Dynamic regulation and control system for cooking fume emission

Abstract

The invention relates to the technical field of catering oil fume emission time sequence accurate cooperative control, in particular to a catering oil fume emission dynamic regulation and control system which comprises a monitoring module for collecting oil fume concentration, environmental parameters and execution mechanism state signals; the dynamic modeling module generates an emission trend predicted value; the response quantification module is combined with the delay parameters to construct a physical dynamic response model; the time sequence compensation module is used for calculating fan lag offset, purification unit advance offset and air valve interlocking offset based on real-time working conditions; the cooperative control module outputs a synchronization instruction carrying an offset parameter, so that the fan speed is increased, and the activation time of the purification unit is aligned with the opening and closing action time of the air valve; the closed-loop correction module extracts an actual response curve of the execution mechanism; and the model updating module compares the theoretical output value and updates the delay parameter when the deviation exceeds the threshold value. Through physical response modeling and a dynamic time sequence compensation mechanism, the problem of time sequence mismatch of cooperative control of the fan, the purification unit and the air valve is solved, the purification efficiency is improved, and invalid energy consumption is reduced.

Description

Technical Field

[0001] This invention relates to the field of precise and coordinated control of catering fume emissions, and in particular to a dynamic control system for catering fume emissions. Background Technology

[0002] Dynamic control of restaurant fume emissions is an intelligent control method based on real-time data monitoring and prediction models. By installing fume concentration sensors and environmental parameter monitoring devices (such as temperature, humidity, and wind speed), the system collects and analyzes kitchen operation data. Algorithms are used to construct emission fluctuation prediction models, generating dynamic control strategies to drive adaptive adjustments in exhaust fan speed or purification equipment power. Its operating logic is as follows: fixed emission thresholds are difficult to adapt to the cyclical changes in cooking activities and external environmental influences, often leading to energy waste or the risk of pollutant exceedances; dynamic control optimizes resource allocation based on real-time emission demands, thereby improving treatment efficiency, reducing ineffective treatment practices, and meeting environmental standards.

[0003] In practical applications, dynamic control systems for restaurant fume emissions struggle to accurately coordinate the action logic and response timing of parallel or series actuators, such as fan speed adjustment, purification unit start / stop / power, and valve opening / closing, based on real-time changes in fume emission characteristics, environmental disturbances, and equipment operating status. For example, when sensors detect a sudden surge in fume concentration (such as during stir-frying), the control system needs to simultaneously increase the fan's suction negative pressure and activate the ionization function of the high-voltage electrostatic field. If the fan speed increase lags behind the purifier's high-voltage start-up, the uncaptured fumes will escape and diffuse. Conversely, if the purifier's power increase lags behind, oil droplets in the high-speed airflow will not be effectively charged and adsorbed, leading to a decrease in purification efficiency. This control timing mismatch stems from the differences in the physical response time constants of each actuator (fan inertial delay of approximately 2-5 seconds, and purifier electric field establishment requiring 0.5-1 second) and the simplification defects of the multivariable coupled control model, resulting in instantaneous purification efficiency loss and energy waste. Therefore, it is urgent to optimize the collaborative control mechanism through high-order nonlinear dynamic compensation algorithms. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a dynamic control system for restaurant fume emissions, which solves the problems of decreased purification efficiency and energy waste caused by mismatch in the timing of multi-actuator coordinated control.

[0005] To solve the above-mentioned technical problems, the specific technical solution of the present invention is as follows: The dynamic control system for restaurant fume emissions provided by this invention includes: The monitoring module collects signals of oil fume concentration, environmental parameters, and actuator status. The dynamic modeling module connects to the monitoring module, receives oil fume concentration signals and environmental parameter signals, and generates predicted values ​​for oil fume emission trends. The response quantization module receives the actuator status signal from the monitoring module, and combines it with the stored fan mechanical acceleration delay parameters, purification unit power ramp-up delay parameters, and damper opening adjustment delay parameters to generate a physical dynamic response model. The timing compensation module connects the dynamic modeling module and the response quantization module. It receives the predicted value of oil fume emission trend and the physical dynamic response model. Based on the real-time operating condition changes in the actuator status signal, it calculates the fan command lag offset value, the purification unit command advance offset value, and the valve interlock offset value. The collaborative control module, connected to the timing compensation module, outputs a synchronous trigger command carrying offset parameters to the actuator, aligning the fan speed-up action, the purification unit activation action, and the air valve opening and closing action in the time dimension. The closed-loop correction module receives feedback values ​​of fan speed, electric field current, and damper position, and extracts the actual response curve of the actuator. The model update module connects the closed-loop correction module and the response quantization module. It inputs the actual response curve of the actuator into the physical dynamic response model to obtain the theoretical output value. When the response deviation exceeds the tolerance threshold, it sends the updated delay parameter data to the response quantization module and the timing compensation module.

[0006] Furthermore, in the dynamic control system for catering fume emissions described in this invention, the response quantification module is configured such that the mechanical acceleration delay parameter of the fan is 2-5 seconds, the power ramp-up delay parameter of the purification unit is 0.5-1 seconds, and the damper opening adjustment delay parameter is 0.2-0.8 seconds.

[0007] Furthermore, in the dynamic control system for restaurant fume emissions described in this invention, the timing compensation module is configured as follows: When identifying a surge in oil fume concentration, based on the upper limit of 0.5-1 seconds for the power ramp-up delay parameter of the purification unit in the response quantification module, the advance offset value of the purification unit command is set to 0.8 seconds; Based on the lower limit of the wind turbine mechanical acceleration delay parameter of 2-5 seconds, the wind turbine command lag offset value is set to 2 seconds.

[0008] Furthermore, the dynamic control system for catering fume emissions described in this invention includes a multi-level decision-making unit in its collaborative control module. The multi-level decision-making unit is used to respond to the stir-frying condition event defined by the instantaneous change rate of fume concentration exceeding 15% per second, which is output by the dynamic modeling module. The first stage outputs commands to increase negative pressure in the fan and adjust the opening of the damper. The second stage uses a closed-loop correction module to detect the fan output pressure in real time. When the pressure value reaches 90% of the theoretical pressure value of the physical dynamic response model in the response quantification module, a high-voltage electric field activation command for the purification unit is output.

[0009] Furthermore, the dynamic control system for restaurant fume emissions described in this invention includes a simulation verification unit in its collaborative control module. The simulation verification unit is used to pre-simulate the output command sequence of the multi-level decision-making unit. The simulation results showed that the oil fume adsorption efficiency was lower than the set standard value, which triggered the genetic algorithm optimization unit to adjust the branch weights of the multi-level decision units. The optimized decision weights are fed back to the multi-level decision-making units to update their control strategy execution rules.

[0010] Furthermore, in the dynamic control system for restaurant fume emissions described in this invention, the deviation detection operation of the closed-loop correction module is configured as follows: Calculate the absolute value of the difference between the actual response time of the fan speed and the delay value of the theoretical curve of the physical dynamic response model in the response quantization module; The standard deviation between the settling time of the electric field current in the detection purification unit and the theoretical ramp-up curve output by the timing compensation module is measured.

