A smoke removal system for the forming chamber used in additive manufacturing of rocket engine thrust chambers

Abstract

The present invention relates to the field of additive manufacturing technology for rocket engine thrust chambers, and discloses a fume removal system for a forming chamber used for additive manufacturing of rocket engine thrust chambers. The system acquires multi-dimensional fume parameters through a multi-source data acquisition module, decomposes the features through a fume feature extraction module, generates a filtration strategy and an airflow optimization matrix through a dynamic filtration module, coordinates equipment operating parameters to achieve energy consumption balance through a collaborative control module, and identifies anomalies through an abnormality handling module that triggers a self-repair mechanism. The system can accurately collect and analyze fume data, dynamically optimize filtration and airflow paths, efficiently coordinate equipment operation, promptly handle anomalies, improve fume removal effects, reduce energy consumption, ensure operator safety and stable equipment operation, and promote the development of additive manufacturing technology for rocket engine thrust chambers.

Description

Technical Field

[0001] The present invention relates to the technical field of additive manufacturing of rocket engine thrust chambers, and in particular to a fume removal system for a forming chamber used for additive manufacturing of rocket engine thrust chambers. Background Art

[0002] In modern aerospace, rocket engines serve as core propulsion units, and their performance and quality play a decisive role in the success or failure of space missions. With the continuous advancement of science and technology, additive manufacturing technology, due to its unique advantages such as strong ability to manufacture complex structures, high material utilization, and short production cycles, has gradually been applied to the manufacture of rocket engine thrust chambers. However, this manufacturing process generates a large amount of complex fumes, which poses numerous challenges to environmental control and equipment operation within the build chamber. This makes an efficient build chamber fume removal system a key component in the development of additive manufacturing technology for rocket engine thrust chambers.

[0003] The additive manufacturing process for rocket engine thrust chambers involves the high-temperature melting and layer-by-layer deposition of multiple materials, generating fumes with complex compositions. From a temperature perspective, the temperature distribution within the molding chamber is extremely uneven, with significant temperature gradients. High temperatures in different areas not only affect the molding quality of the material but can also cause thermal damage to key components of the equipment, reducing its service life. For example, in the additive manufacturing process of some metal materials, excessively high local temperatures can lead to coarse grains in the material, affecting the mechanical properties of the product.

[0004] Regarding particulate matter, the concentration and size distribution of particulate matter in flue gas are complex. Small particles easily adhere to equipment surfaces, clogging precision components such as sensors and filters, leading to reduced measurement accuracy or even failure. Large particles, on the other hand, can settle under gravity at the bottom of the molding chamber, impacting normal operation and increasing cleaning and maintenance costs.

[0005] Gas composition is also not to be ignored. The additive manufacturing process produces a variety of harmful gases, such as carbon monoxide and nitrogen oxides. These gases not only pose a threat to the health of operators, but can also chemically react with metal components in the build chamber, accelerating component corrosion and reducing equipment reliability and safety.

[0006] Existing flue gas removal technologies have numerous shortcomings when addressing these issues. Traditional filtration methods often utilize fixed filter media and a single filter structure, making it impossible to dynamically adjust to real-time changes in the flue gas. When the concentration of particulate matter in the flue gas suddenly increases or the gas composition changes, the filtration effect decreases significantly, making it difficult to meet production needs. Furthermore, traditional systems lack effective coordination mechanisms when controlling the exhaust fan, filter device, and cooling unit. The operating parameters of each device cannot be optimally matched, resulting in excessive system energy consumption and low operating efficiency.

