Kitchen smoke exhaust space vector control and airflow organization collaborative optimization management method and system

Abstract

The invention discloses a kitchen smoke exhaust space vector control and airflow organization collaborative optimization management method and system, and relates to the technical field of data processing. According to the method, kitchen space topology is constructed, multiple sensors are deployed to collect oil smoke concentration, temperature, wind speed and equipment operation data, and a structured environment tensor is generated through interpolation; solving a high-precision oil smoke motion vector field, generating a dynamic diffusion path, positioning an oil smoke core area, dynamically marking a personnel risk area, and synthesizing a thermal plume main path; an LSTM model is adopted to predict equipment load and fault probability, and a dynamic power consumption strategy is generated; constructing a multi-objective function of clearance rate, energy consumption and comfort, outputting a Pareto optimal solution set through a genetic algorithm, and dynamically selecting a control scheme according to working conditions; and finally, an instruction is converted into coordinated action of air exhaust and air supplement, energy consumption is reduced through PID real-time parameter adjustment and closed-loop correction, and meanwhile the effects of improving the oil smoke removal efficiency and accurately controlling the air speed of a personnel area within a comfortable range are achieved.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and more specifically, relates to a method and system for collaborative optimization management of kitchen smoke exhaust space vector control and airflow organization. Background Art

[0002] With the expansion of the catering industry and stricter environmental regulations, commercial kitchens face higher demands for air quality and energy efficiency management. Traditional ventilation systems suffer from the common pain points of high energy consumption and low accuracy. Exhaust air volume is often estimated based on the number of stoves and a fixed air volume, resulting in high energy waste during non-cooking periods. Furthermore, the diffusion of oil smoke is affected by factors such as the dynamic path of the heat plume and interference from human activity. Static ventilation solutions struggle to effectively capture the core areas of oil smoke, leading to localized oil smoke exceeding standards and occupational health risks. The integration of environmental sensing, fluid simulation, and intelligent decision-making technologies is crucial to achieve refined airflow organization and control.

[0003] In the existing technology, traditional fan speed regulation only relies on simple temperature sensing or timing strategies, and cannot dynamically respond to the spatial diffusion characteristics of oil smoke; local exhaust causes the air flow field to be disordered, the oil smoke retention rate is high, and the air flow organization is unbalanced; the kitchen exhaust system lacks an energy efficiency collaborative optimization mechanism, resulting in serious energy waste. Summary of the Invention

[0004] (1) Technical problems solved In response to the problems in the related art, the present invention provides a method for collaborative optimization management of kitchen smoke exhaust space vector control and airflow organization to overcome the above-mentioned technical problems existing in the existing related art.

[0005] (2) Technical solution To solve the above technical problems, the present invention is achieved through the following technical solutions: S1. Deploy multiple sensors based on kitchen space topology data; collect real-time environmental data sets and kitchen real-time operation data through multiple sensors; S2. Based on the real-time environmental data set, structured field data is obtained by using the central difference method and the Sobel operator. Based on the structured field data, the final oil fume motion vector field is obtained by solving the dynamic equation. Based on the final oil fume motion vector field, the oil fume features are extracted and dynamic trajectory modeling is performed to obtain a dynamic vector diagram. S3. Based on the real-time environmental data set, perform oil fume concentration field threshold filtering and cluster analysis to obtain the oil fume core area set; Through exposure risk assessment, the dynamic personnel zone is obtained; based on the dynamic vector diagram, the thermal plume characteristics are quantified to obtain the main path of the thermal plume; S4. Input the real-time kitchen operation data into the LSTM model trained with historical kitchen operation data to perform load prediction and failure probability prediction to obtain a real-time dynamic power consumption strategy; S5. Construct a multi-objective function based on a real-time dynamic power consumption strategy; obtain a Pareto optimal solution set using a multi-objective genetic algorithm based on the oil smoke core area set, the dynamic personnel area, the main path of the heat plume, and the multi-objective function; and select the optimal solution from the Pareto optimal solution set based on the real-time operating conditions. S6. Convert the optimal solution into execution instructions, and perform coordinated optimization of the airflow organization of exhaust control and supply air control to obtain the final control instruction set; S7. Performing deviation analysis on the execution effect data of the final control instruction set to obtain a control parameter correction; and executing control based on the control parameter correction; The present invention uses multiple sensors to collect kitchen environment data in real time, combines dynamic equations to construct an oil fume motion vector field, and accurately identifies the oil fume core area, personnel activity area and main path of the heat plume; based on the LSTM prediction dynamic power consumption strategy, a multi-objective genetic algorithm is used to optimize the exhaust / air supply scheme, generate collaborative control instructions, and dynamically correct parameters through closed-loop feedback, thereby improving the oil fume removal rate, reducing energy consumption and dynamic power consumption strategy response delay; closed-loop optimization ensures the stability of complex working conditions and reduces wind speed fluctuations in the personnel area; while achieving the goal of reducing energy consumption, improving the oil fume removal efficiency and accurately controlling the wind speed in the personnel area within a comfortable range.

[0006] Preferably, the S1 comprises the following steps: S11. By importing the BIM model or laser scanning the point cloud, collect the precise coordinates of the stove, the boundary coordinates of the personnel activity area, the location and size of the exhaust vents and air supply vents, and the coordinates of obstacles to obtain spatial topology data; Establish a three-dimensional coordinate system, mark the stove area, personnel activity area, and ventilation vent locations in the spatial topology data, and obtain the kitchen space coordinate mapping table; S12. Deploy multiple sensors based on the kitchen space coordinate mapping table to obtain a sensor node location table; the multiple sensors include a laser scattering fume sensor, a three-dimensional ultrasonic anemometer, an infrared temperature array, and a PM2.5 sensor; The NTP protocol is used to align the timestamps of all sensors in the sensor node location table, and coordinate tags are added to each sampling point. Real-time data on oil smoke concentration, wind speed, temperature, and PM2.5 are collected to obtain the original sensor data stream. Energy consumption data, operating hours, and equipment status data of the kitchen are collected to obtain real-time kitchen operation data. S13, reorganizing the original sensor data stream by time slice, performing Kriging spatial interpolation on the missing points in the original sensor data, and constructing a structured tensor to obtain a real-time environmental dataset including the oil smoke concentration field, temperature field, and wind speed vector field; The present invention constructs the millimeter-level spatial topology of the kitchen through BIM or laser scanning, deploys a multi-type sensor network based on a coordinate mapping table, and uses NTP timestamp synchronization and Kriging spatial interpolation to generate a structured environmental tensor; it achieves dynamic perception of oil smoke in the entire area, eliminates monitoring blind spots, and improves data integrity, laying a spatial data foundation for precise wind control.

