Energy absorption and discharge efficiency real-time optimization system and method based on multivariable frequency conversion technology

Abstract

The application provides an energy absorption and discharge efficiency real-time optimization system and method based on multivariate frequency conversion technology, and belongs to the field of multivariate frequency conversion technology. The method comprises the following steps: collecting in real time through a multi-source sensor, generating a state vector and a wind resistance particle index through synchronous processing; establishing a wind resistance prediction model based on the state vector and the wind resistance particle index, and outputting a current predicted wind resistance and an actual motor output torque; combining the current predicted wind resistance, the actual motor output torque and a temperature array to generate a two-dimensional heat zone atlas and dynamically mark a heat zone position needing intervention as a control sensitive zone; according to the two-dimensional heat zone atlas and the control sensitive zone, constructing an optimization objective function with the heat zone risk minimization, energy consumption minimization and control stability as targets, and coordinating output fan speed, torque and guide wind angle control instructions. The application is suitable for the energy absorption and discharge system scene with the following characteristics: intense oil fume fluctuation, complex fume exhaust space and sensitive wind resistance feedback delay.

Description

Technical Field

[0001] This invention belongs to the field of multivariable frequency conversion technology, and particularly relates to a real-time optimization system and method for intake and exhaust energy efficiency based on multivariable frequency conversion technology. Background Technology

[0002] With increasingly stringent requirements for energy efficiency, intelligence, and safety in kitchen exhaust systems, traditional suction and exhaust equipment relying on fixed-frequency motors and fixed control logic is struggling to meet the dynamic operational needs of complex kitchen environments, such as multiple stoves operating in parallel, high oil fume loads, and frequent heat flow disturbances. In actual use cases, oil fume particle size, duct resistance, fan load, and local temperature rise exhibit a strongly coupled nonlinear relationship, changing drastically with cooking activities. Most existing systems only use oil fume concentration as a reference for fan control, neglecting the dynamic interaction between particle size distribution, pressure difference, heat accumulation, and electromagnetic torque. This leads to problems such as fan response lag, motor overload, reduced exhaust efficiency, and uncontrolled heat diffusion in highly fluctuating and disturbed scenarios. Furthermore, although some systems have introduced variable-frequency motors and PID control mechanisms for energy-saving regulation, they still lack modeling mechanisms for the multiple physical states of the suction and exhaust process, lack online-updated motor-resistance load prediction capabilities, and fail to introduce thermal zone spatial feedback mechanisms and multi-objective optimization strategies at the control layer. Consequently, the overall system energy efficiency remains difficult to improve, and response speed and accuracy are insufficient.

[0003] To achieve true "on-demand output, structural adaptation, dynamic response, and multi-objective collaboration" in the suction and exhaust system, a new type of suction and exhaust control system with multi-variable state perception capability, online identification of coupling models, spatial thermal zone identification capability, and high real-time predictive control framework is urgently needed. Summary of the Invention

[0004] The purpose of this invention is to propose a real-time optimization system and method for intake and exhaust energy efficiency based on multivariable frequency conversion technology, in order to solve the above-mentioned problems.

[0005] To achieve the above objectives, a first aspect of the present invention provides an intelligent optimization method for an enterprise management system based on digital twins, the method comprising the following steps:

[0006] The average fan speed, average duct pressure difference, motor power sliding average, total particle size distribution, oil fume concentration sliding average, and duct temperature average are collected in real time by multi-source sensors. The data are then processed synchronously to generate a state vector and wind resistance particle index.

[0007] Based on the state vector and the wind resistance particle index, a wind resistance prediction model is established to express the nonlinear relationship between electromagnetic torque and wind resistance in the suction and exhaust system in real time, and output the current predicted wind resistance and the actual motor output torque.

[0008] Combining the current predicted wind resistance, actual motor output torque, and temperature array, a two-dimensional thermal map is generated and the locations of the thermal zones requiring intervention are dynamically marked as control sensitive areas.

[0009] Based on the two-dimensional thermal map and the control sensitive area, an optimization objective function is constructed with the goals of minimizing thermal risk, minimizing energy consumption, and ensuring control stability, and the output of control commands for fan speed, torque, and air guide angle are coordinated.

[0010] Furthermore, the drag particle index represents the drag particle index per unit speed, which is calculated based on the average pressure difference in the duct, the total concentration of particle size distribution, and the average speed of the fan, and is used to reflect the combined effect of pressure difference and particle size concentration on wind resistance.

[0011] Furthermore, the wind resistance prediction model calculates the current predicted wind resistance by weighted summation based on the preset resistance, combined with the average pressure difference in the duct, the total concentration of the particle size distribution, the average temperature of the duct, and the average speed of the fan, and the square of the first derivative of the wind resistance particle index.