[0007] Furthermore, when it comes to handling abnormal situations, existing technologies mostly rely on manual inspections and subsequent repairs. When abnormal situations such as smoke leaks and equipment failures occur, they cannot be detected and handled promptly and accurately, which can easily lead to production interruptions and increase production costs and production cycles. Furthermore, existing technologies lack effective means for addressing potential risks, such as predicting the life of filter media and preventing toxic gases, making it difficult to ensure the safe and stable progress of the production process. In summary, the development of an efficient, intelligent, and stable smoke removal system for the additive manufacturing of rocket engine thrust chambers is urgent. Summary of the Invention

[0008] The object of the present invention is to provide a smoke removal system for a forming chamber used for additive manufacturing of a rocket engine thrust chamber, so as to solve the problems raised in the above-mentioned background technology.

[0009] To achieve the above-mentioned object, the present invention provides the following technical solution: a smoke removal system for a forming chamber for additive manufacturing of a rocket engine thrust chamber, the system comprising:

[0010] Multi-source data acquisition module: used to obtain multi-dimensional parameters of the flue gas in the molding chamber in real time, including temperature gradient distribution, particle concentration spectrum, gas component ratio and airflow velocity field;

[0011] Smoke feature extraction module: performs nonlinear feature decomposition on the multi-dimensional parameters to generate smoke diffusion pattern characteristics, particle aggregation trend characteristics and thermodynamic coupling response characteristics;

[0012] Dynamic filtration module: Based on the improved adaptive filtering algorithm, a multi-layer filtration network is constructed, the characteristics are input into the dynamic weight allocation unit, and the filter medium switching strategy and airflow path optimization matrix are generated;

[0013] Collaborative control module: Based on the filter medium switching strategy and the airflow path optimization matrix, the distributed reinforcement learning model is used to coordinate the operating parameters of the exhaust fan, filter device, and cooling unit to generate global energy consumption balance control instructions;

[0014] Abnormal processing module: Based on the density peak detection algorithm, it identifies abnormal fluctuations in the smoke removal process, triggers the self-repair mechanism and updates the filter network topology.

[0015] Preferably, the multi-source data acquisition module includes:

[0016] A distributed fiber optic sensor array is used to capture the temperature gradient distribution, and a spatiotemporal interpolation algorithm is used to reconstruct the three-dimensional thermal field evolution map.

[0017] The particle concentration spectrum is measured using polarization scattering measurement technology to separate the distribution density of particles of different particle sizes, and the measurement error is corrected in combination with the Kalman filter algorithm;

[0018] The spectral absorption feature fusion method is used to extract the key component proportions of the gas component ratio, and the gas diffusion boundary is predicted based on the chaos mapping model.

[0019] Preferably, the smoke feature extraction module further includes:

[0020] The turbulence core area is extracted from the airflow velocity field using a vortex decomposition algorithm, and the particle motion trajectories are correlated using a phase locking technique;

[0021] The temperature-airflow interaction is modeled using a tensor decomposition method for the thermodynamic coupling response characteristics, and a multi-scale energy transfer path characteristic matrix is generated.

[0022] Preferably, the dynamic filtering module further includes:

[0023] Construct a filter media life prediction model based on a double-layer gated circulation unit to calculate the filter layer porosity decay rate in real time;

[0024] The filter medium switching strategy is optimized by the gradient back-propagation algorithm, and the airflow path optimization matrix is integrated to generate dynamic pressure drop compensation instructions.

[0025] Preferably, the collaborative control module further includes:

[0026] The exhaust fan speed, filter pressure difference threshold, and cooling unit power threshold are defined as the joint action space, and a multi-objective Pareto optimization algorithm is used to solve the global optimal parameter combination.

[0027] Balance local control instruction conflicts through asynchronous policy update mechanism.

[0028] Preferably, the exception handling module further includes:

[0029] Using a density clustering algorithm to classify the abnormal fluctuations into risk levels, generate graded warning signals and associate them with trigger conditions for the self-repair mechanism;

[0030] When the particulate matter concentration exceeds the limit, the redundant filter layer deployment strategy is activated and the priority weight of the airflow path optimization matrix is reconstructed.