[0011] Furthermore, the dynamic control system for catering fume emissions described in this invention also includes a model update module that sends a parameter iteration instruction to the response quantization module when the standard deviation is greater than 5%. The response quantization module uses the gradient descent method to establish the time standard deviation based on the current detected by the closed-loop correction module, corrects the stored fan inertia coefficient and purification unit response order, and the updated parameters are synchronously fed back to the offset value calculation process of the timing compensation module.

[0012] Furthermore, the dynamic control system for restaurant fume emissions described in this invention also includes: The multi-level decision-making unit determines the frying condition event based on the continuous low-amplitude concentration fluctuation characteristics output by the dynamic modeling module, and executes the operation sequence: The initial instruction set consists of power preheating instruction for the purification unit, half-open instruction for the air valve, and low-speed operation instruction for the fan. When the rate of change of oil fume concentration gradient exceeds 10% per second, the collaborative control module is triggered to output a fully open interlock command for the air valve.

[0013] Furthermore, the dynamic control system for restaurant fume emissions described in this invention also includes: The timing compensation module is embedded in the feedforward controller, and its input port receives outdoor temperature disturbance data and stove thermal imaging activity index. The feedforward controller outputs a compensation gain factor, which is used to calculate the offset value of the timing compensation module, dynamically correcting the linkage offset between the fan and the damper.

[0014] Furthermore, the dynamic control system for restaurant fume emissions described in this invention includes a dual-redundant architecture for the synchronous trigger command transmission channel, with the main channel using the RS485 bus protocol and the backup channel using the CAN bus protocol. The collaborative control module monitors the packet loss rate of the main channel in real time. When the packet loss rate exceeds 5%, it triggers seamless switching of the backup channel to maintain the integrity of the transmission of offset parameter commands.

[0015] Beneficial effects of this invention; This invention precisely quantifies the mechanical delay characteristics of the fan, purification unit, and air valve through a physical dynamic response model. Based on real-time operating conditions, it calculates the command offset, ensuring that the activation action of the purification unit precedes the fan acceleration action, thus eliminating the problems of delayed electric field establishment and airflow turbulence caused by differences in response speed. A multi-level decision-making unit outputs control commands in stages, and a closed-loop correction module provides real-time feedback on execution deviations. The gradient descent algorithm dynamically iterates response parameters, simultaneously optimizing timing compensation accuracy and control strategy. Dual redundant command channels and a feedforward environmental compensation mechanism work together to ensure control robustness, significantly improving oil fume adsorption efficiency by more than 15%, while reducing ineffective fan power consumption and purification unit idle loss, achieving synergistic optimization of energy consumption and treatment efficiency. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0017] Figure 1 The system architecture diagram of the dynamic control system for restaurant fume emissions provided in the embodiments of the present invention is shown. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The technical solutions provided by various embodiments of this invention will be described in detail below with reference to the accompanying drawings. To better understand the objectives of this invention, it will be described in further detail below.

[0019] Please see Figure 1 The dynamic control system for restaurant fume emissions provided by this invention includes: The monitoring module collects signals of oil fume concentration, environmental parameters, and actuator status; the dynamic modeling module connects to the monitoring module, receives oil fume concentration and environmental parameters, and generates predicted values ​​of oil fume emission trends. The response quantization module receives the actuator status signal from the monitoring module, and combines it with the stored fan mechanical acceleration delay parameters, purification unit power ramp-up delay parameters, and damper opening adjustment delay parameters to generate a physical dynamic response model. The timing compensation module connects the dynamic modeling module and the response quantization module. It receives the predicted value of oil fume emission trend and the physical dynamic response model. Based on the real-time operating condition changes in the actuator status signal, it calculates the fan command lag offset value, the purification unit command advance offset value, and the valve interlock offset value. The collaborative control module, connected to the timing compensation module, outputs a synchronous trigger command carrying offset parameters to the actuator, aligning the fan speed-up action, the purification unit activation action, and the air valve opening and closing action in the time dimension. The closed-loop correction module receives feedback values ​​of fan speed, electric field current, and damper position, and extracts the actual response curve of the actuator. The model update module connects the closed-loop correction module and the response quantization module. It inputs the actual response curve of the actuator into the physical dynamic response model to obtain the theoretical output value. When the response deviation exceeds the tolerance threshold, it sends the updated delay parameter data to the response quantization module and the timing compensation module.

[0020] The dynamic control system for restaurant fume emissions provided by this invention solves the timing mismatch problem of the actuators through the collaborative work of multiple modules. The technical solutions and logical relationships of each module are described below: The monitoring module, acting as the system's data acquisition terminal, acquires real-time oil fume particle concentration data through an oil fume concentration sensor, collects kitchen environmental parameters through temperature and humidity sensors, and simultaneously monitors the real-time operating status of the fan, purification unit, and air valves. These raw signals form the basis for data processing in subsequent modules.

[0021] The dynamic modeling module receives oil fume concentration and environmental parameter data transmitted from the monitoring module and processes the signal fluctuation characteristics using time series analysis. This module identifies the periodic patterns of oil fume emissions through a sliding window algorithm, generating trend predictions of future oil fume concentration changes. The prediction results reflect changes in cooking activity intensity, providing a basis for predicting operating conditions for time-series compensation.

[0022] The response quantization module receives real-time status signals from the actuators collected by the monitoring module and constructs a dynamic model by combining them with pre-stored physical response parameters. The fan mechanical acceleration delay parameter reflects the inertial time it takes for the motor to reach its rated speed from rest; the purification unit power ramp-up delay parameter characterizes the time required for the high-voltage electric field to establish a stable ionization field; and the valve opening adjustment delay parameter records the mechanical transmission response time of the valve. The module integrates real-time status and historical parameters to generate a physical dynamic response model, quantifying the time constants of each actuator's actions.

[0023] The timing compensation module synchronously receives the trend prediction values ​​from the dynamic modeling module and the physical dynamic response model from the response quantification module. This module analyzes the real-time operating characteristics in the actuator status signals. When a sharp increase in oil fume concentration is detected, it calculates the command trigger time offset: it sets an advance offset value based on the upper limit of the power ramp-up delay of the purification unit, so that the high-voltage electric field is established before the oil fume peak arrives; it sets a lag offset value based on the lower limit of the mechanical acceleration delay of the fan, so that the acceleration action avoids the period of airflow turbulence; and the valve interlock offset value is dynamically calculated based on the pipeline pressure fluctuation.

[0024] The collaborative control module embeds the offset parameters calculated by the timing compensation module into the control commands, generating synchronized trigger commands aligned with the time dimension. These commands carry parameters for the purification unit activation lead time, fan speed-up lag, and valve opening interlock, ensuring coordinated action across all three at the physical execution level. This module outputs the commands to the actuator terminal, eliminating purification efficiency losses caused by differences in response speed.

[0025] The closed-loop correction module acquires the actual rotational speed of the fan impeller through a speed sensor, records the electric field establishment curve of the purification unit through a current sensor, and obtains the valve opening feedback value through a position encoder. The module extracts the actual response curve of the actuator, compares it with the theoretical model in the time domain, and identifies the action delay deviation.

[0026] The model update module inputs the actual response curves collected by the closed-loop correction module into the physical dynamic response model for iterative verification. When the standard deviation of the current settling time of the purification unit exceeds the tolerance threshold, the parameter update mechanism is triggered. The updated fan inertia coefficient and the purification unit response order parameters are synchronously fed back to the model storage unit of the response quantization module and the calculation process of the timing compensation module, forming a dynamic optimization closed loop.

