements a fuzzy PID control algorithm. Through the synergistic effect of a multi-stage centrifugal compressor and a buffer tank, it achieves precise pressure stabilization of the output pressure and generates a pressure-stabilized inert gas flow. The pressure control module includes a pressure sensing submodule, a fuzzy control submodule, and an airflow buffer submodule; The quality analysis module, based on a stable inert gas flow, uses laser spectroscopy detection technology to monitor changes in gas composition in real time through multi-wavelength absorption characteristic analysis and generates a purity analysis report. The quality analysis module includes a laser detection submodule, a spectral analysis submodule, and a data calibration submodule; The safety control module, based on the purity analysis report, adopts a risk probability tree assessment model and dynamically adjusts the system operating parameters through a multi-parameter threshold interlocking mechanism to generate safety control commands. The safety control module comprises a risk assessment submodule, an emergency response submodule, and an instruction issuing submodule. The energy efficiency optimization module, based on the safety control instruction, performs digital twin simulation optimization, dynamically adjusts the device operating state through real-time mapping of energy consumption data and physical models, and generates a system energy efficiency map. The energy efficiency optimization module comprises a digital modeling submodule, an energy consumption monitoring submodule, and a dynamic optimization submodule.

[0017] The air filtration submodule, based on environmental air input, uses a multi-stage gradient filtration algorithm to achieve hierarchical particle interception through electrostatic precipitation devices, generating preliminary purified gas. The environmental air input is processed through a three-stage gradient filtration structure. The first stage uses a metal woven filter screen to intercept large particles, the second stage uses an electret filter material to adsorb medium-sized suspended particles, and the third stage configures a bipolar plate electrostatic precipitation device. In the application of electronic component packaging workshops, for tin smoke particles generated during the welding process, the plate spacing and voltage gradient parameters are set, and the corona discharge intensity is dynamically adjusted according to real-time particle concentration monitoring data. When the optical sensor detects that the number of particles above 0.3 μm in the airflow exceeds the set threshold, the pulse vibration mechanism is triggered to remove dust from the plates. Through multi-stage cooperation, the output gas meets the cleanliness requirements, generating preliminary purified gas. The temperature and humidity adjustment submodule, based on the preliminary purified gas, uses dew point control technology to adjust gas humidity through condensation-reheating cycles, generating a humidity-stable gas source. The preliminary purified gas enters the temperature and humidity control unit. The condensation section reduces the gas temperature below the dew point through a plate heat exchanger, causing gaseous moisture to condense and precipitate. The reheating section uses a PID control algorithm to adjust the power of the heating element. In the application of pharmaceutical sterile workshops, the target humidity range is set according to the moisture permeability coefficient of pharmaceutical packaging materials. When the humidity sensor detects a deviation exceeding the tolerance, the compensation ratio of condensing agent flow and heating power is calculated through fuzzy logic, the adaptive model trained from historical data is used to predict the environmental temperature trend, and the cooperative working period of the dehumidification and humidification modules is dynamically adjusted. Finally, the output gas dew point fluctuation is stabilized within the process allowed interval, generating a humidity-stable gas source.

[0018] The pressure stabilization submodule, based on the humidity-stable gas source, implements a pressure feedback adjustment method, uses a proportional valve to achieve dynamic balance of the inlet pressure, and generates a standard gas source.

[0019] The humidity stable gas source is processed by a pressure closed loop control system, a plurality of differential pressure sensors are arranged in the main pipeline to collect dynamic pressure signals, when it is detected that the pressure fluctuation amplitude exceeds the stable threshold, the proportional valve controller calculates the valve core opening adjustment amount according to the composite index of the pressure change rate and the amplitude, in the glass coating production line gas supply system, the pressure reference value is set combining the gas flow demand curve, the pressure change trend of the pipeline network is predicted through the feedforward-feedback composite control strategy, when the instantaneous pressure deviation reaches the adjustment trigger condition, the actuator completes the valve position correction within the set response time, the pipeline resistance change is compensated by using the pressure-flow coupling model, the output pressure fluctuation amplitude is controlled within the process specification range, and the standard gas source is generated.