[0007] Preferably, said S2 comprises the following steps: S21, unify all sensor timestamps and interpolate the discrete point data in the real-time environmental data set into the grid to obtain processed environmental data; Based on the processed environmental data, the concentration gradient field, temperature gradient field and wind speed gradient field are calculated by the central difference method and the Sobel operator respectively, and the structured field data are obtained; S22, discretizing the continuous equations in the structured field data into differential equations on grid points to obtain a discretized set of equations; Calculate the concentration gradient, wind speed and temperature cross product terms of each equation set in the discretized equation set, and calculate the concentration time derivative to obtain the oil smoke motion vector field at the current moment; Adding the κ-ε model correction term to the oil fume motion vector field at the current moment to obtain the corrected oil fume motion vector field; The corrected oil fume motion vector field is set with the wall boundary and the vent boundary to obtain the final oil fume motion vector field; S23. Based on the final oil smoke motion vector field, perform source intensity calculation, source location, and uncertainty analysis to obtain oil smoke source characteristics; Based on the final oil smoke motion vector field and oil smoke source characteristics, particle tracking, thermal plume identification and diffusion direction analysis are performed on the oil smoke source to obtain a dynamic vector diagram; The present invention calculates the concentration, temperature, and wind speed gradient fields through the central difference method and the Sobel operator, and combines the κ-ε turbulence model to correct the boundary conditions to accurately solve the oil smoke motion vector field. It generates a dynamic vector diagram based on particle tracking and thermal plume identification to achieve accurate prediction of the oil smoke diffusion path. Compared with traditional models, the accuracy is improved, the airflow dead zone is eliminated, and the core modeling foundation is provided for precise wind control.

[0008] Preferably, the step S3 includes the following steps: S31, performing concentration unit normalization processing on the oil fume concentration field in the real-time environmental data set and matching the timestamp of the final oil fume motion vector field to obtain a standardized oil fume concentration field; Create a high oil smoke mask; mark the standardized oil smoke concentration field based on the high oil smoke mask to obtain a marked concentration field; S32, setting the neighborhood radius and the minimum number of cluster points of the DBSCAN algorithm to obtain the parameters of the DBSCAN algorithm; Cluster analysis of the labeled concentration field is performed based on the parameters of the DBSCAN algorithm to obtain spatially adjacent high-concentration point groups; Remove noise points from spatially adjacent high-concentration point clusters, and calculate the centroid position, equivalent radius, and peak concentration of each cluster to obtain the oil smoke core area set; S33. Based on the kitchen space topology and the dynamic tracking of thermal imaging by the infrared positioning system, the actual activity area of ​​the personnel is obtained; the final oil smoke motion vector field is superimposed on the actual activity area of ​​the personnel, and the exposure index is calculated to obtain the real-time dynamic personnel area; S34, extracting the stove heat source from the dynamic vector diagram and performing coefficient calibration to obtain the stove heat power; calculating the plume intensity in the dynamic vector diagram based on the stove heat power to obtain a plume intensity set; A plume intensity threshold is set, and the plume intensities with plume intensity concentration greater than the plume intensity threshold are screened to obtain a high-intensity plume set; the plumes in the high-intensity plume concentration are path-synthesized to obtain the main path of the thermal plume; The present invention uses high-concentration mask marking and DBSCAN spatial clustering to accurately locate the core area of ​​oil fume; combines infrared thermal imaging to dynamically track the position of personnel and calculate the real-time exposure index to generate a risk warning area; and synthesizes the main path of the thermal plume based on the thermal power calibration plume intensity. This improves the efficiency of targeted removal of oil fume sources and the speed of personnel risk response, and eliminates exhaust blind spots.

[0009] Preferably, the S4 comprises the following steps: S41. Collect historical energy consumption data, operating duration, and equipment status data to obtain historical operating data; divide the historical operating data into operating conditions according to the power change rate to obtain labeled historical operating data; S42. Build an LSTM model; set the weights of the LSTM model; use historical operation data to train the LSTM model, and use an optimization algorithm to find the weights of the LSTM model during the training process to obtain the optimal solution; use the optimal solution as the weights of the LSTM model to obtain an optimized LSTM model; S43. Input the real-time kitchen operation data into the final LSTM model to perform load prediction and fault probability prediction to obtain load prediction results and fault probability prediction results; generate a real-time dynamic power consumption strategy based on the load prediction results and fault probability prediction results; The LSTM model of the present invention intelligently predicts equipment load and failure probability, and generates a dynamic power consumption strategy that includes fault avoidance, load response, and energy efficiency optimization; it achieves precise control of energy consumption, reduces failure rates, saves energy during idle periods, and ensures performance requirements during the busy period.