[0027] Specifically, in the dynamic control system for catering fume emissions described in this invention, the response quantification module has a fan mechanical acceleration delay parameter of 2-5 seconds, a purification unit power ramp-up delay parameter of 0.5-1 seconds, and a damper opening adjustment delay parameter of 0.2-0.8 seconds. The values ​​of each delay parameter are obtained by dynamic calibration through analysis and monitoring of historical actuator status signals collected by the monitoring module.

[0028] The response quantization module continuously updates the actuator delay parameters through a dynamic calibration mechanism to maintain model accuracy. The module first receives historical actuator status signal data packets transmitted by the monitoring module. These packets include the fan's speed ramp-up curve during startup, the purification unit's electric field establishment current curve, and the damper opening adjustment sequence. The data parsing unit extracts the fan's acceleration response time from command reception to a stable speed of 90% of the rated value, the purification unit's current setup time from low-voltage standby to high-voltage rated power output, and the damper's position adjustment time from closed to the target opening.

[0029] Historical data feature analysis employs a sliding window algorithm, with a window period set to a continuous 24-hour operating period. Within each window, a sample set of fan acceleration response times is statistically analyzed, and the mean and standard deviation are calculated. Simultaneously, a sample set of current establishment times from the purification unit is processed to identify the power ramp-up rate distribution. The sample set of damper adjustment times is then correlated with pipeline pressure fluctuations and normalized. The analysis results output the fan mechanical acceleration delay parameter range as 2-5 seconds, the purification unit power ramp-up delay parameter range as 0.5-1 seconds, and the damper opening adjustment delay parameter range as 0.2-0.8 seconds.

[0030] The dynamic calibration trigger mechanism sets a deviation fluctuation threshold. When the fan acceleration response time in the newly acquired actuator status signal exceeds the standard deviation of the current parameter range by 15% three times consecutively, or when the current settling time distribution of the purification unit deviates, the parameter recalibration program is activated. The recalibration program extracts the most recent 72 hours of valid operating data, removes outliers using the quartile method, recalculates the parameter range, and updates the storage unit of the response quantization module.

[0031] The parameter update results are synchronized to the offset calculation engine of the timing compensation module. The updated fan mechanical acceleration delay parameters are used to correct the baseline value of the lag offset of the acceleration command, the purification unit power ramp-up delay parameters are associated with the high-pressure activation command advance compensation coefficient, and the damper opening adjustment delay parameters are mapped to the interlock offset response gradient. The dynamic calibration mechanism of delay parameters overcomes the control inaccuracy problems caused by equipment aging and operating condition drift, and maintains the stability of timing coordinated control.

[0032] Specifically, in the dynamic control system for restaurant fume emissions described in this invention, the timing compensation module is configured as follows: When identifying a surge in oil fume concentration, based on the upper limit of 0.5-1 seconds for the power ramp-up delay parameter of the purification unit in the response quantification module, the advance offset value of the purification unit command is set to 0.8 seconds; Based on the lower limit of the wind turbine mechanical acceleration delay parameter of 2-5 seconds, the wind turbine command lag offset value is set to 2 seconds.

[0033] The timing compensation module performs dynamic command offset correction for surges in oil fume concentration. This module monitors the rate of change of the oil fume concentration gradient in real time. When a surge event with an instantaneous change rate exceeding 15% per second is detected, the offset calculation program is activated. The program calls the physical response parameter database stored in the response quantization module and extracts the upper limit of the purification unit's power ramp-up delay parameter (1 second) as a baseline reference.

[0034] The command advance offset value setting for the purification unit adopts a safety margin design principle. Based on the maximum response delay of 1 second for the purification unit, multiplied by an 80% advance compensation coefficient, a command trigger advance of 0.8 seconds is generated. This offset ensures that the high-voltage electric field activation command is issued 800 milliseconds before the theoretical peak oil fume is reached, compensating for the power ramp-up time required for the electrostatic field to establish an ionization field, thus preventing oil fume escape.

[0035] The fan speed-up command lag offset is set based on the most conservative operating condition. A minimum mechanical acceleration delay parameter of 2 seconds is selected as the baseline, ensuring the speed-up command is triggered 2000 milliseconds after a surge in oil fume concentration. This lag design avoids the initial turbulent airflow period in the duct, allowing the speed to increase only after the air pressure stabilizes, thus preventing high-speed airflow from interfering with electrostatic adsorption efficiency.

[0036] After the offset parameters are generated, a synchronous trigger instruction structure is embedded. The timing compensation module outputs an instruction sequence carrying timestamps: the purification unit activation instruction time = event detection time - 800 milliseconds, and the fan speed-up instruction time = event detection time + 2000 milliseconds. The timestamp difference dynamically corrects the phase difference between the fan and the purification unit's actions, achieving precise alignment of the physical execution effects in the time dimension.

[0037] A self-checking mechanism is established during the offset parameter execution process. After each 24-hour operation cycle, the module statistically analyzes the deviation samples between the command trigger time and the actual action start time. When the purification unit start deviation continues to exceed ±100 milliseconds or the fan response lag exceeds ±300 milliseconds, the parameter calibration procedure is triggered. The calibration procedure, in conjunction with the response quantization module, updates the delay parameter range values ​​to maintain offset compensation accuracy.

[0038] Specifically, the dynamic control system for restaurant fume emissions described in this invention includes a collaborative control module comprising multi-level decision-making units. This module responds to a stir-frying condition event defined by the instantaneous change rate of fume concentration exceeding 15% per second, as output by the dynamic modeling module. The response includes: The first stage outputs commands to increase negative pressure in the fan and adjust the opening of the damper. The second stage uses a closed-loop correction module to detect the fan output pressure in real time. When the pressure value reaches 90% of the theoretical pressure value of the physical dynamic response model in the response quantification module, a high-voltage electric field activation command for the purification unit is output.

[0039] The multi-level decision-making units of the collaborative control module respond to stir-frying conditions by executing phased control strategies. The stir-frying condition is determined based on the real-time output of the instantaneous change rate of oil fume concentration from the dynamic modeling module. When this value continuously exceeds a threshold of 15% per second and remains above 300 milliseconds, an event flag is triggered. The event determination result is input into the multi-level decision-making units to activate the control sequence.

[0040] The first stage of control commands synchronously outputs a fan negative pressure increase command and a valve opening adjustment command. The fan negative pressure increase command drives the motor speed to rise to more than 85% of the rated value within 500 milliseconds, while the valve opening adjustment command adjusts the valve opening angle to 80%-90% of the maximum stroke. The two commands are simultaneously issued through the hardware driver interface to establish the initial negative pressure duct environment.

[0041] The second phase initiates the wind pressure compliance detection mechanism. The closed-loop calibration module collects data from the wind turbine outlet pressure sensor in real time, with a sampling frequency set to 100 milliseconds / time. The collected data is compared with the theoretical wind pressure value output by the physical dynamic response model in the response quantification module. When three consecutive sampling values ​​reach 90% of the theoretical wind pressure value threshold, the wind pressure stability condition is determined to be met.