[0020] The membrane stack array sub-module generates preliminary separation gas based on the standard gas source and adopts hollow fiber membrane stacking technology to realize gas component separation through osmotic pressure difference. The standard gas source enters the spiral membrane stack group, the hollow fiber membranes are arranged in layers according to the osmotic coefficient difference, in the petrochemical industry cracking gas processing scene, the pressure gradient parameters between the membrane layers are set according to the ethylene and nitrogen separation requirements, when the raw gas passes through the membrane surface, oxygen preferentially permeates the membrane wall due to the higher osmotic coefficient, nitrogen is retained on the high pressure side, the permeation rate difference of different components is controlled by adjusting the membrane stack temperature and the gas flow ratio, in the continuous operation process, the pressure difference sensors are used to monitor the pressure change on both sides of each membrane layer, when it is detected that the pressure difference decay of a specific membrane layer exceeds the set threshold, the membrane group rotation mechanism is triggered to switch the working unit, the nitrogen concentration in the output gas reaches the process separation requirements, and the preliminary separation gas is generated.

[0021] The permeation efficiency monitoring sub-module generates permeation efficiency data by using the optical fiber sensing detection method to monitor the permeation rate change of the membrane surface in real time based on the preliminary separation gas. The preliminary separation gas flows through the annular optical fiber sensing array, distributed fiber grating sensors are arranged on the membrane surface, when the metal coating production line needs to monitor the nitrogen purity, the sensors capture the temperature and strain change signals on the membrane surface, a monitoring model is established through the mapping relationship between the light intensity attenuation rate and the permeation rate, a data processing unit converts the optical signal into a permeation efficiency index, when it is detected that the permeation efficiency fluctuation of a region exceeds the stable interval, a local pressure compensation mechanism is started, a time series prediction model trained according to historical data is used to adjust the working parameters of adjacent membrane groups in advance, permeation state data sets containing time stamps and spatial distribution characteristics are output, and the permeation efficiency data is generated.

[0022] The neural network prediction sub-module generates primary inert gas by constructing a convolutional neural network model based on the permeation efficiency data to predict the optimal membrane stack working parameter combination.

[0023] The permeation efficiency data input feature extraction engine extracts key features such as membrane stack temperature, pressure difference fluctuation frequency, and permeation rate gradient in the lithium battery separator production scene by using a sliding window mechanism to intercept time series fragments, constructs a three-dimensional convolution kernel scanning feature matrix, the first convolution layer captures local time correlation, the second convolution layer extracts spatial distribution patterns, and the fully connected layer outputs pressure adjustment, temperature compensation, and flow correction coefficient control parameter combinations after the data dimension is compressed by the pooling layer. When the model detects periodic fluctuation patterns in the permeation efficiency data, it automatically generates an optimized set of working parameters for the membrane stack array, outputs gas that meets the inert gas purity standard, and generates primary inert gas.

[0024] The molecular sieve adsorption sub-module uses a temperature swing adsorption method based on primary inert gas to selectively adsorb residual oxygen by molecular sieves to generate preliminary purified gas. The primary inert gas enters the multi-tower adsorption system, and in the semiconductor wafer manufacturing scene, the adsorption tower temperature gradient and pressure step parameters are set. When it is detected that the oxygen concentration in the gas exceeds the process standard, the temperature swing program is started to raise the molecular sieve bed to a specific temperature range, and the pressure in the tower is adjusted to an appropriate proportion at the same time, which promotes the expansion of the molecular sieve pores to release the adsorbed oxygen. In the continuous operation stage, the infrared spectrometer is used to monitor the desorption gas composition in real time. When the oxygen desorption amount reaches a certain proportion of the saturated adsorption amount, it switches to the cooling stage and restores the working pressure. The residual oxygen concentration in the output gas is reduced to a trace level, and preliminary purified gas is generated.