[0010] Preferably, the S5 comprises the following steps: S51, unifying the coordinates of the final oil smoke motion vector field, the oil smoke core area set, the dynamic personnel area, and the main path of the heat plume into the same coordinate system, and performing time stamp alignment processing to obtain spatiotemporal alignment data; Extract the constraint boundary based on the collective coordinates of the oil smoke core area and the coordinates of the dynamic personnel area in the spatiotemporal alignment data; Based on the device parameter database, the device parameters are formatted to obtain physical limit data; S52, constructing a multi-objective function by minimizing total energy consumption, maximizing removal rate, and minimizing deviation of wind speed in the occupant zone from a comfortable value; and adjusting the weight of total energy consumption in the multi-objective function based on a real-time dynamic power consumption strategy; The upper limit of concentration in the occupant zone, vertical temperature difference constraint, physical restriction data and constraint boundary are used as the constraints of the multi-objective function; the fan speed and air supply angle are used as the decision variables of the multi-objective function; S53, finding the Pareto frontier solution by combining the multi-objective genetic algorithm with the multi-objective function, and obtaining the Pareto frontier solution set; Collect real-time working condition information; select the optimal solution from the Pareto optimal solution set based on the real-time working condition information; The present invention constructs a three-objective function of removal rate, energy consumption and comfort through the alignment of multi-source spatiotemporal data, combines equipment parameter constraints, and uses a multi-objective genetic algorithm to generate a Pareto optimal solution set. The strategy is dynamically selected according to the working conditions to achieve energy reduction while improving the efficiency of oil fume removal and accurately controlling the wind speed in the personnel area within a comfortable range.

[0011] Preferably, the S6 comprises the following steps: S61, converting the optimal solution into an execution instruction; obtaining the actual exhaust system speed based on the final oil smoke motion vector field and the main diameter of the heat plume; Based on the actual exhaust system speed and the exhaust system fan characteristic curve, the exhaust scheme in the execution instruction is adjusted by PID to obtain the exhaust speed instruction; S62. Based on the air supply solution in the execution instruction, the joint angle is solved by solving the momentum equation and combining inverse kinematics, and the servo motor position is controlled to obtain the air supply speed instruction; The exhaust air speed instruction and the supply air speed instruction together constitute the final control instruction set; The present invention dynamically calibrates the exhaust volume through the oil fume motion vector field and the main diameter of the heat plume, and generates a precise speed command in combination with the PID parameter adjustment of the fan characteristic curve; the air supply system drives the servo motor to coordinate air supply based on the inverse solution of the momentum equation; reduces the response delay of exhaust and supply air, improves the matching degree of airflow organization, and eliminates the oil fume retention problem caused by airflow turbulence.

[0012] Preferably, the S7 includes the following steps: S71. Execute the instructions in the final control instruction set and collect execution effect data to obtain actual execution effect data; compare the actual execution effect data with the expected effect of the final control instruction set to obtain a deviation result; S72, performing PID correction according to the deviation result to obtain a control parameter correction amount; and executing control according to the control parameter correction amount; The present invention generates deviation results by comparing the execution data collected in real time with the expected effects, and dynamically adjusts the control parameters using PID closed-loop correction; it achieves continuous improvement in control accuracy, stable maintenance of long-term operating energy consumption and oil fume removal efficiency, and solves the control failure problem caused by operating condition drift in traditional systems.

[0013] A kitchen smoke exhaust space vector control and airflow organization collaborative optimization management system, used to implement the above-mentioned kitchen smoke exhaust space vector control and airflow organization collaborative optimization management method, including a sensor deployment and data acquisition module, a fume movement modeling and dynamic trajectory generation module, a fume core area identification and risk area positioning module, an equipment operation prediction and energy efficiency strategy generation module, a multi-objective collaborative optimization and decision generation module, and a control instruction execution and closed-loop feedback correction module; The sensor deployment and data acquisition module is used for digital modeling and real-time data collection of kitchen spaces. It uses BIM or laser scanning point clouds to obtain stove coordinates, personnel activity area boundaries, ventilation vent locations, and obstacle information, constructs a three-dimensional spatial coordinate system, and generates a coordinate mapping table. Based on this, it deploys laser scattering fume sensors, three-dimensional ultrasonic anemometers, and other equipment to form a sensor node network. It synchronizes timestamps using the NTP protocol, collects fume concentration, wind speed, temperature, PM2.5, and equipment operation data, and generates a raw sensor data stream. The data is then time-sliced ​​and spatially interpolated to construct a structured tensor dataset containing the fume concentration field, temperature field, and wind speed vector field. The fume motion modeling and dynamic trajectory generation module is used for quantitative analysis of fume motion patterns. It calculates concentration, temperature, and wind speed gradient fields using the central difference method and the Sobel operator to generate structured field data. By combining the discretization solution of fluid dynamics equations and introducing the κ-ε turbulence model to correct boundary conditions, it outputs a high-precision fume motion vector field. By extracting fume source features, particle tracking, and thermal plume identification, it generates a dynamic vector diagram that intuitively displays the fume diffusion path and intensity changes. The oil fume core area identification and risk area positioning module is used to achieve dynamic monitoring of oil fume hotspots and personnel risks. After standardizing the oil fume concentration field, a high oil fume mask is generated through threshold filtering. The DBSCAN clustering algorithm is used to identify spatially adjacent high-concentration clusters, and the centroid position and equivalent radius are calculated to form an oil fume core area set. Infrared thermal imaging is used to track personnel activities, and the oil fume motion vector field is superimposed to calculate the exposure index and output the dynamic personnel risk area. The plume intensity is calibrated by extracting the heat source of the stove, and high-intensity plumes are screened and synthesized into the main path of the heat plume to provide a key path basis for ventilation control. The device operation prediction and energy efficiency strategy generation module uses machine learning to optimize device operation strategies; collects historical energy consumption, operating duration, and device status data, and classifies operating condition labels according to power change rate; trains an LSTM model to predict real-time load and failure probability, and combines it with an optimization algorithm to generate a dynamic power consumption strategy. This includes three mechanisms: fault avoidance, load response, and energy efficiency optimization, achieving a balanced control of device safety and energy conservation. The multi-objective collaborative optimization and decision-making module integrates environmental and operational data to achieve global optimization. It unifies the spatiotemporal coordinates of the fume vector field, fume core area, personnel activity area, and heat plume path, extracting physical constraint boundaries. It constructs a multi-objective function with the goals of minimizing energy consumption, maximizing removal rate, and minimizing wind speed deviation in the personnel area. Using fan speed and air supply angle as decision variables, it solves the Pareto front solution set using a multi-objective genetic algorithm and dynamically selects the optimal solution based on real-time operating conditions. The control instruction execution and closed-loop feedback correction module is responsible for instruction conversion and system tuning; the optimal solution is converted into exhaust and supply air control instructions, and the exhaust system adjusts the speed through PID based on the fan characteristic curve; the supply air system drives the servo motor through the inverse solution of the momentum equation; after execution, the actual effect data is collected and compared with the expected value to generate a deviation result, and the control parameters are corrected through PID to form a closed-loop feedback to ensure that the collaborative optimization effect of airflow organization continues to improve.