[0042] The high-voltage electric field activation command of the purification unit is output immediately after the wind pressure stabilizes. The command transmission uses a priority interrupt channel to ensure that the high-voltage power supply starts up within 50 milliseconds. The electric field activation delay control maintains an interval of 2000-2500 milliseconds with the start time of the first-stage command. This interval matches the lower limit of the fan mechanical acceleration delay parameter to achieve synchronization of the establishment timing of the wind speed field and the ionization field.

[0043] A multi-level decision-making unit is configured with an execution timeout protection mechanism. Starting from the issuance of the first-stage command, if no wind pressure compliance signal is detected within 2000 milliseconds, a forced output of a medium-pressure preheating command for the purification unit is implemented as a degradation solution. Simultaneously, the model update module is triggered to mark abnormal operating conditions for subsequent parameter optimization analysis.

[0044] The control sequence termination condition is linked to the monitoring of the oil fume concentration gradient. When the dynamic modeling module detects that the concentration change rate has fallen below 5% per second and remains so for 1000 milliseconds, the multi-level decision unit outputs a command reset sequence. The fan speed is reduced to standby mode in stages, the purification unit enters low-pressure maintenance mode, and the damper opening is restored to the reference position.

[0045] Specifically, the dynamic control system for restaurant fume emissions described in this invention adds a simulation verification unit to the collaborative control module to pre-simulate the output command sequence of multi-level decision-making units; The simulation results showed that the oil fume adsorption efficiency was lower than the set standard value, which triggered the genetic algorithm optimization unit to adjust the branch weights of the multi-level decision units. The optimized decision weights are fed back to the multi-level decision-making units to update their control strategy execution rules.

[0046] The simulation verification unit of the collaborative control module performs a virtual pre-run before the actual issuance of control commands. This unit loads the command sequence generated by the multi-level decision-making unit, including the fan acceleration time point, the valve opening change curve, and the purification unit activation sequence. The pre-run environment imports the physical dynamic response model parameters from the response quantification module and combines them with the oil fume concentration distribution prediction data output by the dynamic modeling module to construct a three-dimensional fluid dynamics virtual scene.

[0047] The simulation process calculates the trajectory of oil fume particles and their charged adsorption efficiency within the air duct. The particle trajectory simulation employs a Lagrange discrete phase model to calculate the capture rate of oil droplets by the high-voltage electrostatic field. Adsorption efficiency is evaluated based on the coupling effect between the electric field intensity distribution and the airflow velocity field. The simulation outputs a quantitative value of the oil fume adsorption efficiency, which is then compared with a preset standard value of 85%.

[0048] When the quantified adsorption efficiency value falls below the standard threshold three times consecutively, the genetic algorithm optimization unit is triggered. The optimization unit uses the branch weights of the multi-level decision-making units as chromosome genes, including parameters such as the fan-valve command interval time and the purification unit activation delay. The fitness function is set as a comprehensive score of adsorption efficiency and energy consumption in the virtual scenario, and a new generation of weight combinations is generated iteratively through selection, crossover, and mutation operations.

[0049] The optimized decision weights are fed back to the strategy execution engine of the multi-level decision-making unit in real time. The weight update process adopts a hot-replacement mechanism: new weight parameters are loaded during the output intervals of the instruction sequence, overwriting the original control rules. The updated control strategy takes effect in the next stir-frying event, and the actual adsorption efficiency data is recorded simultaneously to verify the optimization effect.

[0050] A closed-loop verification system is established between the pre-simulation and actual execution data. After each operational event, the simulation verification unit reloads the actual executed command sequence and sensor feedback data, and performs a backtracking simulation in the same virtual environment. When the deviation between the backtracking result and the actual adsorption efficiency exceeds 8%, the genetic algorithm secondary optimization process is initiated, forming a continuous strategy improvement mechanism.

[0051] Specifically, in the dynamic control system for restaurant fume emissions described in this invention, the deviation detection operation of the closed-loop correction module is configured as follows: Calculate the absolute value of the difference between the actual response time of the fan speed and the delay value of the theoretical curve of the physical dynamic response model in the response quantization module; The standard deviation between the settling time of the electric field current in the detection purification unit and the theoretical ramp-up curve output by the timing compensation module is measured.

[0052] The deviation detection operation of the closed-loop correction module constructs an actuator response accuracy evaluation system. This module acquires pulse signals from the wind turbine speed sensor in real time through a high-speed data acquisition unit, recording the actual response time from the issuance of the acceleration command to the speed stabilizing within ±3% of the target value. This time value is aligned and compared with the theoretical curve output by the physical dynamic response model in the response quantification module. The theoretical curve includes characteristic parameters such as the acceleration slope and stabilization time threshold. The absolute value of the difference between the actual response time and the theoretical delay value is calculated as a quantification index of the wind turbine's mechanical inertia deviation.

[0053] The current establishment process of the purification unit's electric field is analyzed using millisecond-level sampling. The module acquires the current waveform at the output of the high-voltage power supply through a current transformer and extracts the establishment time series of the current rising from 10% to 90% of the rated value. This series is matched for similarity with the theoretical ramp-up curve issued by the timing compensation module. The theoretical curve is generated based on the purification unit's power ramp-up delay parameter and has a preset S-shaped rising trajectory. A sliding window is used to calculate the standard deviation between the actual sequence and the theoretical curve. The window width matches the electric field establishment period, and the standard deviation reflects the stability of the current ramp-up process.

[0054] The deviation quantification results are transmitted to the model update module, triggering a graded response mechanism. A level 1 deviation event is marked when the absolute value of the wind turbine response time difference exceeds 300 milliseconds, and a level 2 deviation event is marked when the standard deviation of the current build-up process exceeds the baseline value by 15%. The graded marking results are associated with data storage areas of different priorities for subsequent parameter optimization analysis and tracing.

[0055] The testing process incorporates a dynamic compensation calibration step. Every 24-hour operating cycle, the module calls upon the historical theoretical curve database for self-calibration testing. A standard test waveform is injected via a digital-analog signal generator to verify the measurement accuracy of the absolute difference and standard deviation calculation unit. If the calibration deviation exceeds the tolerance limit, the sensor calibration procedure is triggered to maintain the reliability of the test results.

[0056] A mapping relationship is established between deviation data and actuator operating conditions. The difference in fan response time is correlated with fan bearing temperature sensor data, and the standard deviation of current is correlated with the carbon deposition index of the purification unit electrodes. Multidimensional data fusion analysis locates the root cause of deviations, determining whether they are due to equipment aging or environmental interference, providing a basis for classification and optimization for model parameter updates.

[0057] Specifically, in the dynamic control system for catering fume emissions described in this invention, the model update module sends a parameter iteration instruction to the response quantization module when the standard deviation is greater than 5%. The response quantization module uses the gradient descent method to establish the time standard deviation based on the current detected by the closed-loop correction module, corrects the stored fan inertia coefficient and purification unit response order, and the updated parameters are synchronously fed back to the offset value calculation process of the timing compensation module.

[0058] The model update module triggers the parameter iteration process through real-time deviation analysis. This module receives the current settling time standard deviation data transmitted by the closed-loop correction module. When the average standard deviation of three consecutive monitoring cycles exceeds the 5% threshold, a parameter iteration command is generated. The command includes the deviation quantization value and the corresponding timestamp index, and is sent to the parameter correction interface of the response quantization module through a high-priority data channel.