[0025] The microwave regeneration sub-module implements microwave radiation regeneration technology based on preliminary purified gas to excite the dissociation of adsorbates on the surface of the molecular sieve, and generates regenerated molecular sieve. The preliminary purified gas flows through the microwave regeneration cavity, and in the lithium battery material sintering furnace gas supply system, a multi-mode resonant cavity structure is configured. When the adsorption capacity of the molecular sieve decreases to a certain threshold, the microwave generator is started, the best irradiation position is determined by three-dimensional field intensity distribution simulation, the pulse duty cycle is adjusted to control the surface temperature rise rate of the molecular sieve, and the dielectric temperature rise characteristics are used to selectively heat the adsorbate. When the desorption gas flow monitoring value is continuously lower than the decay threshold for multiple cycles, it switches to the inert gas purge stage to remove residual desorption materials. The specific surface area of the regenerated molecular sieve is restored to near the initial level, and regenerated molecular sieve is generated.

[0026] The pressure oscillation control sub-module uses a pressure pulse oscillation algorithm based on regenerated molecular sieve to enhance the contact efficiency of gas and adsorbent, and generates high-purity inert gas.

[0027] The regenerated molecular sieve is loaded into a pulse fluidized bed, in a petrochemical hydrogenation device protection gas preparation scene, the pressure oscillation frequency and amplitude parameters are set, when the gas flow rate reaches the set multiple of the critical fluidization speed, the bed pressure drop fluctuation is monitored by a piezoelectric ceramic sensor, the pulse waveform parameters are adjusted according to the pressure drop spectrum analysis result, specific waveform pressure pulse is applied in the adsorption stage to enhance the gas diffusion rate, and corresponding pulse mode is adopted in the desorption stage to promote the desorption of adsorbate, when the bed void fraction monitoring value is stable in the fluidized state standard interval, the impurity component concentration in the output gas is reduced to a very low level, and high-purity inert gas is generated.

[0028] The pressure sensing sub-module is based on high-purity inert gas and uses distributed pressure sensing technology to monitor the pressure fluctuation characteristics of the pipeline in real time and generate a pressure fluctuation spectrum. The high-purity inert gas flows through the distributed sensing network, and in the petroleum storage and transportation pipeline monitoring scene, array-type piezoresistive sensors are arranged equidistantly along the pipeline axis, each monitoring node synchronously collects pressure signals, when the detected pressure fluctuation frequency exceeds the set threshold, a multi-source data fusion algorithm is started, the time domain signals of the adjacent three sensors are phase-aligned and amplitude-normalized, and the characteristic frequency components are extracted by wavelet transform, in the chemical industry park gas supply system, for the pressure sudden change caused by valve opening and closing, a sliding time window is set to analyze the pressure change gradient, when the pressure gradient in the continuous three sampling periods exceeds the stable interval, a data marking mechanism is triggered to generate a pressure fluctuation spectrum containing space-time distribution characteristics.

[0029] The fuzzy control sub-module is based on the pressure fluctuation spectrum and constructs a fuzzy PID control model to calculate the compressor speed adjustment parameters and generate a pressure regulation instruction. The pressure fluctuation spectrum input feature extraction engine decomposes the spectrum into amplitude, frequency, and phase three-dimensional feature vectors in the glass substrate coating equipment application, establishes a fuzzy rule base to define the membership functions of pressure deviation and deviation rate, and when high-frequency small-amplitude fluctuations and low-frequency large-amplitude fluctuations coexist, activates the parallel control strategy, the feedforward channel predicts the adjustment amount based on the pressure propagation model trained from historical data, and the feedback channel corrects the control output through a proportional factor adaptive algorithm, in the semiconductor etching machine gas supply scene, for periodic load changes, the matching coefficient of the control period and the device response time is set, and a multi-dimensional adjustment parameter set containing speed increment and action time is output.