[0031] Preferably, the self-repair mechanism further includes:

[0032] A bidirectional graph neural network is used to model the coupling relationship between the filter device and the smoke exhaust fan, extracting fault conduction characteristics from the forward path and redundant resource scheduling characteristics from the reverse path.

[0033] A repair plan is generated through attention weight fusion, and the risk assessment parameters of the dynamic filtering network are updated.

[0034] Preferably, the system further comprises:

[0035] A flue gas toxicity index prediction model was constructed based on historical process data and real-time gas component parameters. The generation rules of thermal degradation products were integrated with a time series convolutional network to dynamically adjust the chemical adsorption priority of the filter medium.

[0036] Preferably, the smoke toxicity index prediction model further includes:

[0037] A generative adversarial network is used to simulate the gas mixture distribution under extreme working conditions, generate a toxicity risk probability map and embed the redundant filter layer deployment strategy.

[0038] Preferably, the processing of the turbulent core region further comprises:

[0039] An adaptive vortex tracking algorithm is used to balance the measurement accuracy of high-speed airflow and low-speed stagnation area. The nonlinear characteristics of vortex evolution are captured through the residual convolutional network. The airflow stability evaluation matrix is constructed and integrated into the weight allocation unit of the dynamic filtering network.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The smoke removal system for the molding chamber for additive manufacturing of the rocket engine thrust chamber of the present invention has significant beneficial effects in many aspects. In terms of data acquisition and analysis, the multi-source data acquisition module can obtain multi-dimensional parameters such as the temperature gradient distribution, particle concentration spectrum, gas component ratio and airflow velocity field of the smoke in the molding chamber in real time. The use of a distributed fiber optic sensor array to capture the temperature gradient distribution and reconstruct the three-dimensional thermal field evolution map can accurately grasp the temperature change trend in the molding chamber, provide strong data support for preventing material molding defects and equipment thermal damage caused by uneven temperature, and ensure the stability of the additive manufacturing process and product quality. The use of polarization scattering measurement technology combined with the Kalman filter algorithm to measure the particle concentration spectrum can more accurately grasp the particle distribution, effectively avoid failures caused by particle clogging of equipment components, and reduce equipment maintenance costs and downtime. The spectral absorption feature fusion method combined with the chaotic mapping model to obtain the gas component ratio and predict the diffusion boundary is helpful to take protective measures in advance to ensure the safety of operators and the normal operation of equipment.

[0042] In the filtration and control link, the dynamic filtration module constructs a multi-layer filtration network based on the improved adaptive filtering algorithm, and generates a filter medium switching strategy and airflow path optimization matrix according to the smoke characteristics. By constructing a filter medium life prediction model based on a double-layer gated circulation unit, the filter layer porosity attenuation rate can be calculated in real time, the filter medium can be switched in time, high-efficiency filtration performance can be maintained, and the service life of the filter device can be extended. The gradient backpropagation algorithm optimizes the filter medium switching strategy and generates dynamic pressure drop compensation instructions, which can ensure the stable operation of the system under different working conditions and reduce energy consumption. The collaborative control module uses a distributed reinforcement learning model to coordinate the operating parameters of the smoke exhaust fan, filter device and cooling unit, and uses a multi-objective Pareto optimization algorithm to solve the global optimal parameter combination to achieve global energy consumption balance control, improve the overall operating efficiency of the system, and effectively reduce energy consumption while ensuring the smoke removal effect, in line with the green manufacturing concept.