[0014] (3) Beneficial effects The present invention has the following beneficial effects: The present invention realizes dynamic perception and precise modeling, solving the problem of spatial diffusion blind spots; through multi-sensor spatial topology deployment and Kriging spatial interpolation technology, it constructs accurate three-dimensional tensors of oil fume concentration field, temperature field, and wind speed field, solving the problem that traditional single sensors cannot track the migration path of oil fume; combined with the solution of dynamic equations modified by the κ-ε turbulence model, it realizes dynamic prediction of the oil fume motion vector field, achieving improved accuracy compared with the static CFD model.

[0015] The present invention realizes targeted control of oil fume sources and active avoidance of personnel risks; identifies the oil fume core area based on the DBSCAN clustering algorithm, and integrates infrared thermal imaging personnel tracking to generate a dynamic risk map; accurately captures the main diffusion path of the stove heat source through the heat plume main path synthesis technology, thereby improving the directional suction efficiency of the exhaust system; and has a real-time calculation function for the exposure index of the personnel activity area, which shortens the retention time of oil fume and effectively reduces occupational health risks.

[0016] The present invention realizes the coordinated improvement of energy efficiency optimization and equipment life; uses the LSTM working condition prediction model to accurately identify the stir-fry, steady state and idle periods to generate a dynamic power consumption strategy, reduces the equipment loss rate through the fault avoidance mechanism, avoids overload tripping through the load response mechanism, and realizes energy saving through the energy efficiency optimization mechanism; and greatly reduces energy consumption compared with traditional timing control strategies.

[0017] The present invention realizes multi-objective collaborative optimization to achieve airflow organization balance; by constructing a three-objective function of removal rate, energy consumption and comfort, combined with a multi-objective genetic algorithm to generate a Pareto optimal solution set, the removal rate is prioritized to be maximized during the stir-fry period, and the wind speed comfort control in the personnel area is prioritized during the idle period. Through the dynamic switching mechanism of the global energy consumption optimal solution, the exhaust and air supply systems can be operated in coordination, and the airflow organization disorder rate is reduced.

[0018] The present invention is based on an instruction set generated by inverse solution of the momentum equation using PID tuning of the fan characteristic curve. After execution, the control parameters are automatically corrected through real-time deviation analysis. The system implements closed-loop optimization, continuously improving the matching degree between the control strategy and the actual working conditions and enhancing long-term operational stability.

[0019] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying any creative work.