[0030] The gas flow buffer sub-module is based on the pressure regulation instruction and implements a multi-stage volume compensation method to suppress pressure pulsation through an elastic energy storage device to generate a stable pressure inert gas flow.

[0031] Pressure regulating instruction drives multi-stage buffer system, in medical center oxygen supply system, primary buffer tank uses corrugated pipe structure to absorb high frequency pulsation, secondary buffer unit configures variable volume cavity, when detecting that pressure fluctuation amplitude exceeds compensation threshold, hydraulic actuator adjusts cavity volume according to instruction intensity, in steel continuous casting protection gas supply scene, set the correlation coefficient of volume compensation amount and pressure change rate, elastic energy storage device changes energy storage stiffness through pre-tightening force adjusting mechanism, when the system detects continuous low frequency fluctuation, start multi-stage coordinated mode to change cavity volume and deformation amount of energy storage element, output gas flow pressure fluctuation amplitude control within the process allowable range.

[0032] Laser detection sub-module, based on stable pressure inert gas flow, uses tunable laser absorption spectroscopy technology to obtain gas molecule characteristic absorption peak and generate spectral feature data; Stable pressure inert gas flow passes through multi-channel laser detection cavity, in semiconductor wafer processing scene, tunable laser emits light beam according to preset wavelength scanning range, when detecting trace oxygen in nitrogen, laser wavelength is locked near oxygen molecule specific absorption spectral line, photodetector receives transmitted light intensity signal, in glass coating production line application, set wavelength modulation frequency and gas flow rate matching parameters for argon purity monitoring demand, extract second harmonic signal through lock-in amplifier, when detecting that absorption peak intensity exceeds background noise threshold, start adjacent spectral line interference compensation mechanism, use reference gas chamber to collect environmental background spectrum for differential processing, output feature data set containing absorption peak position and intensity, generate spectral feature data.

[0033] Spectral analysis submodule, based on spectral feature data, uses multivariate regression analysis method to analyze oxygen / nitrogen concentration ratio and generates component analysis result; Spectral feature data input feature selection engine, in petrochemical cracking gas analysis scene, extract oxygen 760nm and nitrogen 746nm characteristic absorption peak area ratio, build multivariate regression model containing temperature compensation term and pressure correction factor, when detecting that environmental temperature fluctuation causes spectral line broadening, activate adaptive weight adjustment mechanism, according to historical training set, establish correction curve of absorption intensity and concentration relationship, in lithium battery electrolyte preparation scene, for oxygen and nitrogen mixed gas analysis in argon, use principal component analysis method to eliminate cross-sensitivity interference, solve concentration ratio matrix through iterative least squares method, output gas component analysis value containing confidence interval and error range, generate component analysis result.

[0034] Data calibration submodule, based on component analysis result, implements dynamic baseline correction algorithm to eliminate detection deviation caused by environmental interference, generates purity analysis report.

[0035] The component analysis result enters the dynamic calibration process, in the medical center oxygen supply system monitoring, a moving time window is set to collect the change trend data of the environment temperature and humidity, when it is detected that the change rate of the environment temperature exceeds the set threshold, the baseline drift compensation algorithm is started, the short-term fluctuation interference is eliminated through the sliding average filtering, in the steel heat treatment protective gas analysis scene, the calibration period is set to be an integer multiple of the gas sampling frequency, the Kalman filter is used to predict the baseline drift direction, the correction coefficient is adjusted according to the difference between the adjacent two calibration results, when the deviation amount in the continuous three calibration periods shows a convergence trend, the purity evaluation value compensated by the environmental interference is output, and the purity analysis report is generated.