[0043] The exception handling capability is also a major advantage of this system. The exception handling module identifies abnormal fluctuations based on the density peak detection algorithm, uses the density clustering algorithm to divide the risk level and generate graded warning signals, and associates the trigger conditions of the self-repair mechanism. Once an abnormality is detected, the self-repair mechanism is quickly activated, and the bidirectional graph neural network models the coupling relationship between the filter device and the smoke exhaust fan, generates a repair plan and updates the risk assessment parameters to ensure that the system can still operate stably under abnormal conditions, reduce the number of production interruptions, and improve production efficiency. At the same time, the flue gas toxicity index prediction model dynamically adjusts the chemical adsorption priority of the filter medium according to real-time data, uses a generative adversarial network to simulate the gas mixture distribution under extreme working conditions, generates a toxicity risk probability map and embeds a redundant filter layer deployment strategy, which greatly enhances the system's ability to prevent toxic gases, ensuring the life safety of operators and the safety and reliability of the production environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a working principle diagram of a smoke removal system for a forming chamber used for additive manufacturing of a rocket engine thrust chamber according to the present invention;

[0045] Figure 2 This is the workflow diagram of the multi-source data acquisition module;

[0046] Figure 3 This is a flow chart of the smoke feature extraction module regarding the airflow velocity field and thermodynamic coupling response feature processing;

[0047] Figure 4 The following is the workflow diagram of the self-repair mechanism. DETAILED DESCRIPTION

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0049] See also Figures 1-4 The present invention relates to a smoke exhaust system for a forming chamber used for additive manufacturing of a rocket engine thrust chamber, and its specific implementation method is described in detail below.

[0050] The multi-source data acquisition module is responsible for acquiring multi-dimensional parameters of the flue gas in the molding chamber in real time, including temperature gradient distribution, particle concentration spectrum, gas component ratio, and airflow velocity field. The collection of these parameters provides basic data for subsequent analysis and processing.

[0051] The collected data is transmitted to the smoke feature extraction module, which performs nonlinear feature decomposition on multi-dimensional parameters to generate smoke diffusion pattern characteristics, particle aggregation trend characteristics, and thermodynamic coupling response characteristics, thereby mining key characteristic information of the smoke.

[0052] The dynamic filtration module builds a multi-layer filtration network based on an improved adaptive filtering algorithm. The extracted features are input into a dynamic weight allocation unit to generate a filter media switching strategy and an airflow path optimization matrix, thereby achieving efficient smoke filtration and rational airflow path planning.

[0053] Based on the filter media switching strategy and the airflow path optimization matrix, the collaborative control module coordinates the operating parameters of the smoke exhaust fan, filter device and cooling unit through a distributed reinforcement learning model to generate global energy consumption balance control instructions, ensuring that the system operates efficiently while achieving reasonable energy consumption distribution.

[0054] The anomaly handling module uses a density peak detection algorithm to identify abnormal fluctuations in the smoke removal process. Once an anomaly is detected, it triggers a self-repair mechanism and updates the filter network topology to ensure stable system operation.

[0055] The present invention is further described in detail below through specific examples.

[0056] Example 1:

[0057] The implementation method of the multi-source data acquisition module includes:

[0058] For the collection of temperature gradient distribution, a distributed fiber optic sensor array is used. The distributed fiber optic sensor array can perform multi-point temperature monitoring in the molding chamber. Its working principle is to use the optical time domain reflection characteristics of the optical fiber. When light propagates in the optical fiber, it will produce changes in the reflected light intensity when encountering temperature changes. By analyzing the reflected light intensity, the temperature information at different locations can be obtained. In order to understand the temperature distribution and change trend more intuitively, it is necessary to reconstruct the three-dimensional thermal field evolution map through the space-time interpolation algorithm. The space-time interpolation algorithm is based on the collected discrete temperature data points, and uses spatial interpolation and time series analysis methods to construct a map of temperature changes over time in three-dimensional space. Assume that the collected temperature data points are ,in Indicates the The temperature of the data point, Indicates the spatial coordinates of the data point in the molding chamber, Indicates the acquisition time. The spatiotemporal interpolation algorithm obtains the value of any position in the entire molding chamber by fitting and extrapolating these discrete points. At any time Temperature , thereby realizing the reconstruction of the three-dimensional thermal field evolution map.