[0012] Specifically, the actual motor output torque is calculated by combining the current predicted wind resistance, average fan speed, sliding average motor power, and predicted theoretical motor output power.

[0013] Furthermore, the entire operating logic of the wind resistance prediction model is deployed in the edge control board. The model parameters are updated online using a recursive least squares algorithm. The perturbation derivative based on the wind resistance particle index is obtained through sliding difference. The regularization term for predicting the theoretical output power of the motor is updated once per cycle by the main control logic. The wind resistance prediction model can iterate twice per second.

[0014] Furthermore, the process of combining the current predicted wind resistance, actual motor output torque, and temperature array to generate a two-dimensional thermal map and dynamically marking the locations of the thermal zones requiring intervention as control sensitive areas specifically includes:

[0015] Spatial local difference and time derivative processing is performed on the temperature array to obtain the degree of heat accumulation per unit time in each region;

[0016] Based on the current predicted wind resistance and the actual motor output torque, a nonlinear wind resistance amplification factor is introduced to calculate the two-dimensional thermal map.

[0017] The control sensitive area is extracted based on the two-dimensional thermal map; wherein, the sensitive area is the set of all points exceeding the dynamic threshold;

[0018] The control sensitive area represents the location of the area that currently requires key risk control intervention.

[0019] Furthermore, if the control sensitive area is actually used, it will be mapped to the area corresponding to the stove layout and transformed into a secondary air duct airflow angle adjustment suggestion or an auxiliary target wind speed enhancement area.

[0020] Furthermore, based on the two-dimensional thermal map and the control sensitive area, an optimization objective function is constructed with the goals of minimizing thermal risk, minimizing energy consumption, and ensuring control stability. This function coordinates the output of control commands for fan speed, torque, and airflow angle. Specifically, this includes:

[0021] For the prediction of future windows, the fan speed, motor output torque and auxiliary air duct angle are used as the control variables.

[0022] The design optimizes the control variable set using the objective functions of minimizing total energy consumption, minimizing the risk of the two-dimensional thermal map, and smoothing the controller response. At the same time, the controller is designed with physical constraints to reflect the hardware boundaries of the actual equipment and outputs control commands for fan speed, torque, and air guide angle.

[0023] Furthermore, the physical constraints are that the fan speed is within the upper and lower limits of the equipment's safe wind speed, and the maximum response rate of the auxiliary air duct angle cannot exceed a preset threshold.

[0024] Furthermore, the controller runs an MPC solution every 250ms and updates it in real time using a rolling window; the optimization process is implemented using a single-layer QP structure.

[0025] A second aspect of the present invention provides a real-time energy efficiency optimization system for intake and exhaust based on multivariable frequency conversion technology, the system comprising:

[0026] The sensor acquisition module is used to collect data in real time from multiple sources, including the average speed of the fan, the average pressure difference in the duct, the sliding average of the motor power, the total concentration of the particle size distribution, the sliding average of the oil fume concentration, and the average temperature of the duct. After synchronous processing, a state vector and a wind resistance particle index are generated.

[0027] The online modeling module is used to establish a wind resistance prediction model based on the state vector and wind resistance particle index, so as to express the nonlinear relationship between electromagnetic torque and wind resistance in the suction and exhaust system in real time, and output the current predicted wind resistance and the actual motor output torque.

[0028] The thermal zone analysis module is used to combine the current predicted wind resistance, actual motor output torque and temperature array to generate a two-dimensional thermal zone map and dynamically mark the locations of the thermal zones that need intervention as control sensitive zones.

[0029] The predictive control module is used to construct an optimization objective function based on the two-dimensional thermal map and the control sensitive area, with the goals of minimizing thermal risk, minimizing energy consumption, and ensuring control stability, and to coordinate the output of control commands for fan speed, torque, and air guide angle.

[0030] The beneficial technical effects of the present invention are at least as follows:

[0031] The invention provides a real-time optimization system and method for suction and exhaust energy efficiency based on multivariable frequency conversion technology. It constructs a closed-loop optimization structure from multi-dimensional state acquisition, online modeling of coupled models, thermal zone distribution identification to multi-objective model predictive controller output. It can realize dynamic, precise and coordinated energy efficiency control strategies for typical problems such as high disturbance, high particle size and high heat local accumulation in kitchen suction and exhaust conditions. The system first constructs a unified state vector by integrating real-time sensor data from multiple sources, including fan speed, duct pressure difference, particle size distribution, oil fume concentration, and temperature distribution. Combined with a particle size-drag index construction mechanism based on a sliding window, this effectively improves the dimensionality and predictability of the state representation. Second, based on this state vector and particle size trend term, an online identification model for wind resistance and electromagnetic torque is constructed. The model's ability to suppress rapid operating condition switching and misjudgment scenarios is enhanced by using disturbance derivative term and energy consumption regularization term. Furthermore, by fusing predicted wind resistance, motor load, and thermal array data, a real-time updatable two-dimensional thermal map is constructed. Wind resistance-heat flow coupling term and nonlinear risk amplification factor are introduced to achieve accurate extraction of control-sensitive areas. Finally, by establishing an MPC controller with the goals of minimizing energy consumption, minimizing thermal risk, and ensuring control stability, the system achieves coordinated control of three variables—wind speed, torque, and airflow angle—under local spatial disturbance conditions, forming a complete prediction-identification-execution closed-loop control path. The hierarchical modeling, thermal zone identification, and multi-objective predictive control mechanism proposed in this invention has a clear deployment structure, physical interpretability, and improved control accuracy. It is particularly suitable for smoke extraction and exhaust systems with severe oil fume fluctuations, complex exhaust spaces, and sensitive wind resistance feedback delays. Attached Figure Description