[0036] The risk assessment submodule is based on the purity analysis report, a Bayesian risk probability tree model is constructed, the system operation risk level is calculated, and a risk level evaluation is generated; The purity analysis report is input into the risk modeling engine, in the petrochemical hydrogenation device protective gas monitoring scene, key parameters such as oxygen concentration, pressure fluctuation frequency and equipment running time are extracted, a Bayesian network node structure is constructed, when it is detected that the oxygen concentration exceeds the warning threshold, the conditional probability table updating mechanism is activated, the prior probability distribution is trained through historical fault data, in the lithium battery electrolyte preparation workshop application, for the abnormal situation of nitrogen purity, the risk propagation path weight coefficient is set, when multiple related parameters deviate from the normal interval at the same time, the probability propagation algorithm is executed to calculate the system level risk value, according to the interval where the risk value is located, three levels of safety, warning and danger are divided, and a risk level evaluation is generated.

[0037] The emergency response submodule is based on the risk level evaluation, adopts a multi-threshold interlocking control method, triggers a hierarchical protection mechanism, and generates an emergency response strategy; The risk level evaluation triggers a multi-level response mechanism, in the semiconductor wafer factory gas supply system, three interlocking threshold values are set, when the risk value enters the warning interval, the flow limiting mode is started to reduce the gas supply rate, when the risk value breaks through the danger threshold, the device shutdown protocol is activated and the standby gas source is started, in the glass substrate coating equipment protection scene, different risk level configuration response delay time parameters are configured, when a transient risk pulse is detected, the time weighted average algorithm is started to filter incidental interference, when the risk value is maintained in the high interval for continuous multiple sampling periods, the multi-device cooperative response strategy is triggered, and an emergency response strategy containing device state adjustment instructions and alarm level is generated.

[0038] The instruction issuing submodule is based on the emergency response strategy, implements an industrial bus communication protocol, transmits control instructions to an executing mechanism, and generates a safety control instruction.

[0039] The emergency response strategy is processed by the protocol conversion engine. In the steel continuous casting protective gas control system, the strategy instruction is encoded into an industrial bus standard data frame. When multiple device linkage needs to be executed, a timestamp synchronization mechanism is used to ensure that the instruction reaches the target device at the same time. In the medical center oxygen supply system application, a command retransmission mechanism and a response timeout threshold are set. When the actuator does not feedback the state signal within the specified time, the standby control channel is started to send redundant instructions. The data transmission integrity is ensured through cyclic redundancy check to generate a safe control instruction containing device address, control parameter and execution timing.

[0040] The digital modeling sub-module constructs a digital twin simulation model based on the safe control instruction, establishes a virtual mirror of the physical system, and generates a virtual system mirror. The safe control instruction input model construction engine extracts real-time parameters such as compressor speed, membrane separation temperature, and adsorption tower pressure in the semiconductor wafer factory inert gas supply system, establishes a three-dimensional topological structure mapping physical device space relationship, and synchronously updates the virtual model parameters when detecting device state updates through the OPC-UA protocol. In the petrochemical hydrogenation device protective gas system, the model update frequency is set to be an integer multiple of the device control period. When the actuator receives a new instruction, the virtual model synchronously loads the instruction parameters to preview the execution effect. The differential equation solver simulates the gas flow distribution change to construct a virtual system mirror containing device dynamic characteristics and pipe network topological structure.

[0041] The energy consumption monitoring sub-module uses energy consumption data stream analysis technology to collect device operation energy consumption in real time based on the virtual system mirror to generate an energy consumption feature matrix. The virtual system mirror connects the energy consumption perception network. In the lithium battery production line nitrogen supply scene, non-intrusive sensors are deployed to collect motor current, valve opening, cooling water flow and other indirect energy consumption indicators. The mapping relationship between device operating state and energy consumption is established through grey correlation analysis. When the membrane separation unit pressure fluctuation is detected, the high-frequency sampling mode is activated to capture transient energy consumption features. In the glass substrate coating equipment application, data cleaning rules are set to remove abnormal sampling points. The sliding window mechanism is used to extract time domain statistical features to construct a multi-dimensional feature matrix containing energy consumption intensity, fluctuation frequency and correlation dimension.