[0059] In the measurement of particle concentration spectrum, polarization scattering measurement technology is used. Polarization scattering measurement technology uses the difference in scattering characteristics of polarized light by particles of different particle sizes to separate the distribution density of particles of different particle sizes. When polarized light is irradiated on the particles, particles of different particle sizes will produce scattered light with different polarization states and intensities. By analyzing the polarization state and intensity of the scattered light, the distribution of particles of different particle sizes can be determined. In order to improve the accuracy of the measurement, the Kalman filter algorithm is combined to correct the measurement error. The Kalman filter algorithm is a filtering method based on linear least mean square estimation, which can make the best estimate of the measurement data based on the state equation and observation equation of the system. Assuming that the true value of the particle concentration is , the measured value is , the measurement error is The Kalman filter algorithm continuously updates and predicts the measured data to obtain an estimate closer to the true value. , its core formula is:

[0060] Prediction equation:

[0061]

[0062] in is based on The estimated value of the moment The predicted value at time, is the state transition matrix, yes The best estimate of time, is the control matrix, is the control input.

[0063] Update equation:

[0064]

[0065] in yes The best estimate of time, is the Kalman gain, yes The measured value at the moment, is the observation matrix.

[0066] To obtain the proportion of gas components, the spectral absorption feature fusion method is used. Different gases have unique absorption characteristics in a specific spectral range. By measuring the absorption intensity of flue gas in multiple spectral bands and combining it with the spectral absorption feature fusion algorithm, the proportion of key components can be extracted. For example, for a certain gas component , which is in wavelength The absorption intensities at , through a specific fusion algorithm The proportion of the gas component can be obtained In order to understand the diffusion range of gas in advance, the gas diffusion boundary is predicted based on the chaos mapping model. The chaos mapping model is a nonlinear dynamic model that can describe the complex nonlinear behavior in the gas diffusion process. Assume that the initial state of gas diffusion is , through the chaos mapping function , the diffusion state at different times can be iteratively calculated , thereby predicting the gas diffusion boundary.

[0067] Example 2:

[0068] This embodiment mainly focuses on the processing of the airflow velocity field and the modeling of the thermodynamic coupling response characteristics in the smoke feature extraction module.

[0069] The vortex decomposition algorithm is used to process the airflow velocity field and extract the turbulent core region. Based on the concept of vorticity in fluid mechanics, the vortex decomposition algorithm calculates the curl of the airflow velocity field to determine the location and strength of vortices, thereby identifying the turbulent core region. By analyzing the vorticity field, the region with the largest vorticity is identified, which is the turbulent core region.

[0070] In order to further understand the movement of particles in the core area of turbulence, the phase locking technology is used to correlate the particle movement trajectory. Phase locking technology uses a certain periodic signal in the airflow as a reference to synchronize the movement of particles with the periodic signal. For example, the rotation period of the vortex is used as a reference signal, and the position change of the particles in each period is recorded to obtain the movement trajectory of the particles in the core area of turbulence. Assuming that the rotation period of the vortex is , in In a cycle, the particles are at The location is ,By continuously recording the positions of particles over multiple periods, their ,movement trajectory can be mapped.

[0071] For the thermodynamic coupling response characteristics, the tensor decomposition method is used to model the temperature-airflow interaction. The tensor decomposition method can decompose the high-dimensional temperature-airflow data into a combination of multiple low-dimensional tensors, thereby more clearly describing the interaction relationship between them. Assuming the temperature field is , the air flow velocity field is , combining them into a tensor . Through the tensor decomposition algorithm, Decompose into ,in is the rank of the decomposition, is the coefficient, 、 、 、 In space 、 、 and time Through this decomposition, a multi-scale energy transfer path characteristic matrix can be generated. This matrix reflects the energy transfer relationship between temperature and airflow at different scales, providing an important basis for subsequent analysis and processing.

