Integrated treatment method for industrial waste gas based on multi-layer plate

Abstract

This invention relates to the field of industrial waste gas purification and intelligent control technology, specifically a multi-layer plate-based integrated treatment method for industrial waste gas. The method includes: introducing industrial waste gas into a vertical multi-layer plate adsorption tower; deploying a distributed wireless sensor array within each adsorption plate to collect real-time operating status parameters of the plates; synchronizing and transmitting data in real-time via a time-sensitive network; constructing a dynamic model reflecting the mass transfer process of pollutants between plates using an edge computing gateway; and generating a collaborative control strategy using an improved model predictive control algorithm to regulate the sequence of electric flow guiding devices and regeneration valves. This method enables comprehensive and refined perception of the internal state of the plates, eliminates monitoring blind spots and data timing deviations, accurately characterizes the mass transfer patterns between plates, dynamically matches adsorption and regeneration conditions, improves inter-plate operational synergy, and ensures continuous and stable waste gas treatment.

Description

Technical Field

[0001] This invention relates to the field of industrial waste gas purification and intelligent control technology, and in particular to an integrated treatment method for industrial waste gas based on a multi-layer plate structure. Background Technology

[0002] Traditional industrial waste gas multi-layer plate adsorption treatment devices adopt a single-point monitoring mode at the tower inlet and outlet, only collecting parameters such as overall waste gas concentration and temperature. The sensing devices are centrally arranged at the pipeline location, and the data relies on conventional industrial networks for transmission. Adsorption and regeneration adopt a fixed time sequence control, and the operating conditions are adjusted by switching the regeneration valve and the flow guiding device through a preset cycle.

[0003] Traditional monitoring can only reflect the overall operating status of the tower, failing to acquire real-time operating parameters within individual adsorption plates. Uneven adsorption and localized saturation within the plates cannot be accurately captured. Conventional networks suffer from data latency and time-series asynchrony issues, hindering accurate data support for inter-plate state analysis and making it difficult to establish the correlation between pollutant mass transfer between plates. Furthermore, the scheduling logic for adsorption and regeneration is disconnected from the actual state of the plates, resulting in lag in the operation of electric flow guiding devices and regeneration valves, disordered inter-plate mass transfer processes, and a mismatch between adsorption utilization and regeneration control.

[0004] To address the issues of inability to precisely perceive the internal state of adsorption layers, asynchronous sensor data clocks, inaccurate modeling of interlayer mass transfer, and inability to dynamically coordinate adsorption and regeneration, it is necessary to achieve distributed state perception across the entire adsorption layer, real-time synchronous data transmission, interlayer dynamic modeling, and adaptive coordinated scheduling. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an integrated treatment method for industrial waste gas based on a multi-layer plate structure.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an integrated industrial waste gas treatment method based on a multi-layer plate structure, comprising: Industrial waste gas is introduced into a vertical multi-layer plate adsorption tower, which is equipped with multiple parallel adsorption plates. A distributed wireless sensor array is deployed inside each adsorption layer to monitor the operating status of each layer in real time. The operating status data collected by the distributed wireless sensor array is clock-synchronized and transmitted in real time through a time-sensitive network; In an edge computing gateway, the operational status data is received and used to construct an interlayer kinetic model that reflects the mass transfer process of pollutants between each adsorption layer. Based on the analysis results of the interlayer dynamics model, an improved model predictive control algorithm is used to dynamically generate a synergistic control strategy for regulating the adsorption and regeneration processes. According to the aforementioned coordinated control strategy, the electric flow guiding device and regeneration valve sequence installed in the vertical multi-layer plate adsorption tower are controlled to perform coordinated scheduling of adsorption and regeneration.

[0007] As a further aspect of the present invention, the working principle of the improved model predictive control algorithm includes: At the beginning of each control cycle, the edge computing gateway receives the current operating status data from the time-sensitive network as the current system status. The current system state is input into the interlayer dynamics model to predict the changes in pollutant adsorption amount, temperature, humidity and pressure drop of different adsorption plates over multiple future time steps, forming a predicted state sequence. Based on the predicted state sequence, under the preset constraints, an objective function is solved to obtain the optimal control action sequence. The preset constraints include valve opening limit, temperature limit and adsorbent protection conditions. The objective function aims to maximize the overall adsorption efficiency and minimize the regeneration energy consumption. From the optimal control action sequence obtained by the solution, the control command corresponding to the first time step is extracted as the actual control output of the current cycle and sent to the electric flow guiding device and the regeneration valve sequence. In the next control cycle, the process is repeated, and the internal parameters of the interlayer dynamics model are continuously optimized and corrected based on the newly acquired operating status data.

[0008] As a further aspect of the present invention, the provision of a distributed wireless sensor array within the internal space of each adsorption layer includes: The distributed wireless sensor array integrates at least an airflow distribution sensor for monitoring gas flow, a thin-film temperature sensor for monitoring heat distribution, and a conductivity humidity sensor for monitoring moisture content. A set of the airflow distribution sensor, the thin film temperature sensor, and the electrical conductivity humidity sensor are respectively arranged on the air inlet side, the central area, and the air outlet side of each adsorption plate to form a sensing unit. Connect multiple sensing units within the same adsorption layer to a single data aggregation node for that layer. The data aggregation node at this layer performs preliminary filtering, amplification, and analog-to-digital conversion on the raw monitoring signals from multiple sensing units to generate digitized layer status data packets. The data aggregation node at this layer uploads the layer status data packet to the edge computing gateway via the time-sensitive network.

[0009] As a further aspect of the present invention, the operating status data collected by the distributed wireless sensor array is clock-synchronized and transmitted in real time through a time-sensitive network, including: A global clock source is configured in the edge computing gateway, and a synchronization clock signal is periodically broadcast to the data aggregation nodes of the same layer of all adsorption plates through the time-sensitive network; Each of the data aggregation nodes at this layer calibrates its own local clock after receiving the synchronization clock signal; After collecting the layer status data packet, each data aggregation node of this layer adds a calibrated local timestamp to it and encapsulates the timestamped layer status data packet into a time-sensitive network data frame. The time-sensitive network transmits data frames from different layers to the edge computing gateway according to a preset traffic shaping strategy and priority scheduling strategy, ensuring low latency and determinism of the running status data.