[0042] The dynamic optimization sub-module implements a genetic evolution optimization algorithm based on the energy consumption feature matrix to iteratively solve the optimal operating parameter combination and generate a system energy efficiency map.

[0043] The energy consumption characteristic matrix input optimization solver initializes the population containing compressor frequency, cooling water temperature, adsorption cycle and other parameter combinations in the medical center oxygen supply system optimization scenario, the fitness function comprehensively considers the energy consumption index and purity stability, when local optimum appears in the iteration process, the adaptive mutation operator is started to expand the search space, in the steel continuous casting protective gas system, the elite reservation strategy is set to maintain the population diversity, the non-inferior solution set is screened through the dominance relationship ranking, when the improvement amplitude of the optimal solution of continuous multiple generations is lower than the convergence threshold, the energy efficiency optimization scheme containing the Pareto frontier solution set is output, and the system energy efficiency atlas is generated.

[0044] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, the scope of the present application being defined by the appended claims and their equivalents.

Claims

1. An inert gas generating system characterized by The system comprises the following modules: a gas pretreatment module, an intelligent separation module, a dynamic purification module, a pressure regulation module, a quality analysis module, a safety control module, and an energy efficiency optimization module. The gas pretreatment module is based on ambient air input, adopts a multi-stage gradient filtration algorithm, removes particulate matter through an electrostatic dust removal device, adjusts humidity by using a dew point control technology, forms a standardized gas source, and generates a standard gas source. The gas pretreatment module comprises an air filtration sub-module, a temperature and humidity adjustment sub-module, and a pressure stabilization sub-module. The intelligent separation module is based on the standard gas source, uses a convolutional neural network penetration prediction model, realizes dynamic separation of oxygen / nitrogen through the selective penetration characteristics of hollow fiber membrane components, and generates primary inert gas. The intelligent separation module comprises a membrane stack array sub-module, a penetration efficiency monitoring sub-module, and a neural network prediction sub-module. The dynamic purification module is based on the primary inert gas, uses a microwave-assisted adsorption method, utilizes the dynamic adsorption characteristics of molecular sieve materials, deeply removes residual oxygen through a variable temperature and pressure coupling technology, and generates high-purity inert gas. The dynamic purification module comprises a molecular sieve adsorption sub-module, a microwave regeneration sub-module, and a pressure oscillation control sub-module. The pressure regulation module is based on high-purity inert gas, implements a fuzzy PID control algorithm, realizes precise pressure stabilization of the output pressure through the synergistic effect of multi-stage centrifugal compressors and buffer tanks, and generates a pressure-stabilized inert gas stream. The pressure regulation module comprises a pressure sensing sub-module, a fuzzy control sub-module, and a gas flow buffering sub-module. The quality analysis module is based on the pressure-stabilized inert gas stream, uses a laser spectroscopy detection technology, realizes real-time monitoring of gas composition changes through multi-wavelength absorption characteristic analysis, and generates a purity analysis report. The quality analysis module comprises a laser detection sub-module, a spectral analysis sub-module, and a data calibration sub-module. The safety control module is based on the purity analysis report, adopts a risk probability tree evaluation model, dynamically adjusts system operating parameters through a multi-parameter threshold interlocking mechanism, and generates safety control instructions. The safety control module comprises a risk assessment sub-module, an emergency response sub-module, and an instruction issuing sub-module. The energy efficiency optimization module is based on the safety control instructions, implements digital twin simulation optimization, dynamically adjusts equipment operating states through real-time mapping of energy consumption data and physical models, and generates a system energy efficiency map. The energy efficiency optimization module comprises a digital modeling sub-module, an energy consumption monitoring sub-module, and a dynamic optimization sub-module.