[0072] Example 3:

[0073] In the dynamic filtration module, a filter media life prediction model based on a double-layer gated recurrent unit is constructed. The gated recurrent unit (GRU) is a special recurrent neural network structure that can effectively handle long-term dependencies in time series data. The double-layer gated recurrent unit further enhances the learning ability of the model. The model uses the historical operating data of the filter layer as input, including information such as filtration time, flue gas volume, and particulate matter concentration, and calculates the filter layer porosity attenuation rate in real time. Assuming the filter layer porosity is , time is , filter layer porosity decay rate It can be modeled by a double-layer gated recurrent unit Calculated, where Indicates the historical running data of the input.

[0074] In order to optimize the filter medium switching strategy, the gradient back propagation algorithm is used to optimize it. The gradient back propagation algorithm is an effective method for training neural networks. It calculates the gradient of the loss function with respect to the model parameters and updates the parameters in the opposite direction of the gradient, so that the loss function gradually decreases. In this system, the loss function is constructed with the filtration efficiency and filtration cost as the optimization objectives. Assume that the parameters related to the filter medium switching strategy are , calculated by the gradient back propagation algorithm , and update the parameters according to the gradient , thereby optimizing the filter media switching strategy.

[0075] While optimizing the filter media switching strategy, the airflow path optimization matrix is integrated to generate dynamic pressure drop compensation instructions. The purpose of dynamic pressure drop compensation instructions is to ensure that the pressure drop of the system remains within a reasonable range during the filtration process, avoiding excessive or insufficient pressure drop due to the switching of filter media or the change of airflow path. Assume that the airflow path optimization matrix is , through a specific fusion algorithm Generate dynamic voltage drop compensation instructions This instruction is used to adjust the working parameters of the filtration device, such as filtration speed, filtration area, etc., to achieve dynamic pressure drop compensation.

[0076] Example 4:

[0077] The key task of the collaborative control module is to coordinate the operating parameters of the exhaust fan, filter device and cooling unit to achieve global energy consumption balance control. First, define the exhaust fan speed , filter device pressure difference threshold and cooling unit power thresholds For the joint action space. Exhaust fan speed It directly affects the exhaust speed of the flue gas. The higher the speed, the faster the exhaust speed, but the energy consumption also increases accordingly. The pressure difference threshold of the filter device Determines the working state of the filter device. When the pressure difference exceeds this threshold, it may be necessary to switch the filter medium or adjust the operating parameters of the filter device; cooling unit power threshold Controls the cooling capacity of the cooling unit to ensure the system operates within the appropriate temperature range.

[0078] In order to find the global optimal parameter combination, a multi-objective Pareto optimization algorithm is used. The multi-objective Pareto optimization algorithm can find a set of optimal solutions among multiple conflicting objectives, and these solutions achieve a balance between different objectives. In this system, the goals include reducing energy consumption, improving filtration efficiency, and ensuring system stability. Assume that the energy consumption is , the filtration efficiency is , the system stability index is , a set of optimal parameter combinations is obtained by searching in the joint action space through the multi-objective Pareto optimization algorithm , so that under these parameters, each goal can be better taken into account.

[0079] In actual operation, since there may be conflicts in the control instructions between different devices, it is necessary to balance the conflicts of local control instructions through an asynchronous strategy update mechanism. The asynchronous strategy update mechanism allows each device to update the control strategy according to its own situation at different time points, rather than synchronously. For example, when the smoke exhaust fan detects a sudden increase in the smoke flow, it can adjust the speed immediately without waiting for the filter device and the cooling unit to adjust at the same time. During the asynchronous update process, certain coordination mechanisms, such as message passing and conflict detection algorithms, are used to ensure that the control strategy updates of each device will not conflict with each other, thereby achieving stable operation of the system. Assuming that the device At the moment Based on its own local information Update control strategy , through coordination mechanisms ,in It is the total number of devices, ensuring that the control strategy updates of each device can be coordinated and consistent to avoid conflicts.