[0021] Figure 1 Schematic diagram of the flow of the kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method of the present invention; Figure 2 This is a module diagram of the kitchen smoke exhaust space vector control and airflow organization collaborative optimization management system of the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] In the description of the present invention, it should be understood that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0024] Example 1: See also Figure 1 The present invention discloses a method for collaborative optimization management of kitchen smoke exhaust space vector control and airflow organization, comprising the following steps: S1. Deploy multiple sensors based on kitchen space topology data; collect real-time environmental data sets and kitchen real-time operation data through multiple sensors; Said S1 comprises the following steps: S11. Collect the precise coordinates of the stove, the boundary coordinates of the personnel activity area (polygon vertex set), the location and size of the exhaust / supply vents (including installation height), and the coordinates of obstacles (kitchen cabinets / equipment / columns) by importing the BIM model or laser scanning point cloud (accuracy ±5cm) to obtain spatial topology data. Establish a three-dimensional coordinate system (the origin is usually set in the corner of the kitchen), mark the stove area, personnel activity area, and ventilation vents in the spatial topology data, and obtain the kitchen space coordinate mapping table; S12. Deploy multiple sensors based on the kitchen space coordinate mapping table to obtain a sensor node location table; the multiple sensors include a laser scattering fume sensor, a three-dimensional ultrasonic anemometer, an infrared temperature array, and a PM2.5 sensor; The NTP protocol is used to align the timestamps of all sensors in the sensor node location table, and coordinate tags are added to each sampling point. Real-time data on oil smoke concentration, wind speed, temperature, and PM2.5 are collected to obtain the original sensor data stream. Energy consumption data, operating hours, and equipment status data of the kitchen are collected to obtain real-time kitchen operation data. S13, reorganizing the original sensor data stream by time slice, performing Kriging spatial interpolation on the missing points in the original sensor data, and constructing a structured tensor to obtain a real-time environmental dataset including the oil smoke concentration field, temperature field, and wind speed vector field; S2. Based on the real-time environmental data set, structured field data is obtained through the central difference method and the Sobel operator. Based on the structured field data, the final oil fume motion vector field is obtained by solving the dynamic equation. The oil fume features are extracted and combined with the final oil fume motion vector field to perform dynamic trajectory modeling and obtain a dynamic vector diagram. The S2 comprises the following steps: S21, unify all sensor timestamps and interpolate the discrete point data in the real-time environmental data set into the grid to obtain processed environmental data; Based on the processed environmental data, the concentration gradient field, temperature gradient field and wind speed gradient field are calculated by the central difference method and the Sobel operator respectively, and the structured field data are obtained; S22. Discretize the continuous equations in the structured field data into differential equations on the grid points to obtain a set of discretized equations (each grid point has an independent equation); Calculate the concentration gradient, wind speed and temperature cross product terms of each equation set in the discretized equation set, and calculate the concentration time derivative to obtain the oil smoke motion vector field at the current moment; Adding the κ-ε model correction term to the oil fume motion vector field at the current moment to obtain the corrected oil fume motion vector field; The corrected oil fume motion vector field is set with the wall boundary and the vent boundary to obtain the final oil fume motion vector field; S23. Based on the final oil smoke motion vector field, perform source intensity calculation, source location, and uncertainty analysis to obtain oil smoke source characteristics; Based on the final oil smoke motion vector field and oil smoke source characteristics, particle tracking, thermal plume identification and diffusion direction analysis are performed on the oil smoke source to obtain a dynamic vector diagram; S3. Based on the real-time environmental data set, perform oil fume concentration field threshold filtering and cluster analysis to obtain the oil fume core area set; Through exposure risk assessment, the dynamic personnel zone is obtained; based on the dynamic vector diagram, the thermal plume characteristics are quantified to obtain the main path of the thermal plume; The S3 includes the following steps: S31, performing concentration unit normalization processing on the oil fume concentration field in the real-time environmental data set and matching the timestamp of the final oil fume motion vector field to obtain a standardized oil fume concentration field; Create a high oil smoke mask, such as when the oil smoke concentration exceeds 50 mg / m³; mark the standardized oil smoke concentration field based on the high oil smoke mask to obtain a marked concentration field; S32, setting the neighborhood radius (kitchen scale experience value) and the minimum number of cluster points of the DBSCAN algorithm to obtain the parameters of the DBSCAN algorithm; Cluster analysis of the labeled concentration field is performed based on the parameters of the DBSCAN algorithm to obtain spatially adjacent high-concentration point groups; Remove noise points from spatially adjacent high-concentration point clusters, and calculate the centroid position, equivalent radius, and peak concentration of each cluster to obtain the oil smoke core area set; S33. Based on the kitchen space topology and the dynamic tracking of thermal imaging by the infrared positioning system, the actual activity area of ​​the personnel is obtained; the final oil smoke motion vector field is superimposed on the actual activity area of ​​the personnel, and the exposure index is calculated to obtain the real-time dynamic personnel area; S34, extracting the stove heat source from the dynamic vector diagram and performing coefficient calibration to obtain the stove heat power; calculating the plume intensity in the dynamic vector diagram based on the stove heat power to obtain a plume intensity set; A plume intensity threshold is set, and the plume intensities with plume intensity concentration greater than the plume intensity threshold are screened to obtain a high-intensity plume set; the plumes in the high-intensity plume concentration are path-synthesized to obtain the main path of the thermal plume; S4. Use historical kitchen operation data to train the LSTM model to obtain the final LSTM model. Input the real-time kitchen operation data into the final LSTM model to perform load prediction and failure probability prediction to obtain a real-time dynamic power consumption strategy. The S4 comprises the following steps: S41. Collect historical energy consumption data, operating duration, and device status data to obtain historical operation data; the energy consumption data includes {timestamp, device ID, power (kW)}, the operating duration includes {start time, end time, operating type}, and the device status includes {device ID, timestamp, operating status, fault code}; The historical operating data is divided into operating conditions according to the power change rate to obtain the historical operating data with labels; the operating conditions are as follows: if power > 0.5kW / s: operating condition = "frying period", if 0.1 < power ≤ 0.5: operating condition = "steady state", else: operating condition = "idle"; S42. Build an LSTM model; set the weights of the LSTM model; use historical operation data to train the LSTM model, and use an optimization algorithm to find the weights of the LSTM model during the training process to obtain the optimal solution; use the optimal solution as the weights of the LSTM model to obtain an optimized LSTM model; S43. Input the real-time kitchen operation data into the final LSTM model to perform load prediction and fault probability prediction to obtain load prediction results and fault probability prediction results; generate a real-time dynamic power consumption strategy based on the load prediction results and fault probability prediction results; For example, under the fault avoidance mechanism, when the probability of equipment failure exceeds 70%, the power is limited to 80% of the rated value during the hot-spot period, the steady-state operating power is reduced to 60%, and the idle power is limited to 20% (fault isolation is achieved through step-by-step power reduction); Under the load response mechanism, when the predicted power peak exceeds 8kW, full power is maintained during stir-frying, steady-state power is reduced to 80%, and idle power is limited to 40% (prioritizing core cooking performance). Under the energy efficiency optimization mechanism, under normal operating conditions, the stir-fry period runs at full power (meeting thermodynamic requirements), the steady-state operating power is reduced to 70%, and the idle period power is limited to 30% (achieving a conventional energy saving effect of 30%); S5. Construct a multi-objective function based on the real-time dynamic power consumption strategy; obtain a Pareto optimal solution set using a multi-objective genetic algorithm based on the final oil fume motion vector field, the oil fume core area set, the dynamic personnel area, the main path of the heat plume, and the multi-objective function; and select the optimal solution from the Pareto optimal solution set based on the real-time operating conditions. The S5 comprises the following steps: S51, unifying the coordinates of the final oil smoke motion vector field, the oil smoke core area set, the dynamic personnel area, and the main path of the heat plume into the same coordinate system, and performing time stamp alignment processing to obtain spatiotemporal alignment data; Extract the constraint boundary based on the collective coordinates of the oil smoke core area and the coordinates of the dynamic personnel area in the spatiotemporal alignment data; Based on the equipment parameter database (wind turbine power curve, angle range, etc.), the equipment parameters are formatted to obtain physical constraint data; the physical constraint data is used to set equipment physical constraints in multi-objective optimization; S52: Construct a multi-objective function by minimizing total energy consumption, maximizing removal rate, and minimizing deviation of wind speed from the comfort value in the personnel area; the multi-objective function is specifically: Minimize F ( x ) = [ f 1(x) , f 2(x) , f 3(x) ] Among them, Minimize F ( x ) represents minimizing the multi-objective function, f 1(x) =Σ(fan power + air supply system power)*Δ t / Total energy consumption, Δ t represents the time step; f 2(x) = -(∂C / ∂t)*V zone , ∂C / ∂t represents the partial derivative of the pollutant concentration C with respect to time t; it represents the rate of change of the pollutant concentration at a certain moment. In a ventilation system, we hope that this value is negative (the concentration is decreasing). The larger the absolute value of (∂C / ∂t), the faster the pollutants are removed. The unit is mg / m³ / s. V zone Indicates the volume of the kitchen; f 3(x) =Σ|vperson - vcomfort|, where vperson represents the actual wind speed in the personnel area (e.g., the height and position of a person standing or sitting), calculated using sensors or simulation. This speed is determined by the fan speed and the air outlet angle, and is expressed in meters per second. vcomfort represents the comfortable wind speed, a preset, ideal wind speed value. The comfortable wind speed range varies depending on the season and activity. For example, in winter, the comfortable wind speed may be lower (e.g., 0.15 m / s), while in summer, it may be slightly higher (e.g., 0.25 m / s). vcomfort is this target value, expressed in meters per second. x is the decision variable, x = [ n 1, n 2,..., n i ,... n k , i 1, i 2,..., i j ,..., i m ],in n i Indicates the i The speed of the fan, k Indicates the total number of fans, i j Indicates the j The angle of the air outlet, m Indicates the total number of air outlets; Adjust the weight of total energy consumption in the multi-objective function based on the real-time dynamic power consumption strategy; specifically: dynamically adjust the weight coefficient of each sub-objective in the multi-objective function according to the working condition type (hot-spicing period, steady state period, idle period) in the real-time dynamic power consumption strategy w i , weighted aggregation is used to solve a single objective function or to sort individuals in a multi-objective genetic algorithm, that is, F weighted(x) =w 1* f 1(x) + w 2* f 2(x) + w 3* f 3(x) , the weight adjustment strategy is: increase f 2(x) The weight (i.e. w 2 increases), steady-state period increases f 1(x) The weight (i.e. w 1 increases), idle period increases f 3(x) The weight (i.e. w 3 increase); The upper limit of concentration in the occupant zone, vertical temperature difference constraint, physical restriction data and constraint boundary are used as the constraints of the multi-objective function; the fan speed and air supply angle are used as the decision variables of the multi-objective function; S53, finding the Pareto frontier solution by combining the multi-objective genetic algorithm with the multi-objective function, and obtaining the Pareto frontier solution set; The specific steps of the multi-objective genetic algorithm include: S531. Randomly generate an initial population of N individuals, each of which represents a set of decision variables. x (Combination of fan speed and air supply angle); set the maximum number of iterations S532. Calculate the multi-objective function value corresponding to each individual in the population F ( x ); Stratify and sort individuals in a population according to the Pareto dominance relationship; In the same non-dominated layer, calculate the crowding distance of each individual in the objective function space; Select parent individuals based on non-dominated rank and crowding degree, and perform crossover and mutation operations to generate offspring populations; Merge the parent population and the offspring population, perform non-dominated sorting and crowding calculation, and select the first N individuals as the new generation population; S533, repeat S532 until the maximum number of iterations is reached, and finally output the non-dominated solution set as the Pareto front solution set; S54, collecting real-time working condition information; selecting the optimal solution from the Pareto optimal solution set based on the real-time working condition information; if the working condition is stir-frying, prioritizing solutions with high removal rates; if the working condition is steady-state, prioritizing solutions with low energy consumption; if the working condition is idle, prioritizing solutions with high comfort; S6. Convert the optimal solution into execution instructions, and perform coordinated optimization of the airflow organization of exhaust control and supply air control based on the final oil fume motion vector field and the main diameter of the heat plume to obtain the final control instruction set; The S6 comprises the following steps: S61, converting the optimal solution into an execution instruction; obtaining the actual exhaust system speed based on the final oil smoke motion vector field and the main diameter of the heat plume; Based on the actual exhaust system speed and the exhaust system's fan characteristic curve (maximum air volume / power ratio), PID adjustment is performed on the exhaust plan in the execution instruction to obtain the exhaust speed instruction; S62, based on the air supply scheme in the execution instruction, the joint angle is solved by solving the momentum equation combined with inverse kinematics, and the servo motor position is controlled to obtain the air supply speed instruction; specifically, the air supply angle decision variable in the optimal solution is i j The joint angle of the air louver is solved by inverse kinematics. The specific solution process is based on the wind momentum equation M = r * A * v 2 * sin i j , combined with the expected air supply speed v and angles i j , reversely calculate the joint angle α that the servo motor needs to rotate, the formula is α= g( i j , v ), and through the servo motor position control, the air supply speed instruction is obtained; where M represents the torque, r represents the air density, A represents the effective force area of ​​the air supply louver joint, and g() represents the inverse kinematics mapping function; The exhaust air speed instruction and the supply air speed instruction together constitute the final control instruction set; S7. Performing deviation analysis on the execution effect data of the final control instruction set to obtain a control parameter correction; and executing control based on the control parameter correction; The S7 comprises the following steps: S71. Execute the instructions in the final control instruction set and collect execution effect data to obtain actual execution effect data; compare the actual execution effect data with the expected effect of the final control instruction set to obtain a deviation result; S72. Perform PID correction according to the deviation result to obtain a control parameter correction amount; and execute control using the control parameter correction amount.