[0059] The response quantization module initiates the gradient descent optimization engine. The algorithm uses the fan inertia coefficient and the purification unit response order as optimization variables, and minimizes the standard deviation of the current settling time as the objective function. The initial step size is set to 0.5% of the current parameter values. In each iteration, the partial derivative of the objective function with respect to the parameters is calculated, and the parameter values ​​are updated in the opposite direction of the gradient. The optimization process is set to a maximum of 20 iterations, and the calculation automatically terminates when the objective function descent rate is less than 0.1%.

[0060] The parameter correction results undergo a dual verification mechanism. The corrected fan inertia coefficient is first simulated using a virtual wind field model to verify the matching degree between the speed response curve and the theoretical model; the response order of the purification unit is then verified using a circuit simulation model to examine the electric field establishment time distribution characteristics. After successful verification, the parameter database of the storage unit is updated, and discarded parameters are transferred to the historical version cache.

[0061] The updated parameters are transmitted synchronously via a bidirectional channel. The primary channel sends a parameter update flag to the offset calculation engine of the timing compensation module, triggering the offset recalculation procedure; the secondary channel writes the parameter change log to non-volatile memory, recording the version number, effective time, and associated deviation event code. The parameter version management mechanism supports rapid rollback operations in abnormal situations.

[0062] The offset value recalculation adopts an incremental update strategy. After receiving the new parameter version flag, the timing compensation module extracts the currently processed real-time operating condition data packet. Based on the updated fan inertia coefficient, it regenerates the acceleration command lag compensation formula and corrects the electric field activation advance calculation matrix according to the new purification unit response order. The calculation results immediately overwrite the original offset value cache, enabling it to take effect in the next control cycle.

[0063] A post-evaluation mechanism for parameter iteration performance is established. Within 24 hours of each update taking effect, the model update module continuously monitors the changing trend of the current settling time standard deviation. When the standard deviation decrease rate fails to reach the expected target of 30%, a secondary optimization process is triggered, automatically adjusting the step size parameters and iteration termination conditions of the gradient descent method to improve the effectiveness of parameter correction.

[0064] Specifically, the dynamic control system for restaurant fume emissions described in this invention uses multi-level decision-making units to determine frying conditions based on the continuous low-amplitude concentration fluctuation characteristics output by the dynamic modeling module, and then executes an operation sequence: The initial instruction set consists of power preheating instruction for the purification unit, half-open instruction for the air valve, and low-speed operation instruction for the fan. When the rate of change of oil fume concentration gradient exceeds 10% per second, the collaborative control module is triggered to output a fully open interlock command for the air valve.

[0065] Multi-level decision-making units implement a progressive control strategy for frying conditions. The frying condition is determined based on the characteristics of the oil fume concentration data output by the dynamic modeling module. When the concentration fluctuation range remains within ±15% of the baseline value and the fluctuation frequency is less than twice per minute, it is determined to be a continuous low-amplitude concentration fluctuation mode. After this mode lasts for more than 300 seconds, the frying condition event flag is activated, triggering the issuance of the initial instruction set.

[0066] The initial instruction set is configured for energy-saving operation. In the purification unit, the power preheating instruction controls the high-voltage power supply to output 50%-60% of its rated voltage, establishing a weak ionization field between the electrodes without activating the adsorption function; the half-open valve instruction fixes the valve opening to 40%-60% of its maximum stroke, maintaining basic ventilation; and the low-speed fan operation instruction drives the motor speed to remain at 30%-40% of its rated value, generating a low-speed airflow field. These three instructions are issued synchronously through a parallel channel, establishing a low-energy standby state.

[0067] The oil fume concentration gradient abrupt change monitoring uses a dynamic threshold. The monitoring module calculates the concentration change rate in real time, with a sampling period of 200 milliseconds. When the average change rate of five consecutive sampling points exceeds the 10% threshold per second, and the peak value reaches more than 15% per second, it is determined to be an oil fume outbreak event caused by a sudden rise in oil temperature. The event signal is transmitted to the priority processing unit of the collaborative control module via an interrupt request.

[0068] The full-open interlock command for the damper has a millisecond-level response. Upon receiving a sudden event signal, the collaborative control module sends a full-open command to the damper actuator within 50 milliseconds. The command includes the target opening degree of 100% and the maximum opening rate command, driving the valve to complete its stroke within 300-500 milliseconds. During the interlock operation, the power preheating state in the purification unit remains unchanged, and the fan speed is increased in steps to 60% of the rated value to avoid interference from sudden airflow changes.

[0069] The interlock termination mechanism is linked to concentration gradient decline monitoring. When the rate of change remains below the 5% threshold for 30 seconds, the collaborative control module outputs a reset sequence. The damper opening returns to the half-open reference position within 10 seconds, the fan speed drops to the initial low speed state, and the purification unit maintains preheating standby. The reset process records valve action delay data, which is used to update the damper adjustment delay parameters of the response quantification module.

[0070] Safety interlock logic is implemented during the switching process. When switching from frying to stir-frying, the high-voltage electric field is only activated after the air valve has been forcibly reset. The interlock signal is implemented through hardware logic circuitry to prevent electrode contamination caused by the high-voltage start-up of the purification unit before the air valve is in the correct position.

[0071] Specifically, the dynamic control system for catering fume emissions described in this invention has a timing compensation module embedded in a feedforward controller, whose input port receives outdoor temperature disturbance data and stove thermal imaging activity index. The feedforward controller outputs a compensation gain factor, which is used to calculate the offset value of the timing compensation module, dynamically correcting the linkage offset between the fan and the damper.

[0072] The feedforward controller of the timing compensation module establishes an environmental disturbance compensation mechanism. This controller receives outdoor temperature data transmitted from the building management system via a Modbus protocol interface, with a temperature sampling frequency of once per minute. Simultaneously, it accesses the stove thermal imaging activity index output from an infrared thermal imager. This index quantifies the area ratio of the high-temperature zone on the stove and the integral value of thermal radiation intensity, with an update cycle of 2 frames per second. The environmental data packets are filtered and denoised before being input into the feedforward compensation model.

[0073] The feedforward compensation model is constructed using a multiple linear regression algorithm. Outdoor temperature disturbance data is mapped to an air density compensation coefficient, with each 10°C increase in temperature corresponding to a 0.8%-1.2% increase in duct resistance loss. The thermal imaging activity index is converted into a predicted value for the rate of oil fume generation, with each 10% increase in activity corresponding to a 0.3% / second increase in the baseline value of the oil fume concentration gradient. The model outputs a compensation gain factor matrix, including a fan pressure loss correction coefficient K1 and a valve flow rate compensation coefficient K2.

[0074] The core algorithm for calculating the offset value is embedded in the compensation gain factor. The timing compensation module multiplies the K1 factor by the original fan command lag offset to generate the temperature-compensated lag offset value; it then adds the K2 factor to the flow regulation term of the valve interlock offset to form the linkage offset parameter for heat load adaptation. The strength of the gain factor is positively correlated with the disturbance amplitude. When the outdoor temperature change exceeds 5℃ / minute or the thermal imaging activity fluctuation exceeds 15% / second, the enhanced compensation mode is activated.