[0080] Example 5:

[0081] This embodiment mainly introduces the specific implementation of the exception handling module and the related self-repair mechanism and the smoke toxicity index prediction model.

[0082] The abnormality handling module identifies abnormal fluctuations in the smoke removal process based on the density peak detection algorithm. The density peak detection algorithm calculates the local density of the data point and the distance to other high-density points, and identifies data points with high local density and far distance from other high-density points. These points usually represent abnormal values. Assume that the data point is , its local density The calculation formula is:

[0083]

[0084] in is a data point and The distance between is the cutoff distance, is the total number of data points. Distance to other high-density points ,in is the local density of data point j. and Analysis to determine abnormal fluctuation points.

[0085] When abnormal fluctuations are detected, a density clustering algorithm is used to classify risk levels. This algorithm divides data points into clusters based on their density distribution, with each cluster representing a risk level. Based on the risk level, a graded warning signal is generated and associated with the triggering conditions for the self-healing mechanism. For example, when the risk level reaches a high level, the self-healing mechanism is triggered.

[0086] The self-repair mechanism uses a bidirectional graph neural network to model the coupling relationship between the filter device and the smoke exhaust fan. The bidirectional graph neural network can simultaneously learn the relationship between nodes from both the forward and reverse paths. In this system, the filter device and the smoke exhaust fan are nodes of the graph, and the connection between them represents the coupling relationship between them. Fault conduction characteristics are extracted from the forward path, and redundant resource scheduling characteristics are extracted from the reverse path. Assuming the graph neural network model is ,in It is the adjacency matrix of the graph, describing the connection relationship between nodes. is the feature matrix of the node. The fault conduction characteristics are obtained through the bidirectional graph neural network and redundant resource scheduling characteristics , a repair plan is generated through attention weight fusion, and the risk assessment parameters of the dynamic filtering network are updated.

[0087] In terms of smoke toxicity index prediction, a smoke toxicity index prediction model is constructed. This model is based on historical process data and real-time gas component parameters, and uses a time series convolutional network to integrate the generation rules of thermal degradation products. Assuming that the historical process data is , the real-time gas composition parameters are , the generation law of thermal degradation products is expressed by the function Indicates that the temporal convolutional network model is , the smoke toxicity index is predicted by this model In order to more comprehensively evaluate the toxicity risk, a generative adversarial network is used to simulate the gas mixture distribution under extreme working conditions, generate a toxicity risk probability map and embed a redundant filter layer deployment strategy. and the discriminator Composition, the generator generates a simulated gas mixture distribution based on random noise , the discriminator determines whether the generated distribution is real. Through adversarial training between the two, a gas mixture distribution closer to the real extreme working conditions is generated, thereby obtaining a more accurate toxicity risk probability map and providing a more reliable basis for the redundant filter layer deployment strategy.

[0088] Through the collaborative work of the exception handling module, self-repair mechanism and smoke toxicity index prediction model, abnormal situations during system operation can be discovered and handled in a timely manner, ensuring the safe and stable operation of the system while improving the ability to respond to smoke toxicity.

[0089] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0090] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A smoke removal system for a forming chamber for additive manufacturing of a rocket engine thrust chamber, characterized in that: include: Multi-source data acquisition module: used to obtain multi-dimensional parameters of the flue gas in the molding chamber in real time, including temperature gradient distribution, particle concentration spectrum, gas component ratio and airflow velocity field; Smoke feature extraction module: performs nonlinear feature decomposition on the multi-dimensional parameters to generate smoke diffusion pattern characteristics, particle aggregation trend characteristics and thermodynamic coupling response characteristics; Dynamic filtration module: Based on the improved adaptive filtering algorithm, a multi-layer filtration network is constructed, the characteristics are input into the dynamic weight allocation unit, and the filter medium switching strategy and airflow path optimization matrix are generated; Collaborative control module: Based on the filter medium switching strategy and the airflow path optimization matrix, the distributed reinforcement learning model is used to coordinate the operating parameters of the exhaust fan, filter device, and cooling unit to generate global energy consumption balance control instructions; Abnormal processing module: Based on the density peak detection algorithm, it identifies abnormal fluctuations in the smoke removal process, triggers the self-repair mechanism and updates the filter network topology.