[0025] Example 2: See also Figure 2, a kitchen smoke exhaust space vector control and airflow organization collaborative optimization management system, which is used to implement the above-mentioned kitchen smoke exhaust space vector control and airflow organization collaborative optimization management method, including a sensor deployment and data acquisition module, a fume movement modeling and dynamic trajectory generation module, a fume core area identification and risk area positioning module, an equipment operation prediction and energy efficiency strategy generation module, a multi-objective collaborative optimization and decision generation module, and a control instruction execution and closed-loop feedback correction module; The sensor deployment and data acquisition module is used for digital modeling and real-time data collection of kitchen spaces. It uses BIM or laser scanning point clouds to obtain stove coordinates, personnel activity area boundaries, ventilation vent locations, and obstacle information, constructs a three-dimensional spatial coordinate system, and generates a coordinate mapping table. Based on this, it deploys laser scattering fume sensors, three-dimensional ultrasonic anemometers, and other equipment to form a sensor node network. It synchronizes timestamps using the NTP protocol, collects fume concentration, wind speed, temperature, PM2.5, and equipment operation data, and generates a raw sensor data stream. The data is then time-sliced ​​and spatially interpolated to construct a structured tensor dataset containing the fume concentration field, temperature field, and wind speed vector field. The fume motion modeling and dynamic trajectory generation module is used for quantitative analysis of fume motion patterns. It calculates concentration, temperature, and wind speed gradient fields using the central difference method and the Sobel operator to generate structured field data. By combining the discretization solution of fluid dynamics equations and introducing the κ-ε turbulence model to correct boundary conditions, it outputs a high-precision fume motion vector field. By extracting fume source features, particle tracking, and thermal plume identification, it generates a dynamic vector diagram that intuitively displays the fume diffusion path and intensity changes. The oil fume core area identification and risk area positioning module is used to achieve dynamic monitoring of oil fume hotspots and personnel risks. After standardizing the oil fume concentration field, a high oil fume mask is generated through threshold filtering. The DBSCAN clustering algorithm is used to identify spatially adjacent high-concentration clusters, and the centroid position and equivalent radius are calculated to form an oil fume core area set. Infrared thermal imaging is used to track personnel activities, and the oil fume motion vector field is superimposed to calculate the exposure index and output the dynamic personnel risk area. The plume intensity is calibrated by extracting the heat source of the stove, and high-intensity plumes are screened and synthesized into the main path of the heat plume to provide a key path basis for ventilation control. The device operation prediction and energy efficiency strategy generation module uses machine learning to optimize device operation strategies; collects historical energy consumption, operating duration, and device status data, and classifies operating condition labels according to power change rate; trains an LSTM model to predict real-time load and failure probability, and combines it with an optimization algorithm to generate a dynamic power consumption strategy. This includes three mechanisms: fault avoidance, load response, and energy efficiency optimization, achieving a balanced control of device safety and energy conservation. The multi-objective collaborative optimization and decision-making module integrates environmental and operational data to achieve global optimization. It unifies the spatiotemporal coordinates of the fume vector field, fume core area, personnel activity area, and heat plume path, extracting physical constraint boundaries. It constructs a multi-objective function with the goals of minimizing energy consumption, maximizing removal rate, and minimizing wind speed deviation in the personnel area. Using fan speed and air supply angle as decision variables, it solves the Pareto front solution set using a multi-objective genetic algorithm and dynamically selects the optimal solution based on real-time operating conditions. The control instruction execution and closed-loop feedback correction module is responsible for instruction conversion and system tuning; the optimal solution is converted into exhaust and supply air control instructions, and the exhaust system adjusts the speed through PID based on the fan characteristic curve; the supply air system drives the servo motor through the inverse solution of the momentum equation; after execution, the actual effect data is collected and compared with the expected value to generate a deviation result, and the control parameters are corrected through PID to form a closed-loop feedback to ensure that the collaborative optimization effect of airflow organization continues to improve.