[0075] The dynamic correction process includes boundary protection conditions. The fan linkage offset correction range is limited to ±30% of the initial value to avoid overcompensation leading to operational timing disorder. During valve offset correction, pipeline pressure sensor data is monitored simultaneously; correction is paused and system diagnostics are triggered when pressure fluctuations exceed the baseline value by 20%. Correction parameter version numbers are linked to environmental data timestamps, supporting backtracking analysis of control effects.

[0076] The compensation effectiveness verification employs a dual-channel comparison mechanism. Every 8-hour operating cycle, the feedforward controller switches to bypass mode for 1 minute, recording the offset execution data under uncompensated conditions. This data is then compared with the fan response time and valve positioning accuracy under compensated mode. If the deviation consistently exceeds 15%, a self-learning process for the feedforward model parameters is triggered. This process calls upon the environmental data and offset execution records from the past 72 hours to update the regression algorithm's weight coefficients.

[0077] Specifically, the dynamic control system for restaurant fume emissions described in this invention has a dual-redundancy architecture for the synchronous trigger command transmission channel, with the main channel using the RS485 bus protocol and the backup channel using the CAN bus protocol. The collaborative control module monitors the packet loss rate of the main channel in real time. When the packet loss rate exceeds 5%, it triggers seamless switching of the backup channel to maintain the integrity of the transmission of offset parameter commands.

[0078] The synchronous trigger command transmission channel employs a primary and backup dual-channel redundancy architecture to ensure reliability. The primary channel uses the RS485 bus protocol to build the physical layer transmission link, with a communication baud rate set to 115200bps. The data frame structure includes a command type code, offset parameter value, and CRC check field. The backup channel uses the CAN bus protocol, with a baud rate configured at 250kbps, and the data frame identifier maps to the command type code. The dual-channel hardware interfaces are connected in parallel to the command output port of the collaborative control module.

[0079] The packet loss rate detection mechanism uses a sliding time window for statistical analysis. The collaborative control module calculates the number of lost packets based on the receiving end's confirmation signal every 10 command data packets sent. The statistical window covers the most recent 100 command transmission cycles. If the packet loss rate consistently exceeds the 5% threshold within the window and remains above it for three consecutive statistical cycles, a channel switching condition is triggered. A packet loss rate exceeding the limit event generates a channel fault flag and writes it to the system log.

[0080] The seamless handover process executes a four-stage protocol conversion. The first stage suspends the transmission of new commands on the primary channel and buffers the command queue. The second stage sends a link activation request to the backup channel to establish a handshake connection. The third stage converts the buffered commands on the primary channel into CAN protocol frame format for transmission. The fourth stage sends a handover confirmation signal back to the collaborative control module, updating the channel status flags. The entire handover process is completed within 50 milliseconds.

[0081] Transmission integrity is ensured using instruction sequence number tracking technology. Each instruction is embedded with a globally unique, incrementing sequence number, and the sequence number count continues from the primary channel after the backup channel is activated. The receiving end detects missing instructions by checking the sequence number continuity, and initiates a retransmission request to the coordination control module when a sequence number gap is detected. The retransmission mechanism covers the instruction buffer interval during the handover process, preventing instruction loss.

[0082] The channel recovery mechanism is based on periodic automatic diagnostics. During standby channel operation, the system sends diagnostic test frames to the primary channel every minute. When 10 consecutive test frames respond normally and the packet loss rate drops below 1%, the reverse handover process is initiated. The reverse handover uses the same four-stage protocol conversion. After the handover is completed, the packet loss rate statistics window is reset, and the primary channel transmission priority is restored.

[0083] The dual-channel physical layer features isolation protection circuitry. The RS485 interface chip is equipped with surge suppression and electrostatic discharge (ESD) protection modules, and an EMC filter is added to the CAN bus terminal. Electrically isolated optocouplers provide potential isolation between the collaborative control module and the channel interface, preventing ground loop interference from affecting transmission stability. A fault in the isolation circuit triggers a hardware alarm and forces the activation of the backup channel.

[0084] The algorithms and models involved in the technical solution of this invention are explained and described as follows: The time series analysis algorithm of the dynamic modeling module processes the data streams of oil fume concentration and temperature and humidity through a sliding window (30-second period), and uses an autoregressive integral moving average model to identify the periodicity of concentration fluctuations and trend inflection points. Based on the gradient change rate of the monitoring data from the previous 60 seconds (e.g., the instantaneous change rate of concentration) and the influence coefficient of ambient temperature and humidity on oil fume diffusion, the algorithm predicts the emission peak and duration for the next 5-10 seconds. Those skilled in the art will understand that this predicted value is used to calculate the lead time for triggering control timing, rather than for precise numerical matching.

[0085] The parameter calibration mechanism of the response quantification module adopts a dynamic calibration method combining historical data analysis and real-time feedback. The module has a built-in statistical analysis unit that continuously collects a sample set of acceleration times (e.g., 100 sets of historical data) from the time the fan receives a command to the time it reaches 90% of its rated speed. After removing outliers using the quartile method, the confidence interval for the delay parameters (2-5 seconds for the fan) is calculated. The calibration of the purification unit's power ramp-up delay is based on the established time series of the electric field current rising from 10% to 90% of its rated value. Moving average filtering is used to eliminate noise interference, generating a baseline parameter of 0.5-1 seconds. This process is essentially a digital representation of the equipment's physical response characteristics, and the parameter values ​​are automatically updated as the equipment ages.

[0086] The offset calculation algorithm of the timing compensation module converts physical delay parameters into executable time corrections. For events involving a surge in oil fume concentration, a 0.8-second command advance (with a 20% safety margin) is set based on the maximum 1-second delay of the purification unit, and a 2-second lag offset corresponds to the minimum 2-second delay of the fan, forming a staggered triggering strategy for purification unit activation and fan acceleration. Those skilled in the art will understand that this staggered design ensures that the completion of the electric field establishment coincides with the generation of the fan's maximum negative pressure on the time axis, eliminating the control window caused by differences in response speed.

[0087] The multi-level decision-making logic of the collaborative control module adopts an event-driven state machine model. During the stir-frying operation, the synchronous output of the first-stage fan speed-up and damper adjustment commands is implemented based on a hardware parallel drive interface. The trigger condition for the second-stage purification unit activation command is set to "the actual fan pressure reaches 90% of the theoretical value after three consecutive samplings." This threshold is determined through fluid simulation and measured data verification, ensuring that the electric field is activated after the air pressure stabilizes. The progressive control during the frying operation triggers the damper interlock action through an event threshold where the concentration gradient change rate exceeds 10% per second. This threshold is set based on the typical characteristics of a sudden rise in oil temperature leading to an explosion of fumes.

[0088] The deviation detection algorithm of the closed-loop correction module performs a time-domain comparison between the theoretical model and the actual response. The fan response time deviation is calculated using the absolute difference between the actual acceleration time and the theoretical value of the physical dynamic response model. The standard deviation of the current ramp-up curve of the purification unit is calculated using a sliding window (the window width matches the electric field establishment period) to determine the deviation between the actual sequence and the theoretical S-shaped curve. When the standard deviation consistently exceeds 5%, it is determined to be a model parameter mismatch, triggering the iteration process.