[0010] As a further aspect of the present invention, the construction of an interlayer kinetic model reflecting the mass transfer process of pollutants between each adsorption layer includes: The edge computing gateway parses the received timestamped operating status data and extracts the real-time airflow velocity distribution, temperature field distribution, humidity field distribution, and pressure difference data for each layer. Based on fluid mechanics, mass transfer and adsorption equilibrium theory, a physical model framework is established that includes multilayer plate structure, adsorbent characteristics and pollutant properties. The physical model framework is incorporated as a constraint into the neural network training process using a physical information neural network. The physical information neural network is trained using historical operating data and real-time parsed data to obtain an interlayer dynamic model that simulates the changes in concentration gradient, heat transfer, and adsorption saturation process of pollutants between multilayer plates.

[0011] As a further aspect of the present invention, based on the analysis results of the interlayer dynamics model, an improved model predictive control algorithm is used to dynamically generate a synergistic control strategy for regulating the adsorption and regeneration processes, including: The edge computing gateway calls the interlayer dynamics model to calculate the adsorbent saturation rate of each adsorption layer at the current moment and predict the change of saturation rate in the future. Adsorption plates with saturation rates exceeding a high threshold are identified and marked as plates to be regenerated. Adsorption plates with saturation rates below a low threshold are identified and marked as plates with high adsorption capacity. Based on the spatial distribution of the regenerated plate and the high adsorption capacity plate, the optimal airflow guidance path is calculated so that when the regenerated plate is being regenerated, more waste gas is guided to flow through the high adsorption capacity plate. Based on the optimal airflow guidance path, a sequence of control commands is generated for the opening degree of each guide plate in the electric flow guiding device, as well as timing commands for controlling the opening and closing of the corresponding valves in the regeneration valve sequence.

[0012] As a further aspect of the present invention, according to the aforementioned coordinated control strategy, the electric flow guiding device and regeneration valve sequence installed in the vertical multi-layer plate adsorption tower are controlled to perform coordinated scheduling of adsorption and regeneration, specifically including: The edge computing gateway sends the control command sequence to the electric flow guiding device, and the electric flow guiding device adjusts the tilt angle of the corresponding flow guiding plate according to the command, thereby changing the flow field distribution of the exhaust gas in the tower; Simultaneously, the edge computing gateway sends the timing command to the regeneration valve sequence, controlling the opening of the regeneration medium valve connected to the layer to be regenerated, and introducing the regeneration medium into the layer to be regenerated for desorption and regeneration, while the regeneration medium valves corresponding to other layers remain closed; During the regeneration process, the temperature, humidity and outlet pollutant concentration changes of the regenerated plate are continuously monitored by the distributed wireless sensor array. When the regeneration is completed and the indicators meet the standards, the corresponding regeneration medium valve is closed and the regenerated plate is put back into the adsorption process.

[0013] As a further aspect of the present invention, the method further includes: during the scheduling process, when a specific indicator in the operating status data meets a preset triggering condition, initiating a low-temperature plasma activation procedure for in-situ activation of the adsorbent, including: The edge computing gateway monitors the voltage drop data of each board in real time and calculates the rate of change of its voltage drop over time. When the pressure drop change rate of any adsorption plate continuously exceeds the preset change rate threshold, it is determined that the adsorption plate is blocked or the adsorbent is severely deactivated. Simultaneously, monitor the temperature sensor data of any of the adsorption layers to confirm whether there is an abnormal temperature rise gradient; When both the pressure drop rate exceeding the limit and the abnormal temperature rise gradient are met simultaneously, the low-temperature plasma activation procedure is triggered. The low-temperature plasma activation procedure includes: cutting off the waste gas supply to any of the adsorption plates, introducing a specific gas into any of the adsorption plates, and activating the built-in plasma generator to generate low-temperature plasma to bombard and activate the adsorbent in situ, so as to restore its adsorption performance.

[0014] As a further aspect of the present invention, the edge computing gateway parses the received timestamped operating status data to extract real-time airflow velocity distribution, temperature field distribution, humidity field distribution, and pressure difference data for each layer, including: The edge computing gateway receives time-stamped layer status data packets from a time-sensitive network; The data packet representing the shelf status is decoded to separate multiple sets of sensor data from the inlet side, the central region, and the outlet side within the same adsorption shelf. The multiple sets of sensor data are spatiotemporally aligned to ensure that data from different locations on the same adsorption plate correspond to the same sampling time. Based on the data from the airflow distribution sensors on the inlet side, the central region, and the outlet side, a real-time airflow velocity distribution map on the cross-section of the adsorption plate is reconstructed using a spatial interpolation algorithm. Based on the data from the thin film temperature sensors from the inlet side, the central region and the outlet side, the temperature field distribution map inside the adsorption layer is reconstructed by the heat conduction inversion algorithm. Based on the data from the electrical conductivity humidity sensors from the air inlet side, the central region, and the air outlet side, the humidity field distribution map within the adsorption plate is reconstructed using a humidity diffusion model. Based on the pressure sensor data from the inlet and outlet sides, the real-time pressure difference data of the adsorption plate at the current moment is calculated.