2. The inert gas generating system according to claim 1, wherein: The air filtration sub-module is based on ambient air input, adopts a multi-stage gradient filtration algorithm, and realizes hierarchical interception of particulate matter through an electrostatic dust removal device to generate preliminary purified gas. The temperature and humidity adjustment sub-module is based on the preliminary purified gas, uses a dew point control technology, adjusts gas humidity through condensation-reheating circulation, and generates a humidity-stabilized gas source. The pressure stabilization sub-module is based on the humidity-stabilized gas source, implements a pressure feedback regulation method, uses a proportional valve to achieve dynamic balance of the inlet gas pressure, and generates a standard gas source.

3. The inert gas generation system of claim 1, wherein: The membrane stack array sub-module is based on the standard gas source, adopts a hollow fiber membrane stacking technology, realizes gas component separation through a penetration pressure difference, and generates preliminary separation gas. The permeation efficiency monitoring submodule, based on the primary separation gas, uses an optical fiber sensing detection method to monitor the permeation rate change on the membrane surface in real time, and generates permeation efficiency data; The neural network prediction submodule, based on the permeation efficiency data, constructs a convolutional neural network model to predict the optimal membrane stack operating parameter combination, and generates primary inert gas.

4. The inert gas generating system according to claim 1, wherein: The molecular sieve adsorption submodule, based on the primary inert gas, uses a temperature swing adsorption method to selectively adsorb residual oxygen by molecular sieves, and generates preliminary purified gas; The microwave regeneration submodule, based on the preliminary purified gas, implements microwave radiation regeneration technology to excite the dissociation of adsorbed substances on the surface of the molecular sieve, and generates regenerated molecular sieve; The pressure oscillation control submodule, based on the regenerated molecular sieve, uses a pressure pulse oscillation algorithm to enhance the contact efficiency of gas and adsorbent, and generates high-purity inert gas.

5. The inert gas generation system of claim 1, wherein: The pressure sensing submodule, based on the high-purity inert gas, uses a distributed pressure sensing technology to monitor the pressure fluctuation characteristics of the pipeline in real time, and generates a pressure fluctuation spectrum; The fuzzy control submodule, based on the pressure fluctuation spectrum, constructs a fuzzy PID control model to calculate the compressor speed adjustment parameters and generates pressure regulation instructions; The airflow buffer submodule, based on the pressure regulation instructions, implements a multi-stage volume compensation method to smooth pressure pulsations through elastic energy storage devices, and generates stable pressure inert gas flow.

6. The inert gas generation system of claim 1, wherein: The laser detection submodule, based on the stable pressure inert gas flow, uses tunable laser absorption spectroscopy technology to obtain gas molecular characteristic absorption peaks, and generates spectral feature data; The spectral analysis submodule, based on the spectral feature data, uses multivariate regression analysis methods to analyze the oxygen / nitrogen concentration ratio and generates component analysis results; The data calibration submodule, based on the component analysis results, implements a dynamic baseline correction algorithm to eliminate detection deviations caused by environmental interference, and generates a purity analysis report.

7. The inert gas generation system of claim 1, wherein: The risk assessment submodule, based on the purity analysis report, constructs a Bayesian risk probability tree model to calculate the system operation risk level and generates a risk level assessment; The emergency response submodule, based on the risk level assessment, uses a multi-threshold interlocking control method to trigger a hierarchical protection mechanism and generates an emergency response strategy; The instruction issuing submodule, based on the emergency response strategy, implements an industrial bus communication protocol to transmit control instructions to the actuator and generates safety control instructions.

8. The inert gas generation system of claim 1, wherein: The digital modeling submodule, based on the safety control instructions, constructs a digital twin simulation model to establish a virtual image of the physical system and generates a virtual system image; The energy consumption monitoring submodule, based on the virtual system image, uses energy consumption data stream analysis technology to collect device operation energy consumption in real time and generates an energy consumption feature matrix; The dynamic optimization submodule, based on the energy consumption feature matrix, implements a genetic evolution optimization algorithm to iteratively solve the optimal operating parameter combination and generates a system energy efficiency spectrum.

Images