[0026] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0027] The preferred embodiments of the invention disclosed above are intended only to help illustrate the invention. These preferred embodiments do not exhaust all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for collaborative optimization management of kitchen smoke exhaust space vector control and airflow organization, characterized in that: The following steps are involved: S1. Deploy multiple sensors based on kitchen space topology data; collect real-time environmental data sets and kitchen real-time operation data through multiple sensors; S2, based on the real-time environmental data set, obtain structured field data through the central difference method and Sobel operator; Based on the structured field data, the final fume motion vector field is obtained by solving the dynamic equations. Extract the characteristics of the oil smoke based on the final oil smoke motion vector field and perform dynamic trajectory modeling to obtain a dynamic vector diagram; S3. Based on the real-time environmental data set, perform oil smoke concentration field threshold filtering and cluster analysis to obtain the oil smoke core area set; Through exposure risk assessment, dynamic personnel areas are obtained; Based on the dynamic vector diagram, the thermal plume characteristics are quantified to obtain the main diameter of the thermal plume; S4. Input the real-time kitchen operation data into the LSTM model trained with historical kitchen operation data to perform load prediction and failure probability prediction to obtain a real-time dynamic power consumption strategy; S5. Construct a multi-objective function based on a real-time dynamic power consumption strategy; obtain a Pareto optimal solution set using a multi-objective genetic algorithm based on the oil smoke core area set, the dynamic personnel area, the main path of the heat plume, and the multi-objective function; and select the optimal solution from the Pareto optimal solution set based on the real-time operating conditions. S6. Convert the optimal solution into execution instructions, and perform coordinated optimization of the airflow organization of exhaust control and supply air control to obtain the final control instruction set; S7. Perform deviation analysis on the execution effect data of the final control instruction set to obtain a control parameter correction amount; and perform control using the control parameter correction amount.

2. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: S11. By importing the BIM model or laser scanning the point cloud, collect the precise coordinates of the stove, the boundary coordinates of the personnel activity area, the location and size of the exhaust vents and air supply vents, and the coordinates of obstacles to obtain spatial topology data; Establish a three-dimensional coordinate system, mark the stove area, personnel activity area, and ventilation vent locations in the spatial topology data, and obtain the kitchen space coordinate mapping table; S12. Deploy multiple sensors based on the kitchen space coordinate mapping table to obtain a sensor node location table; the multiple sensors include a laser scattering fume sensor, a three-dimensional ultrasonic anemometer, an infrared temperature array, and a PM2.5 sensor; The NTP protocol is used to align the timestamps of all sensors in the sensor node location table, and coordinate tags are added to each sampling point. Real-time data on oil smoke concentration, wind speed, temperature, and PM2.5 are collected to obtain the original sensor data stream. Energy consumption data, operating hours, and equipment status data of the kitchen are collected to obtain real-time kitchen operation data. S13. Reorganize the original sensor data stream by time slice, perform Kriging spatial interpolation on the missing points in the original sensor data, and construct a structured tensor to obtain a real-time environmental data set containing the oil fume concentration field, temperature field, and wind speed vector field.

3. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: The S2 comprises the following steps: S21, unify all sensor timestamps and interpolate the discrete point data in the real-time environmental data set into the grid to obtain processed environmental data; Based on the processed environmental data, the concentration gradient field, temperature gradient field and wind speed gradient field are calculated by the central difference method and the Sobel operator respectively, and the structured field data are obtained; S22, discretizing the continuous equations in the structured field data into differential equations on grid points to obtain a discretized set of equations; Calculate the concentration gradient, wind speed and temperature cross product terms of each equation set in the discretized equation set, and calculate the concentration time derivative to obtain the oil smoke motion vector field at the current moment; Adding the κ-ε model correction term to the oil fume motion vector field at the current moment to obtain the corrected oil fume motion vector field; The corrected oil fume motion vector field is set with the wall boundary and the vent boundary to obtain the final oil fume motion vector field; S23. Based on the final oil smoke motion vector field, perform source intensity calculation, source location, and uncertainty analysis to obtain oil smoke source characteristics; Based on the final oil smoke motion vector field and oil smoke source characteristics, particle tracking, thermal plume identification and diffusion direction analysis are performed on the oil smoke source to obtain a dynamic vector diagram.

4. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: The S3 includes the following steps: S31, performing concentration unit normalization processing on the oil fume concentration field in the real-time environmental data set and matching the timestamp of the final oil fume motion vector field to obtain a standardized oil fume concentration field; Create a high oil smoke mask; mark the standardized oil smoke concentration field based on the high oil smoke mask to obtain a marked concentration field; S32, setting the neighborhood radius and the minimum number of cluster points of the DBSCAN algorithm to obtain the parameters of the DBSCAN algorithm; Cluster analysis of the labeled concentration field is performed based on the parameters of the DBSCAN algorithm to obtain spatially adjacent high-concentration point groups; Remove noise points from spatially adjacent high-concentration point clusters, and calculate the centroid position, equivalent radius, and peak concentration of each cluster to obtain the oil smoke core area set; S33. Based on the kitchen space topology and the dynamic tracking of thermal imaging by the infrared positioning system, the actual activity area of ​​the personnel is obtained; the final oil smoke motion vector field is superimposed on the actual activity area of ​​the personnel, and the exposure index is calculated to obtain the real-time dynamic personnel area; S34, extracting the stove heat source from the dynamic vector diagram and performing coefficient calibration to obtain the stove heat power; calculating the plume intensity in the dynamic vector diagram based on the stove heat power to obtain a plume intensity set; A plume intensity threshold is set, and the plume intensities with a plume intensity concentration greater than the plume intensity threshold are screened to obtain a high-intensity plume set; the plumes in the high-intensity plume concentration are path-synthesized to obtain the main path of the thermal plume.

5. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: The S4 comprises the following steps: S41. Collect historical energy consumption data, operating duration, and equipment status data to obtain historical operating data; divide the historical operating data into operating conditions according to the power change rate to obtain labeled historical operating data; S42. Build an LSTM model; set the weights of the LSTM model; use historical operation data to train the LSTM model, and use an optimization algorithm to find the weights of the LSTM model during the training process to obtain the optimal solution; use the optimal solution as the weights of the LSTM model to obtain an optimized LSTM model; S43. Input the real-time kitchen operation data into the final LSTM model to perform load prediction and fault probability prediction to obtain load prediction results and fault probability prediction results; based on the load prediction results and fault probability prediction results, generate a real-time dynamic power consumption strategy.

6. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: The S5 comprises the following steps: S51, unifying the coordinates of the final oil smoke motion vector field, the oil smoke core area set, the dynamic personnel area, and the main path of the heat plume into the same coordinate system, and performing time stamp alignment processing to obtain spatiotemporal alignment data; Extract the constraint boundary based on the collective coordinates of the oil smoke core area and the coordinates of the dynamic personnel area in the spatiotemporal alignment data; Based on the device parameter database, the device parameters are formatted to obtain physical limit data; S52, constructing a multi-objective function by minimizing total energy consumption, maximizing removal rate, and minimizing deviation of wind speed in the occupant zone from a comfortable value; and adjusting the weight of total energy consumption in the multi-objective function based on a real-time dynamic power consumption strategy; The upper limit of concentration in the occupant zone, vertical temperature difference constraint, physical restriction data and constraint boundary are used as the constraints of the multi-objective function; the fan speed and air supply angle are used as the decision variables of the multi-objective function; S53, finding the Pareto frontier solution by combining the multi-objective genetic algorithm with the multi-objective function, and obtaining the Pareto frontier solution set; Collect real-time working condition information; select the optimal solution from the Pareto optimal solution set based on the real-time working condition information.

7. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: The S6 comprises the following steps: S61, converting the optimal solution into an execution instruction; obtaining the actual exhaust system speed based on the final oil smoke motion vector field and the main diameter of the heat plume; Based on the actual exhaust system speed and the exhaust system fan characteristic curve, the exhaust scheme in the execution instruction is adjusted by PID to obtain the exhaust speed instruction; S62. Based on the air supply solution in the execution instruction, the joint angle is solved by solving the momentum equation and combining inverse kinematics, and the servo motor position is controlled to obtain the air supply speed instruction; The exhaust air speed instruction and the supply air speed instruction together constitute the final control instruction set.

8. The kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to claim 1 is characterized in that: The S7 comprises the following steps: S71. Execute the instructions in the final control instruction set and collect execution effect data to obtain actual execution effect data; compare the actual execution effect data with the expected effect of the final control instruction set to obtain a deviation result; S72. Perform PID correction according to the deviation result to obtain a control parameter correction amount; and execute control using the control parameter correction amount.

9. Kitchen exhaust space vector control and airflow organization coordinated optimization management system, characterized by: A method for collaborative optimization management of kitchen smoke exhaust space vector control and airflow organization as described in any one of claims 1 to 8 is implemented, wherein the system includes a sensor deployment and data acquisition module, an oil smoke movement modeling and dynamic trajectory generation module, an oil smoke core area identification and risk area positioning module, an equipment operation prediction and energy efficiency strategy generation module, a multi-objective collaborative optimization and decision generation module, and a control instruction execution and closed-loop feedback correction module.

10. A storage medium, characterized in that: A program is stored thereon, and when the program is executed by the processor, the kitchen smoke exhaust space vector control and airflow organization coordinated optimization management method according to any one of claims 1 to 8 is implemented.

Images