[0015] As a further aspect of the present invention, the optimal airflow guiding path is calculated based on the spatial distribution of the regenerated layer and the high adsorption capacity layer, including: Based on the interlayer dynamics model, the spatial coordinates of each adsorption plate in the vertical multilayer plate adsorption tower and its current adsorbent saturation rate are obtained. The spatial coordinate set of the plate to be regenerated and the spatial coordinate set of the high adsorption capacity plate are used as inputs; An equivalent flow resistance network model containing all adsorption plates is constructed between the total inlet and total outlet of the vertical multi-layer plate adsorption tower. The optimization objective is to maximize the total flow rate of the waste gas in the high adsorption capacity plate, while the constraint is to keep the total pressure drop in the tower within a safe range. In the equivalent flow resistance network model, the flow resistance distribution scheme of the airflow channels between each layer is solved by an iterative search algorithm; Based on the flow resistance allocation scheme obtained by the solution, the target flow resistance value of each channel is mapped to the opening command of the guide plate in the electric flow guiding device at the corresponding position, and the set of opening commands defines the optimal airflow guiding path.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: A distributed wireless sensor array is deployed within the interior space of each adsorption layer. The sensing units cover the entire interior area of ​​the layer, collecting real-time operating status parameters to comprehensively reflect the adsorption conditions and medium distribution throughout the layer. This avoids the blind spots present in traditional overall monitoring, achieving refined and comprehensive perception of the internal state of each individual layer. Sensor data from each layer is collected and transmitted independently, clearly distinguishing the differences in operating states between different layers. The spatial resolution and real-time performance of the status parameters match the changing characteristics of the adsorption process within the layer, providing accurate in-situ status information for subsequent operational analysis.

[0017] Time-Sensitive Networking (TSN) synchronizes and transmits multi-node sensor data in real time, eliminating timing discrepancies between different sensor nodes, reducing data transmission latency, and ensuring the consistency and stability of status data. The edge computing gateway utilizes the synchronized, precise data to construct an interlayer pollutant mass transfer kinetic model, realistically reflecting the dynamic correlation characteristics of pollutant transfer and adsorption between multiple adsorption layers, and accurately characterizing the real-time changes in interlayer mass transfer. An improved model predictive control algorithm outputs control commands based on model analysis results, dynamically forming a coordinated control logic for adsorption and regeneration. The electric flow guide device and regeneration valve sequence act synchronously according to the commands, and the airflow distribution and regeneration sequence can adaptively adjust to the real-time status of the layers. The interlayer mass transfer process remains continuous and stable, and the adsorption and regeneration processes are mutually matched. The internal operating condition adjustment of the multi-layer plate adsorption tower maintains a high degree of coordination with the actual operating state of the layers, and the real-time performance and matching of the operation adjustment closely align with the equipment's operating characteristics. Attached Figure Description

[0018] Figure 1 This is a state diagram of the integrated industrial waste gas treatment method based on multi-layer plates according to the present invention. Figure 2 A fishbone diagram illustrating the working principle of the improved model predictive control algorithm; Figure 3 This is a flowchart illustrating the complete operation of distributed sensing and data uploading within the adsorption layer. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0021] See Figure 1 This invention provides an integrated treatment method for industrial waste gas based on a multi-layer plate structure, the overall implementation of which is as follows: Industrial waste gas is introduced into a vertical multi-layer plate adsorption tower, which has multiple parallel adsorption plates arranged along its height. A distributed wireless sensor array is deployed within the internal space of each adsorption plate to monitor its status in real time during operation. The operational status data collected by this distributed wireless sensor array is synchronized and transmitted in real time via a time-sensitive network. An edge computing gateway receives this data and constructs an interlayer dynamics model reflecting the mass transfer process of pollutants between the adsorption plates. Based on the analysis results of this model, an improved model predictive control algorithm is used to dynamically generate a coordinated control strategy for adsorption and regeneration operations. Finally, according to this coordinated control strategy, the electric flow guiding device and regeneration valve sequence installed in the adsorption tower are driven to perform actions, achieving coordinated scheduling of the adsorption and regeneration processes.

[0022] In one embodiment of the present invention, see [reference] Figure 2 At the start of each control cycle, the edge computing gateway receives the current operating status data transmitted by the time-sensitive network, which serves as the current system state. This current system state is then input into the interlayer dynamics model to predict the changes in pollutant adsorption capacity, temperature, humidity, and pressure drop of different adsorption plates over multiple future time steps, forming a predicted state sequence. Based on this predicted state sequence, and under the premise of satisfying valve opening limits, temperature limits, and adsorbent protection conditions, the objective function aimed at maximizing overall adsorption efficiency and minimizing regeneration energy consumption is solved to obtain the optimal control action sequence. The control command corresponding to the first time step is extracted from this optimal control action sequence and used as the actual control output for the current cycle, which is then sent to the electric flow guide device and the regeneration valve sequence. Upon entering the next control cycle, the above steps are repeated, and the internal parameters of the interlayer dynamics model are continuously optimized and corrected using newly acquired operating status data.

[0023] In specific implementation, the improved model predictive control algorithm involved in the integrated treatment method for industrial waste gas based on multi-layer plates is implemented as follows: At the beginning of each control cycle, the edge computing gateway receives the current operating status data of each adsorption plate sent by the time-sensitive network. This operating status data includes, but is not limited to, the airflow velocity distribution, temperature distribution, humidity distribution, pressure drop data, and pollutant adsorption amount of each plate. This operating status data is defined as the current system state. The edge computing gateway inputs the current system state into the pre-established interlayer dynamics model. Based on the current system state, the interlayer dynamics model predicts the changes in pollutant adsorption amount, temperature and humidity, and pressure drop of each adsorption plate within several discrete time steps in the future, forming a predicted state sequence. The predicted state sequence describes the overall operating trajectory of the adsorption tower over a period of time.

[0024] In some embodiments, the edge computing gateway solves an objective function based on a predicted state sequence to obtain an optimal control action sequence under preset constraints. These preset constraints include limitations on the opening degree of the electric flow guide plate, the opening range of the regeneration valve, the maximum allowable operating temperature of each adsorption layer, and the maximum allowable temperature rise rate of the adsorbent. The objective function is designed to maximize adsorption efficiency and minimize regeneration energy consumption, and its mathematical expression is: in: Represents the objective function value. Represents the length of the prediction time domain. Representing the Each prediction time step Indicates the first The overall adsorption efficiency of the adsorption tower at each time step. Indicates the first The power consumed by the regeneration process at each time step and These are the weighting coefficients for the adsorption efficiency term and the energy consumption term, respectively; the edge computing gateway uses an optimization algorithm to solve for the control action sequence that minimizes the objective function.