2. The fume removal system for the molding chamber according to claim 1 is characterized in that: The multi-source data acquisition module includes: A distributed fiber optic sensor array is used to capture the temperature gradient distribution, and a spatiotemporal interpolation algorithm is used to reconstruct the three-dimensional thermal field evolution map. The particle concentration spectrum is measured using polarization scattering measurement technology to separate the distribution density of particles of different particle sizes, and the measurement error is corrected in combination with the Kalman filter algorithm; The spectral absorption feature fusion method is used to extract the key component proportions of the gas component ratio, and the gas diffusion boundary is predicted based on the chaos mapping model.

3. The fume removal system for the molding chamber according to claim 2 is characterized in that: The smoke feature extraction module also includes: The turbulence core area is extracted from the airflow velocity field using a vortex decomposition algorithm, and the particle motion trajectories are correlated using a phase locking technique; The temperature-airflow interaction is modeled using a tensor decomposition method for the thermodynamic coupling response characteristics, and a multi-scale energy transfer path characteristic matrix is generated.

4. The fume removal system for the molding chamber according to claim 1 is characterized in that: The dynamic filtering module also includes: Construct a filter media life prediction model based on a double-layer gated circulation unit to calculate the filter layer porosity decay rate in real time; The filter medium switching strategy is optimized by the gradient back-propagation algorithm, and the airflow path optimization matrix is integrated to generate dynamic pressure drop compensation instructions.

5. The fume removal system for the molding chamber according to claim 4 is characterized in that: The collaborative control module also includes: The exhaust fan speed, filter pressure difference threshold, and cooling unit power threshold are defined as the joint action space, and a multi-objective Pareto optimization algorithm is used to solve the global optimal parameter combination. Balance local control instruction conflicts through asynchronous policy update mechanism.

6. The fume removal system for the molding chamber according to claim 1 is characterized in that: The exception handling module also includes: Using a density clustering algorithm to classify the abnormal fluctuations into risk levels, generate graded warning signals and associate them with trigger conditions for the self-repair mechanism; When the particulate matter concentration exceeds the limit, the redundant filter layer deployment strategy is activated and the priority weight of the airflow path optimization matrix is reconstructed.

7. The fume removal system for the molding chamber according to claim 6, characterized in that: The self-repair mechanism also includes: A bidirectional graph neural network is used to model the coupling relationship between the filter device and the smoke exhaust fan, extracting fault conduction characteristics from the forward path and redundant resource scheduling characteristics from the reverse path. A repair plan is generated through attention weight fusion, and the risk assessment parameters of the dynamic filtering network are updated.

8. The fume removal system for the molding chamber according to claim 6, characterized in that: The system further comprises: A flue gas toxicity index prediction model was constructed based on historical process data and real-time gas component parameters. The generation rules of thermal degradation products were integrated with a time series convolutional network to dynamically adjust the chemical adsorption priority of the filter medium.

9. The fume removal system for the molding chamber according to claim 8, characterized in that: The smoke toxicity index prediction model also includes: A generative adversarial network is used to simulate the gas mixture distribution under extreme working conditions, generate a toxicity risk probability map and embed a redundant filter layer deployment strategy.

10. The fume removal system for the molding chamber according to claim 3, characterized in that: The processing of the turbulent core region further includes: An adaptive vortex tracking algorithm is used to balance the measurement accuracy of high-speed airflow and low-speed stagnation area. The nonlinear characteristics of vortex evolution are captured through the residual convolutional network. The airflow stability evaluation matrix is constructed and integrated into the weight allocation unit of the dynamic filtering network.

Images