[0089] The parameter iteration mechanism of the model update module employs gradient descent to optimize the physical response parameters. Using the standard deviation of the current settling time of the purification unit as the objective function, partial derivatives are calculated with respect to the fan inertia coefficient and the order of the purification unit response. Parameter values ​​are then adjusted in 0.5% steps along the opposite direction of the gradient. The iteration process is validated by both a virtual wind field model (to calculate the speed response matching degree) and circuit simulation (to verify the electric field settling characteristics). Those skilled in the art will understand that this process effectively corrects the mathematical representation accuracy of the physical model through feedback data.

[0090] The feedforward controller's environmental compensation model establishes a mapping relationship between external disturbances and offsets. A linear relationship exists between the wind pressure loss compensation coefficient (0.8%-1.2%) and the actual wind pressure loss for every 10°C increase in outdoor temperature, obtained through regression analysis of historical operating data. The correlation between the stove's thermal imaging activity index and the rate of oil fume generation is determined based on correlation analysis between infrared thermal imager data and concentration gradients. Output compensation factors dynamically correct temporal offsets; for example, increasing the fan lag offset in high-temperature environments to compensate for acceleration delays caused by reduced air density.

[0091] The genetic algorithm optimization unit uses the weights of multi-level decision branches as gene codes (such as the instruction interval time between the fan and the purification unit), and the oil fume adsorption efficiency output by the virtual simulation as the fitness function. New weight combinations are generated through selection, crossover, and mutation operations. This process is essentially a self-optimization of the control strategy in a digital twin environment. After optimization, the weights directly update the execution rules of the multi-level decision units.

[0092] The dual-channel communication redundancy mechanism achieves seamless switching through global tracking of command sequence numbers. The determination of a main channel packet loss rate exceeding 5% is based on the sliding window statistics of the most recent 100 command transmission cycles. During switching, the command sequence number continues in the CAN channel. The receiving end detects gaps according to the sequence number and triggers retransmission, ensuring the integrity and timing consistency of the offset parameter commands.

[0093] The dynamic control system for restaurant fume emissions provided by this invention addresses the timing mismatch problem caused by differences in the physical response times of fans, purification units, and air valves in restaurant kitchens, achieving precise coordinated control through the following technical solution: The system hardware deployment includes oil fume concentration sensors (such as laser particle counters), environmental parameter sensors (temperature, humidity, and wind pressure sensors), and actuator status feedback units (fan speed encoders, purification unit current transformers, and damper opening position sensors). The monitoring module collects oil fume concentration signals (sampling frequency ≥10Hz), ambient temperature and humidity, and duct pressure data in real time, and simultaneously acquires actuator status signals such as fan start / stop status, purification unit operating mode, and damper opening position. Data is transmitted to the central controller via a dual-redundant RS485 / CAN bus. When the main channel packet loss rate exceeds 5%, the backup channel is automatically switched to ensure command integrity.

[0094] The dynamic modeling module uses time series analysis to process monitoring data, identifying the trend of oil fume concentration fluctuations based on a sliding window algorithm (30-second window period) and predicting the emission peak in the next 5-10 seconds. The response quantification module has a built-in historical data analysis unit. Based on the fan acceleration curve (time taken to reach 0-90% of rated speed), the purification unit current ramp-up curve (time to establish high-voltage rated current from low-voltage standby), and the damper position change sequence, it dynamically calibrates the following delay parameters: fan mechanical acceleration delay 2-5 seconds, purification unit power ramp-up delay 0.5-1 second, and damper opening adjustment delay 0.2-0.8 seconds. The parameter update condition is that three consecutive measured delay values ​​deviate from the current range by ±15%.

[0095] The time-series compensation module receives emission trend predictions and a physical response model. When a stir-frying condition with an instantaneous change rate exceeding 15% per second is identified: Purification unit command advance offset: Based on the maximum delay of 1 second for the purification unit, the command is set to be triggered 0.8 seconds in advance, so that the high-voltage electric field establishes an ionization field before the peak of oil fume; Fan command lag offset: Based on the minimum fan delay of 2 seconds, the speed-up command is set to be triggered with a 2-second lag to avoid the initial airflow turbulence period; Air valve interlock offset: The opening adjustment advance is dynamically calculated based on real-time air duct pressure data to match changes in air speed.

[0096] The multi-level decision-making units of the collaborative control module execute in stages: Phase 1: Synchronously output commands to increase the fan speed to 85% of the rated speed and to adjust the damper opening to 80%-90%; The second stage involves real-time monitoring of the fan outlet air pressure via a closed-loop calibration module. When the air pressure reaches 90% of the theoretical value (after three consecutive successful samplings), a high-pressure activation command for the purification unit is immediately triggered.

[0097] The closed-loop correction module collects the actual fan speed response time, the purification unit current establishment curve, and the damper position feedback, and compares them with the theoretical values ​​of the physical dynamic response model. When the absolute value of the wind turbine response time deviation exceeds 300 milliseconds, a Level 1 deviation event is marked. When the standard deviation of the current ramp-up curve of the purification unit exceeds 5%, a level 2 deviation event is marked.

[0098] For secondary deviation events, the model update module triggers parameter iteration in the response quantization module: the gradient descent method is used to correct the fan inertia coefficient and the purification unit response order. After the update, the parameters are synchronized to the timing compensation module to recalculate the offset value. The iteration results take effect after being verified by virtual wind field model and circuit simulation.

[0099] Frying condition control: When the dynamic modeling module identifies continuous low-amplitude concentration fluctuations (fluctuation range ±15%), the multi-level decision unit outputs the initial instruction set for power preheating (50%-60% of rated voltage), half-open air valve (40%-60% opening degree), and low-speed operation of the fan (30%-40% of rated speed) in the purification unit; when the concentration gradient change rate is detected to exceed 10% per second, the air valve fully open interlock command is immediately triggered.

[0100] Feedforward Environmental Compensation: The timing compensation module is embedded in the feedforward controller, receiving outdoor temperature data (sampled every minute) and stove thermal imaging activity index (2 frames per second). For every 10°C increase in outdoor temperature, a wind pressure loss compensation coefficient (0.8%-1.2%) is generated; for every 10% increase in thermal activity, a fume generation rate correction factor (+0.3% / second) is generated, dynamically adjusting the fan / valve offset (correction range limited to ±30%).

[0101] Communication redundancy mechanism: The collaborative control module monitors the packet loss rate of the RS485 main channel in real time. If the packet loss rate exceeds 5%, it switches to the CAN backup channel within 50 milliseconds. The instruction transmission adopts a global sequence number tracking and retransmission mechanism to ensure integrity.

[0102] This invention solves the problem of mismatch in coordination among multiple actuators by physical response modeling, time offset compensation, multi-level decision control and closed-loop parameter iteration, thereby improving purification efficiency and reducing ineffective energy consumption. It is especially suitable for scenarios with a surge in transient oil fumes, such as stir-frying and deep-frying.

[0103] This invention addresses the timing mismatch problem caused by differences in actuator response speeds by constructing a physical response characteristic quantification and dynamic timing compensation mechanism. The system precisely calibrates the parameters of fan mechanical acceleration delay, purification unit power ramp-up delay, and damper opening adjustment delay in the response quantification module, forming a physical dynamic response model. The timing compensation module calculates the fan command lag offset, purification unit command advance offset, and damper interlock offset based on the model parameters and real-time operating condition changes. This offset design allows the purification unit's high-voltage electric field establishment process to start earlier, the fan acceleration action to be triggered later, and the damper opening / closing to be dynamically matched with airflow changes, achieving physical alignment of these three components in the time dimension and eliminating the phenomenon of oil fume escape or insufficient purification caused by differences in actuator response speeds.