[0025] It is understandable that after the edge computing gateway obtains the optimal control action sequence, it extracts the control command corresponding to the first time step. This control command includes the baffle opening adjustment value and the opening and closing status of the regeneration valve that should be executed immediately. The edge computing gateway sends this control command to the electric baffle device and the regeneration valve sequence in the vertical multi-layer plate adsorption tower. The electric baffle device adjusts the baffle angle according to the opening command, and the regeneration valve sequence opens or closes the corresponding valve channel according to the command.

[0026] In practice, after one control cycle ends, the edge computing gateway enters the next control cycle. In the new control cycle, the edge computing gateway receives the latest operating status data transmitted by the time-sensitive network, updates the current system status, and repeats the entire process of prediction, optimization, and control output. At the same time, the edge computing gateway compares the newly acquired actual operating status data with the prediction output of the interlayer dynamics model, calculates the model prediction error, and uses a rolling optimization method to correct the internal parameters of the interlayer dynamics model online based on the error, so that the interlayer dynamics model can adapt to changes such as adsorbent performance decay or fluctuating exhaust gas conditions.

[0027] In one embodiment of the invention, within the internal space of each adsorption layer, see [reference needed]. Figure 3 The distributed wireless sensor array integrates at least an airflow distribution sensor for monitoring gas flow, a thin-film temperature sensor for monitoring heat distribution, and a conductivity humidity sensor for monitoring moisture content. A set of these three types of sensors is installed on the inlet side, central region, and outlet side of the adsorption layer, forming a sensing unit. Multiple sensing units within the same adsorption layer are connected to a local data aggregation node. This local data aggregation node performs preliminary filtering, amplification, and analog-to-digital conversion on the raw monitoring signals from each sensing unit, generating digitized layer status data packets, which are then uploaded to the edge computing gateway via a time-sensitive network (TSN). A global clock source is configured in the edge computing gateway, and a synchronization clock signal is periodically broadcast to all local data aggregation nodes of the adsorption layers via the TSN. Upon receiving the synchronization clock signal, each local data aggregation node calibrates its local clock and appends a calibrated local timestamp to the acquired layer status data packets, encapsulating them into TSN data frames. The TSN transmits data frames from different layers to the edge computing gateway according to a preset flow shaping and priority scheduling strategy, ensuring low latency and determinism in data transmission of operational status data.

[0028] In practical implementation, when a distributed wireless sensor array is arranged in the internal space of each adsorption layer of a vertical multi-layer plate adsorption tower, the distributed wireless sensor array integrates at least an airflow distribution sensor for monitoring gas flow, a thin-film temperature sensor for monitoring heat distribution, and an electrical conductivity humidity sensor for monitoring moisture content. A set of sensing units consisting of an airflow distribution sensor, a thin-film temperature sensor, and an electrical conductivity humidity sensor are fixedly installed on the inlet side, central area, and outlet side of each adsorption layer. Multiple sensing units in the same adsorption layer are connected to a local data aggregation node located on the side wall of the layer via wires. The local data aggregation node performs low-pass filtering, signal amplification, and analog-to-digital conversion on the raw analog monitoring signals from multiple sensing units to generate a digital layer status data packet. Then, the local data aggregation node uploads the layer status data packet to the edge computing gateway via a time-sensitive network.

[0029] In some embodiments, the edge computing gateway is configured with a high-precision global clock source. The global clock source broadcasts a synchronization clock signal at a fixed period to all local data aggregation nodes of the adsorbed layers via a time-sensitive network. Each local data aggregation node calibrates its local clock counter after receiving the synchronization clock signal. After collecting and generating layer status data packets, each local data aggregation node adds a calibrated local timestamp to the layer status data packets and encapsulates the timestamped layer status data packets into data frames conforming to the time-sensitive network protocol. The time-sensitive network processes data frames from different layers according to a preset traffic shaping strategy and a priority scheduling strategy. The traffic shaping strategy limits the maximum bandwidth of each layer's data, and the priority scheduling strategy assigns a higher transmission level to data from higher-level adsorbed layers, thereby ensuring low latency and determinism in the transmission of operational status data from the local data aggregation node to the edge computing gateway.

[0030] It is understandable that, in order to quantitatively evaluate the quality of data synchronization, the edge computing gateway statistically analyzes the timestamp deviation of data frames arriving at each layer. The clock synchronization accuracy of the time-sensitive network can be characterized by the following formula: in: Indicates the maximum clock synchronization error. This indicates that the edge computing gateway received the first... The moment of each data frame Indicates the first The moment when a data frame is sent by the data aggregation node of this layer. Indicates the nominal value of fixed transmission delay; The smaller the value, the higher the clock synchronization accuracy provided by the time-sensitive network.

[0031] In specific implementation, the layer status data packet contains processed digital sensor readings. The preliminary processing of the sensor unit signal by the data aggregation node of this layer includes removing power frequency interference and random noise. Optionally, when transmission resources are tight, the data aggregation node of this layer performs lossless compression on the layer status data packet before encapsulating and uploading it. Optionally, when the time-sensitive network detects a sudden increase in the data flow of a certain layer, it automatically enables redundant path diversion transmission to avoid single path congestion.