[0032] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0033] Figure 1 This is a flowchart of the real-time optimization method for intake and exhaust energy efficiency based on multivariable frequency conversion technology according to the present invention. Detailed Implementation

[0034] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0035] like Figure 1 As shown in the figure, the real-time optimization method for intake and exhaust energy efficiency based on multivariable frequency conversion technology provided in this embodiment of the invention includes:

[0036] S1. Real-time data collection of fan average speed, duct average pressure difference, motor power sliding average, total particle size distribution, oil fume concentration sliding average, and duct temperature average are obtained through multi-source sensors. These data are then processed synchronously to generate a state vector and wind resistance particle index.

[0037] Specifically, this step aims to construct a multivariable state space that reflects the actual operating state of the kitchen exhaust system, serving as the input basis for subsequent modeling and predictive control. Because the operating conditions of the kitchen exhaust environment change drastically—instantaneous actions such as turning on the stove, putting food in a pan, and stir-frying significantly affect the heat flow structure, wind resistance, and motor load—a state vector with high temporal resolution and structural integrity for key physical variables needs to be established. Furthermore, this state space must rely solely on engineering-featured sensing methods, low-computational-resource processing mechanisms, and online feature processing methods that can be deployed on edge devices (such as the range hood's main control board or fan controller).

[0038] The primary data source for this step is the various sensors deployed in the suction and exhaust system. Their installation locations and sampling methods are based on current kitchen appliance hardware standards. The following are the sources and processing logic for each input variable:

[0039] Current fan speed The speed signal is obtained by a Hall magnetic encoder that is mounted coaxially with the motor spindle. The encoder has a resolution of 600 lines and is converted into a speed signal by the controller counting method. The update frequency is 10Hz.

[0040] Duct pressure difference The pressure difference is measured by two sets of MEMS differential pressure sensors embedded at both ends of the air duct. Taking a certain model as an example (such as the Honeywell HSC series), its measurement range can reach ±500Pa, the update cycle is 100ms, and the pressure difference value is read through the analog-to-digital conversion channel of the main controller.

[0041] motor power The power value is estimated based on the current and voltage sampling modules in the inverter (i.e., direct sampling of drive current and bus voltage). The power value is generated by the controller by calculating the voltage-current product and filtering. The update frequency is synchronized with the PWM carrier (typically 10kHz, but this step uses the 1Hz period average).

[0042] Oil fume particle size distribution An infrared particle light scattering sensor module (such as Plantower PMS5003) deployed directly in front of the air intake provides particle concentrations in six particle size ranges: PM0.3, PM0.5, PM1.0, PM2.5, PM5.0, and PM10.0. It updates once per second and outputs a 6-dimensional vector.

[0043] Total concentration of cooking fumes Also provided by the aforementioned sensors, the weighted average of PM2.5 and PM10 serves as an indicator of oil fume intensity, reflecting the current state of oil fume generation.

[0044] Duct temperature The temperature signal is measured by a bridge circuit consisting of NTC thermistors distributed at two locations on the inner wall of the air duct. The temperature signal is sampled by the ADC of the main control chip and converted into a temperature value by a lookup table method. The average of the two temperatures is taken to represent the overall temperature level inside the air duct.

[0045] Due to differences in the sampling periods and physical response speeds of various sensors, a synchronization window mechanism needs to be established within the main control system. Using a 1-second sliding time window, a moving average is performed on all sampled data in each control cycle (typically 500ms), and the results are combined to construct a time-aligned state vector. The specific state vector construction is as follows (this calculation is dimensionless):

[0046] ;

[0047] in: This represents the average rotational speed of the fan within the current second, representing time t. The average pressure difference in the duct over time t; Let be the moving average of the motor power over time t; The six-dimensional particle size distribution vector at time t ; Let fume concentration be the moving average over time t. Let be the average temperature of the air duct at time t.