[0104] The collaborative control module employs multi-level decision-making units to achieve precise orchestration of action sequences. Under high-pressure stir-frying conditions, the first stage simultaneously outputs commands to increase the negative pressure of the fan and adjust the opening of the damper, establishing a basic air pressure environment. The second stage uses a closed-loop correction module to monitor the fan output air pressure in real time, triggering a high-pressure activation command for the purification unit once it reaches 90% of the theoretical threshold of the physical dynamic response model. The action sequence, combined with the feedforward controller's environmental compensation mechanism, dynamically corrects the offset based on outdoor temperature disturbances and the activity index of the stove's thermal imaging. The simulation verification unit pre-runs the command sequence and evaluates the oil fume adsorption efficiency, triggering a genetic algorithm to optimize decision rules, thus forming a self-evolving control strategy.

[0105] The system establishes a dual optimization closed loop of execution effect feedback and parameter iteration. The closed-loop correction module collects real-time data on fan speed, electric field current, and valve position, comparing it with the theoretical output values ​​of the physical dynamic response model to identify action delay deviations and the standard deviation of the current ramp-up curve. When the deviation exceeds the tolerance threshold, the model update module triggers the gradient descent algorithm to correct the inertia coefficient and response order parameters of the response quantization module. The updated parameters are synchronously fed back to the offset calculation process of the timing compensation module, driving multi-level decision units to adjust the action sequence. Dual redundant command transmission channels ensure complete delivery of control commands, and the primary and backup channels achieve seamless switching within 50 milliseconds based on packet loss rate monitoring, maintaining the continuous reliability of timing-coordinated control.

Claims

1. A dynamic control system for restaurant fume emissions, characterized in that, include: The monitoring module collects signals of oil fume concentration, environmental parameters, and actuator status. The dynamic modeling module connects to the monitoring module, receives oil fume concentration signals and environmental parameter signals, and generates predicted values ​​for oil fume emission trends. The response quantization module receives the actuator status signal from the monitoring module, and combines it with the stored fan mechanical acceleration delay parameters, purification unit power ramp-up delay parameters, and damper opening adjustment delay parameters to generate a physical dynamic response model. The timing compensation module connects the dynamic modeling module and the response quantization module. It receives the predicted value of oil fume emission trend and the physical dynamic response model. Based on the real-time operating condition changes in the actuator status signal, it calculates the fan command lag offset value, the purification unit command advance offset value, and the valve interlock offset value. The collaborative control module, connected to the timing compensation module, outputs a synchronous trigger command carrying offset parameters to the actuator, aligning the fan speed-up action, the purification unit activation action, and the air valve opening and closing action in the time dimension. The closed-loop correction module receives feedback values ​​of fan speed, electric field current, and damper position, and extracts the actual response curve of the actuator. The model update module connects the closed-loop correction module and the response quantization module. It inputs the actual response curve of the actuator into the physical dynamic response model to obtain the theoretical output value. When the response deviation exceeds the tolerance threshold, it sends the updated delay parameter data to the response quantization module and the timing compensation module.

2. The dynamic control system for restaurant fume emissions according to claim 1, characterized in that, The response quantization module is configured such that the fan mechanical acceleration delay parameter is 2-5 seconds, the purification unit power ramp-up delay parameter is 0.5-1 seconds, and the damper opening adjustment delay parameter is 0.2-0.8 seconds.

3. The dynamic control system for restaurant fume emissions according to claim 2, characterized in that, The timing compensation module is configured as follows: When identifying a surge in oil fume concentration, based on the upper limit of 0.5-1 seconds for the power ramp-up delay parameter of the purification unit in the response quantification module, the advance offset value of the purification unit command is set to 0.8 seconds; Based on the lower limit of the wind turbine mechanical acceleration delay parameter of 2-5 seconds, the wind turbine command lag offset value is set to 2 seconds.

4. The dynamic control system for restaurant fume emissions according to claim 3, characterized in that, The collaborative control module includes multi-level decision-making units, which are used to respond to the stir-frying condition event defined by the instantaneous change rate of oil fume concentration exceeding 15% per second, as output by the dynamic modeling module. The first stage outputs commands to increase negative pressure in the fan and adjust the opening of the damper. The second stage uses a closed-loop correction module to detect the fan output pressure in real time. When the pressure value reaches 90% of the theoretical pressure value of the physical dynamic response model in the response quantification module, a high-voltage electric field activation command for the purification unit is output.

5. The dynamic control system for restaurant fume emissions according to claim 4, characterized in that, The collaborative control module includes a simulation verification unit, which is used to pre-simulate the output instruction sequence of the multi-level decision-making unit; The simulation results showed that the oil fume adsorption efficiency was lower than the set standard value, which triggered the genetic algorithm optimization unit to adjust the branch weights of the multi-level decision units. The optimized decision weights are fed back to the multi-level decision-making units to update their control strategy execution rules.

6. The dynamic control system for restaurant fume emissions according to claim 5, characterized in that, The deviation detection operation of the closed-loop correction module is configured as follows: Calculate the absolute value of the difference between the actual response time of the fan speed and the delay value of the theoretical curve of the physical dynamic response model in the response quantization module; The standard deviation between the settling time of the electric field current in the detection purification unit and the theoretical ramp-up curve output by the timing compensation module is measured.

7. The dynamic control system for restaurant fume emissions according to claim 6, characterized in that, It also includes the model update module sending parameter iteration instructions to the response quantization module when the standard deviation is greater than 5%; The response quantization module uses the gradient descent method to establish the time standard deviation based on the current detected by the closed-loop correction module, corrects the stored fan inertia coefficient and purification unit response order, and the updated parameters are synchronously fed back to the offset value calculation process of the timing compensation module.

8. The dynamic control system for restaurant fume emissions according to claim 7, characterized in that, Also includes: The multi-level decision-making unit determines the frying condition event based on the continuous low-amplitude concentration fluctuation characteristics output by the dynamic modeling module, and executes the operation sequence: The initial instruction set consists of power preheating instruction for the purification unit, half-open instruction for the air valve, and low-speed operation instruction for the fan. When the rate of change of oil fume concentration gradient exceeds 10% per second, the collaborative control module is triggered to output a fully open interlock command for the air valve.

9. The dynamic control system for restaurant fume emissions according to claim 8, characterized in that, Also includes: The timing compensation module is embedded in the feedforward controller, and its input port receives outdoor temperature disturbance data and stove thermal imaging activity index. The feedforward controller outputs a compensation gain factor, which is used to calculate the offset value of the timing compensation module, dynamically correcting the linkage offset between the fan and the damper.

10. The dynamic control system for restaurant fume emissions according to any one of claims 1-9, characterized in that, The synchronous trigger command transmission channel includes a dual-redundancy architecture, with the main channel using the RS485 bus protocol and the backup channel using the CAN bus protocol. The collaborative control module monitors the packet loss rate of the main channel in real time. When the packet loss rate exceeds 5%, it triggers seamless switching of the backup channel to maintain the integrity of the transmission of offset parameter commands.

Images