[0032] In one embodiment of the present invention, the edge computing gateway parses the time-stamped layer status data packets received from the time-sensitive network, separates multiple sets of sensor data from the inlet side, central region, and outlet side within the same adsorption layer, and performs spatiotemporal alignment on these data to ensure that data from different locations within the same layer correspond to the same sampling time; based on the airflow distribution sensor data from the inlet side, central region, and outlet side, a real-time airflow velocity distribution map of the adsorption layer cross-section is reconstructed using a spatial interpolation algorithm; based on the thin-film temperature sensor data from the aforementioned locations, a temperature field distribution map within the adsorption layer is reconstructed using a thermal conduction inversion algorithm; based on the conductivity humidity sensor data from the aforementioned locations, a humidity field distribution map within the adsorption layer is reconstructed using a humidity diffusion model; and simultaneously, the real-time pressure difference data of the adsorption layer at the current moment is calculated based on the pressure sensor data from the inlet side and outlet side. Based on this, a physical model framework including multilayer plate structure, adsorbent characteristics and pollutant properties was established according to fluid mechanics, mass transfer and adsorption equilibrium theory. This physical model framework was embedded as a constraint in the training process of the physical information neural network. The physical information neural network was trained using historical operating data and real-time analyzed operating status data, and finally an interlayer dynamic model that can simulate the concentration gradient change, heat transfer and adsorption saturation process of pollutants between multilayer plates was obtained.

[0033] In practical implementation, the edge computing gateway receives time-stamped shelf status data packets from a time-sensitive network, decodes these packets to separate multiple sets of sensor data from the inlet side, central region, and outlet side within the same adsorption shelf. The edge computing gateway then performs spatiotemporal alignment processing on these sensor data packets, correcting minor time shifts caused by differences in transmission paths to ensure that data from different spatial locations within the same adsorption shelf strictly correspond to the same sampling time. Based on the measurement values ​​from the airflow distribution sensors on the inlet side, central region, and outlet side, the edge computing gateway reconstructs the real-time data across the entire cross-section of the adsorption shelf using a bilinear spatial interpolation algorithm. The edge computing gateway reconstructs the temperature field distribution map inside the adsorption layer using a two-dimensional thermal conduction inversion algorithm, based on readings from thin-film temperature sensors on the inlet side, central region, and outlet side. It also reconstructs the humidity field distribution map inside the adsorption layer using a humidity diffusion model derived from Fick's diffusion law, based on signals from conductivity and humidity sensors on the inlet side, central region, and outlet side. Simultaneously, the edge computing gateway reads data from pressure sensors on the inlet and outlet sides, calculates the difference between the two to obtain the real-time pressure difference data of the adsorption layer at the current moment. The reconstructed distribution map and pressure difference data constitute a complete spatial description of the layer's operating state.

[0034] In some embodiments, when constructing an interlayer dynamics model reflecting the mass transfer process of pollutants between adsorption layers, the edge computing gateway establishes a physical model framework based on the Navier-Stokes equations of fluid mechanics, the two-film theory of mass transfer, and the Langmuir isotherm relationship of adsorption equilibrium theory. This framework includes the geometric parameters of the multilayer plates, the specific surface area and pore volume characteristics of the adsorbent, and the molecular weight and diffusion coefficient of the pollutants. The continuity equation and momentum conservation equation of the physical model framework describe the flow behavior of the exhaust gas between the multilayer plates, and the mass transfer equation characterizes the transfer process of pollutants from the gas phase to the solid phase of the adsorbent. The edge computing gateway utilizes a physical information neural network to directly integrate the partial differential equations in the above physical model framework as soft constraints into the loss function design of the neural network, forcing the learning process of the neural network to comply with known physical laws.

[0035] It is understandable that the training process of the physical information neural network uses both historical operating datasets and real-time parsed operating status data. The historical operating datasets cover the state records of the layers under different exhaust gas flow rates, different inlet concentrations, and different temperature and humidity conditions. During training, the input of the physical information neural network is the airflow velocity distribution, temperature field distribution, humidity field distribution, and pressure difference data of each layer, and the output is the pollutant concentration distribution and adsorbent loading of each layer at the next moment. The physical information neural network minimizes the mean square error between the predicted output and the actual measurement value and the residual norm of the physical equation through the backpropagation algorithm. After sufficient training, the interlayer dynamic model can accurately simulate the diffusion mass transfer process of pollutants driven by concentration gradients between multiple layers, the heat transfer process accompanied by adsorption exothermia, and the dynamic process of adsorbent saturation.

[0036] In practical implementation, the structural configuration parameters of the physical information neural network are shown in Table 1: Table 1: Main Structural Parameters of Physical Information Neural Network Among them, the physical residual weighting coefficient To balance the contributions of data fitting error and physical equation constraint residuals to the total loss function, the expression for the total loss function is: in: This represents the total number of training samples. Indicates the first The true state vector of each sample The physical information neural network represents the first... The predicted state vector of each sample. This represents the partial differential operators corresponding to the physical model framework. The parameters to be optimized in the physical information neural network are represented by this loss function, which ensures that the trained interlayer dynamics model both matches the observed data and follows the physical conservation laws.

[0037] In one embodiment of the present invention, the edge computing gateway invokes the interlayer dynamics model to calculate the adsorbent saturation rate of each adsorption layer at the current moment and predict its changing trend in the future period; adsorption layers with saturation rates higher than a set high threshold are identified as layers to be regenerated, while adsorption layers with saturation rates lower than a set low threshold are identified as layers with high adsorption capacity. Based on the interlayer dynamics model, the spatial coordinates of each adsorption layer in the adsorption tower and its current adsorbent saturation rate are obtained. Using the set of spatial coordinates of the regenerating layer and the high adsorption capacity layer as input, an equivalent flow resistance network model including all adsorption layers is constructed between the total inlet and outlet of the exhaust gas. With maximizing the total flow rate of exhaust gas in the high adsorption capacity layer as the optimization objective and maintaining the total pressure drop in the tower within a safe range as the constraint, the flow resistance distribution scheme of the airflow channel between each layer is solved by an iterative search algorithm in the equivalent flow resistance network model. The obtained flow resistance distribution scheme is mapped to the opening command of the guide plate in the corresponding electric flow guiding device. This set of opening commands defines the optimal airflow guiding path, thereby generating a control command sequence for the opening of the guide plate of the electric flow guiding device and an opening and closing timing command for the regeneration valve sequence.

[0038] In practical implementation, the edge computing gateway invokes the interlayer dynamics model to calculate the adsorbent saturation rate of each adsorption plate in the vertical multilayer plate adsorption tower at the current moment, and extrapolates the trend of adsorbent saturation rate changes of each plate over a future period based on the current operating conditions. The edge computing gateway compares the calculated adsorbent saturation rate with preset high and low thresholds, identifies adsorption plates with adsorbent saturation rates greater than or equal to the high threshold and marks them as plates to be regenerated, and identifies adsorption plates with adsorbent saturation rates less than or equal to the low threshold and marks them as high adsorption capacity plates. The marking information of the plates to be regenerated and the high adsorption capacity plates, together with their spatial coordinates, constitute the decision basis for collaborative control.

[0039] In some embodiments, the edge computing gateway obtains the three-dimensional spatial coordinates of all adsorption plates in the vertical multi-layer plate adsorption tower and their current adsorbent saturation values ​​based on the interlayer dynamics model, using the set of spatial coordinates of the marked plates to be regenerated and the set of spatial coordinates of the high adsorption capacity plates as key inputs. Between the total inlet node and the total outlet node of the exhaust gas in the vertical multi-layer plate adsorption tower, the edge computing gateway constructs an abstract equivalent flow resistance network model. The equivalent flow resistance network model regards each adsorption plate as a flow resistance node and the airflow channel between adjacent adsorption plates as a connecting branch with variable flow resistance. The topology of the equivalent flow resistance network model reflects the relative positions and connectivity of each plate in the actual physical structure of the adsorption tower.

[0040] It is understandable that the edge computing gateway sets the optimization objective to maximize the total flow volume of waste gas in the high adsorption capacity plate nodes, with the hard constraint of maintaining the overall pressure drop of the vertical multi-layer plate adsorption tower within the safe operating range of the equipment. In the equivalent flow resistance network model, the edge computing gateway uses a gradient descent-type iterative search algorithm to solve the flow resistance distribution scheme of each connecting branch. The flow resistance distribution scheme determines the distribution ratio of waste gas among each plate. After the solution is completed, the edge computing gateway converts the calculated target flow resistance value into the mechanical opening command of the guide plate in the corresponding electric flow guiding device through a linear mapping relationship. This set of opening commands defines the optimal airflow guiding path at the physical level. Based on the optimal airflow guiding path, the edge computing gateway generates a sequence of control commands to adjust the angle of each guide plate in the electric flow guiding device and a timing pulse command to control the opening and closing state of each valve in the regeneration valve sequence.

[0041] For the optimization solution parameter configuration of the equivalent flow resistance network model in specific implementation, please refer to Table 2: Table 2: Optimization Parameters for Equivalent Flow Resistance Network Model The mathematical expression for the optimization objective of the equivalent flow resistance network model is: in: This indicates optimizing the objective function value. This represents the set of indices for all high adsorption capacity plate layers. Indicates the flow through the first The volumetric flow rate of a high-adsorption-capacity plate; the iterative search process continuously adjusts the branch flow resistance until... Achieve maximum value and total pressure drop of tower body not exceeding .

[0042] In one embodiment of the present invention, the edge computing gateway sends the control command sequence generated in the embodiment to the electric flow guiding device, causing it to adjust the tilt angle of the corresponding flow guiding plate and change the flow distribution of exhaust gas in the tower; at the same time, it sends the timing command of the regeneration valve sequence to open the regeneration medium valve connected to the plate to be regenerated, and introduce the regeneration medium to it for desorption and regeneration, while the regeneration medium valves corresponding to the other plates remain closed; during the regeneration process, the temperature, humidity and outlet pollutant concentration of the plate to be regenerated are continuously monitored by a distributed wireless sensor array. When the regeneration completion index reaches the set standard, the corresponding regeneration medium valve is closed and the plate is reintroduced into the adsorption process. Meanwhile, the edge computing gateway monitors the pressure drop data of each plate and its rate of change over time in real time; if the pressure drop rate of a certain adsorption plate is found to continuously exceed the preset threshold and the temperature sensor data of the plate shows an abnormal temperature rise gradient, the low-temperature plasma activation program is triggered; the exhaust gas supply to the layer is cut off, a specific gas is introduced into it and the built-in plasma generator is activated to generate low-temperature plasma to bombard and activate the adsorbent in situ.

[0043] In practical implementation, the edge computing gateway transforms the generated collaborative control strategy into execution instructions. The control instruction sequence for the electric flow guiding device includes the target tilt angle values ​​for each flow guiding plate. The servo motor drive mechanism in the electric flow guiding device adjusts the mechanical tilt angle of the corresponding flow guiding plate according to the received control instruction sequence, changing the flow direction and velocity distribution of the exhaust gas inside the vertical multi-layer plate adsorption tower, so that the exhaust gas preferentially flows through the area marked as a high adsorption capacity plate. At the same time, the edge computing gateway sends timing instructions to the regeneration valve sequence. The timing instructions include the opening time and opening duration of the regeneration medium valve corresponding to the plate to be regenerated, as well as the closing and locking commands for the valves corresponding to other non-regeneration plates. The solenoid valves or pneumatic valves in the regeneration valve sequence open the pipelines connected to the plate to be regenerated according to the timing instructions, injecting high-temperature steam or hot air into the plate to be regenerated as the regeneration medium, triggering the desorption reaction of pollutants on the adsorbent. At this time, the regeneration medium valves of other plates remain in a normally closed state to maintain normal adsorption function.

[0044] In some embodiments, during the regeneration operation, a distributed wireless sensor array continuously collects the readings of the film temperature sensor and the conductivity humidity sensor of the plate to be regenerated. At the same time, a miniature gas detection module located on the gas outlet side of the plate to be regenerated monitors the fluctuation of pollutant concentration at the outlet. The edge computing gateway receives the above monitoring data in real time and calculates the regeneration process indicators. When it is detected that the internal temperature rise curve of the plate to be regenerated tends to be stable, the humidity change rate drops to the set lower limit, and the outlet pollutant concentration is below the threshold for several consecutive sampling cycles, it is determined that the regeneration completion indicators have been met. The edge computing gateway then issues an instruction to close the corresponding regeneration medium valve to stop the flow of regeneration medium and reconnects the exhaust gas passage of the plate to be regenerated, thus reintegrating the plate into the adsorption operation process.

[0045] It is understandable that the edge computing gateway synchronously executes status monitoring tasks during the scheduling process, reads the pressure sensor data on the inlet and outlet sides of each adsorption layer in real time, calculates the instantaneous pressure difference value of each adsorption layer, and obtains the derivative of the pressure difference with time based on time series differential operation as the pressure drop change rate. When the pressure drop change rate of a certain adsorption layer continuously exceeds the preset change rate threshold, the edge computing gateway determines that the adsorption layer has adsorbent pore blockage or severe adsorbent deactivation. At the same time, the edge computing gateway retrieves the historical data of the thin film temperature sensor of the adsorption layer to analyze whether there is a local abnormal temperature rise gradient that does not conform to the normal adsorption exothermic law. If the pressure drop change rate exceeding the limit event and the abnormal temperature rise gradient event are both true, the edge computing gateway immediately triggers the low-temperature plasma activation program for the adsorption layer.

[0046] In specific implementation, the execution steps of the low-temperature plasma activation procedure include: the edge computing gateway first cuts off the exhaust gas inlet valve leading to the target adsorption plate, stopping the exhaust gas supply; then, it controls the opening of the dedicated gas pipeline valve to fill the target adsorption plate with nitrogen or inert gas as the discharge atmosphere gas; after the atmosphere replacement is completed, the plasma generator embedded in the inner wall of the adsorption plate is activated, and a high-frequency, high-voltage electric field is applied to ionize the gas and generate low-temperature plasma; the low-temperature plasma bombards the surface of the adsorbent in situ, removing carbon deposits and polymer deposits, and restoring the adsorbent activity; the activation duration is determined by a preset energy dose formula. in: Indicates the plasma processing energy dose. Indicates the total running time of the activation procedure. Indicates the plasma generator at time... ; when the cumulative dose Once the set standard is reached, the plasma generator stops, and the plate is restored to normal operation after being purged with air.

[0047] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A multi-layer plate-based integrated treatment method for industrial waste gas, characterized in that, include: Industrial waste gas is introduced into a vertical multi-layer plate adsorption tower, which is equipped with multiple parallel adsorption plates. A distributed wireless sensor array is deployed inside each adsorption layer to monitor the operating status of each layer in real time. The operating status data collected by the distributed wireless sensor array is clock-synchronized and transmitted in real time through a time-sensitive network; In an edge computing gateway, the operational status data is received and used to construct an interlayer kinetic model that reflects the mass transfer process of pollutants between each adsorption layer. Based on the analysis results of the interlayer dynamics model, an improved model predictive control algorithm is used to dynamically generate a synergistic control strategy for regulating the adsorption and regeneration processes. According to the aforementioned coordinated control strategy, the electric flow guiding device and regeneration valve sequence installed in the vertical multi-layer plate adsorption tower are controlled to perform coordinated scheduling of adsorption and regeneration.

2. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 1, characterized in that, The working principle of the improved model predictive control algorithm includes: At the beginning of each control cycle, the edge computing gateway receives the current operating status data from the time-sensitive network as the current system status. The current system state is input into the interlayer dynamics model to predict the changes in pollutant adsorption amount, temperature, humidity and pressure drop of different adsorption plates over multiple future time steps, forming a predicted state sequence. Based on the predicted state sequence, under the preset constraints, an objective function is solved to obtain the optimal control action sequence. The preset constraints include valve opening limit, temperature limit and adsorbent protection conditions. The objective function aims to maximize the overall adsorption efficiency and minimize the regeneration energy consumption. From the optimal control action sequence obtained by the solution, the control command corresponding to the first time step is extracted as the actual control output of the current cycle and sent to the electric flow guiding device and the regeneration valve sequence. In the next control cycle, the process is repeated, and the internal parameters of the interlayer dynamics model are continuously optimized and corrected based on the newly acquired operating status data.

3. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 1, characterized in that, The aforementioned arrangement of a distributed wireless sensor array within the internal space of each adsorption layer includes: The distributed wireless sensor array integrates at least an airflow distribution sensor for monitoring gas flow, a thin-film temperature sensor for monitoring heat distribution, and a conductivity humidity sensor for monitoring moisture content. A set of the airflow distribution sensor, the thin film temperature sensor, and the electrical conductivity humidity sensor are respectively arranged on the air inlet side, the central area, and the air outlet side of each adsorption plate to form a sensing unit. Connect multiple sensing units within the same adsorption layer to a single data aggregation node for that layer. The data aggregation node at this layer performs preliminary filtering, amplification, and analog-to-digital conversion on the raw monitoring signals from multiple sensing units to generate digitized layer status data packets. The data aggregation node at this layer uploads the layer status data packet to the edge computing gateway via the time-sensitive network.

4. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 3, characterized in that, The operational status data collected by the distributed wireless sensor array is clock-synchronized and transmitted in real time through a time-sensitive network, including: A global clock source is configured in the edge computing gateway, and a synchronization clock signal is periodically broadcast to the data aggregation nodes of the same layer of all adsorption plates through the time-sensitive network; Each of the data aggregation nodes at this layer calibrates its own local clock after receiving the synchronization clock signal; After collecting the layer status data packet, each data aggregation node of this layer adds a calibrated local timestamp to it and encapsulates the timestamped layer status data packet into a time-sensitive network data frame. The time-sensitive network transmits data frames from different layers to the edge computing gateway according to a preset traffic shaping strategy and priority scheduling strategy, ensuring low latency and determinism of the running status data.

5. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 1, characterized in that, The construction of an interlayer kinetic model reflecting the mass transfer process of pollutants between adsorption layers includes: The edge computing gateway parses the received timestamped operating status data and extracts the real-time airflow velocity distribution, temperature field distribution, humidity field distribution, and pressure difference data for each layer. Based on fluid mechanics, mass transfer and adsorption equilibrium theory, a physical model framework is established that includes multilayer plate structure, adsorbent characteristics and pollutant properties. The physical model framework is incorporated as a constraint into the neural network training process using a physical information neural network. The physical information neural network is trained using historical operating data and real-time parsed data to obtain an interlayer dynamic model that simulates the changes in concentration gradient, heat transfer, and adsorption saturation process of pollutants between multilayer plates.

6. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 1, characterized in that, Based on the analysis results of the interlayer dynamics model, an improved model predictive control algorithm is used to dynamically generate a synergistic control strategy for regulating the adsorption and regeneration processes, including: The edge computing gateway calls the interlayer dynamics model to calculate the adsorbent saturation rate of each adsorption layer at the current moment and predict the change of saturation rate in the future. Adsorption plates with saturation rates exceeding a high threshold are identified and marked as plates to be regenerated. Adsorption plates with saturation rates below a low threshold are identified and marked as plates with high adsorption capacity. Based on the spatial distribution of the regenerated plate and the high adsorption capacity plate, the optimal airflow guidance path is calculated so that when the regenerated plate is being regenerated, more waste gas is guided to flow through the high adsorption capacity plate. Based on the optimal airflow guidance path, a sequence of control commands is generated for the opening degree of each guide plate in the electric flow guiding device, as well as timing commands for controlling the opening and closing of the corresponding valves in the regeneration valve sequence.

7. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 6, characterized in that, According to the aforementioned coordinated control strategy, the electric flow guiding device and regeneration valve sequence installed in the vertical multi-layer plate adsorption tower are controlled to perform coordinated scheduling of adsorption and regeneration, specifically including: The edge computing gateway sends the control command sequence to the electric flow guiding device, and the electric flow guiding device adjusts the tilt angle of the corresponding flow guiding plate according to the command, thereby changing the flow field distribution of the exhaust gas in the tower; Simultaneously, the edge computing gateway sends the timing command to the regeneration valve sequence, controlling the opening of the regeneration medium valve connected to the layer to be regenerated, and introducing the regeneration medium into the layer to be regenerated for desorption and regeneration, while the regeneration medium valves corresponding to other layers remain closed; During the regeneration process, the temperature, humidity and outlet pollutant concentration changes of the regenerated plate are continuously monitored by the distributed wireless sensor array. When the regeneration is completed and the indicators meet the standards, the corresponding regeneration medium valve is closed and the regenerated plate is put back into the adsorption process.

8. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 1, characterized in that, The method further includes: during the scheduling process, when a specific indicator in the operating status data meets a preset triggering condition, initiating a low-temperature plasma activation procedure for in-situ activation of the adsorbent, including: The edge computing gateway monitors the voltage drop data of each board in real time and calculates the rate of change of its voltage drop over time. When the pressure drop change rate of any adsorption plate continuously exceeds the preset change rate threshold, it is determined that the adsorption plate is blocked or the adsorbent is severely deactivated. Simultaneously, monitor the temperature sensor data of any of the adsorption layers to confirm whether there is an abnormal temperature rise gradient; When both the pressure drop rate exceeding the limit and the abnormal temperature rise gradient are met simultaneously, the low-temperature plasma activation procedure is triggered. The low-temperature plasma activation procedure includes: cutting off the waste gas supply to any of the adsorption plates, introducing a specific gas into any of the adsorption plates, and activating the built-in plasma generator to generate low-temperature plasma to bombard and activate the adsorbent in situ, so as to restore its adsorption performance.

9. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 5, characterized in that, The edge computing gateway parses the received timestamped operational status data and extracts the real-time airflow velocity distribution, temperature field distribution, humidity field distribution, and pressure difference data for each layer, including: The edge computing gateway receives time-stamped layer status data packets from a time-sensitive network; The data packet representing the shelf status is decoded to separate multiple sets of sensor data from the inlet side, the central region, and the outlet side within the same adsorption shelf. The multiple sets of sensor data are spatiotemporally aligned to ensure that data from different locations on the same adsorption plate correspond to the same sampling time. Based on the data from the airflow distribution sensors on the inlet side, the central region, and the outlet side, a real-time airflow velocity distribution map on the cross-section of the adsorption plate is reconstructed using a spatial interpolation algorithm. Based on the data from the thin film temperature sensors from the inlet side, the central region and the outlet side, the temperature field distribution map inside the adsorption layer is reconstructed by the heat conduction inversion algorithm. Based on the data from the electrical conductivity humidity sensors from the air inlet side, the central region, and the air outlet side, the humidity field distribution map within the adsorption plate is reconstructed using a humidity diffusion model. Based on the pressure sensor data from the inlet and outlet sides, the real-time pressure difference data of the adsorption plate at the current moment is calculated.

10. The integrated industrial waste gas treatment method based on multi-layer plate as described in claim 6, characterized in that, Based on the spatial distribution of the regenerated layer and the high adsorption capacity layer, the optimal airflow guiding path is calculated, including: Based on the interlayer dynamics model, the spatial coordinates of each adsorption plate in the vertical multilayer plate adsorption tower and its current adsorbent saturation rate are obtained. The spatial coordinate set of the plate to be regenerated and the spatial coordinate set of the high adsorption capacity plate are used as inputs; An equivalent flow resistance network model containing all adsorption plates is constructed between the total inlet and total outlet of the vertical multi-layer plate adsorption tower. The optimization objective is to maximize the total flow rate of the waste gas in the high adsorption capacity plate, while the constraint is to keep the total pressure drop in the tower within a safe range. In the equivalent flow resistance network model, the flow resistance distribution scheme of the airflow channels between each layer is solved by an iterative search algorithm; Based on the flow resistance allocation scheme obtained by the solution, the target flow resistance value of each channel is mapped to the opening command of the guide plate in the electric flow guiding device at the corresponding position, and the set of opening commands defines the optimal airflow guiding path.

Images