[0048] Based on this, in order to construct an indicator variable that can more effectively reflect the actual load pressure, the following composite physical quantity construction formula is proposed (this calculation is dimensionless):

[0049] ;

[0050] in: The drag particle index represents the wind resistance per unit rotational speed. This represents the average pressure difference in the air duct. Indicates the first Particle concentration within each particle size range; This represents the average rotational speed of the fan.

[0051] This formula reflects the combined effect of pressure difference and particle size concentration on wind resistance. When wind speed is low, particle size is large, and concentration is high... This will increase significantly, reflecting an increase in system resistance. To ensure numerical scale consistency, this quantity is normalized together with the state vector using standard deviation normalization.

[0052] Output: The output of this step is the following two quantities, which will serve as the input for the next step of modeling: State vector. Standardized multidimensional state data; wind resistance particle index : This reflects the resistance strength index of the system per unit speed.

[0053] This step involves deploying a mature and mass-producible sensor array at the core nodes of the suction and exhaust system. Combined with a sliding window mechanism and a composite quantity construction method, this creates a realistic, dynamic, and reproducible multivariable state space. Particularly noteworthy is the design of a method to express the relationship between particle size distribution and wind resistance. This independent construct enables the system to evaluate resistance strength in real time without model dependence, providing significantly higher expressive efficiency for subsequent motor-load matching modeling.

[0054] S2. Based on the state vector and the wind resistance particle index, establish a wind resistance prediction model to express the nonlinear relationship between electromagnetic torque and wind resistance in the suction and exhaust system in real time, and output the current predicted wind resistance and the actual motor output torque.

[0055] Specifically, the core task of this step is to construct an adaptive motor-duct coupling model. This paper proposes a model to express the nonlinear relationship between electromagnetic torque and air resistance in a suction and exhaust system in real time. Unlike traditional static matching curves, suction and exhaust systems in actual kitchen environments are continuously affected by physical disturbances such as sudden changes in oil fume particle size, pot obstruction, and temperature rise, resulting in obvious unsteady-state characteristics in the system load. Especially under the condition of "rapid changes in particle size spectrum + sudden increase in duct pressure difference", traditional models cannot reflect the trend of load surge in time, which can easily cause fan control lag or overcompensation, affecting suction and exhaust efficiency and energy consumption. Based on the previous state space, this paper proposes a model with a clear structure, online correction capability, and the ability to incorporate disturbance compensation and coupled nonlinear characteristics. This provides predictive support for subsequent control strategies.

[0056] Furthermore, the model Aimed at predicting current wind tunnel resistance With the required output torque of the motor This invention aims to achieve dynamic feedforward adjustment of the control layer. Traditional wind resistance modeling mainly relies on static function mapping of pressure difference or rotational speed, ignoring the higher-order coupling effect of particle size distribution and the coupling trend among oil fume, heat flow, and pressure difference, making it difficult to capture rapid disturbances in suction and exhaust scenarios. The online modeling scheme proposed in this invention consists of two parts: the main model is a regression structure with physical constraints, supplemented by a disturbance gain correction term to reflect abnormal operating conditions in real time.

[0057] The wind resistance prediction model adopts the following structure (this calculation is dimensionless):

[0058] ;

[0059] in: This is the inherent resistance of the system; The contribution weights of each physical quantity to wind resistance; This refers to the pressure difference in the air duct. Total concentration of particle size distribution; Duct temperature reflects the trends of heat diffusion and flue gas buoyancy; Average fan speed; This is the disturbance gain coefficient; It is the square of the first derivative of the drag particle index, used to capture the sudden increase in drag caused by sudden disturbances (such as the emission of cooking fumes).

[0060] in, Let be the square of the first derivative of the wind resistance particle exponent, calculated as follows:

[0061] Known ,

[0062] Its time derivative is

[0063] ;

[0064] in:

[0065] : The wind resistance particle index defined by the above formula.

[0066] The average value of the duct pressure difference over a time window is obtained by moving average from the differential pressure sensor signal.

[0067] : No. Particle counts for each particle size channel, derived from segmented outputs of an infrared particle sensor.

[0068] The sum of the particle size counts of the six segments is used to simplify the writing.

[0069] The average value of the fan speed over a time window is obtained by moving average from the counts of the shaft-end encoder.

[0070] , , The first derivative of the corresponding signal with respect to time is obtained by the differential approximation of adjacent sampling points in the controller and then directly substituted into the above formula for calculation.

[0071] Next, the wind resistance prediction results will be used to calculate the actual required electromagnetic torque. Here, an energy consumption-load linkage formula based on aerodynamic work and similarity laws is used, with the motor power value in the state vector as a constraint term, and a load coordination regularization term is introduced (this calculation is dimensionless):

[0072] ;

[0073] in: The predicted wind resistance is output from the previous formula; This represents the current average speed of the fan. Depend on The conversion yields, The operating efficiency used as a constant parameter in the current period (the following calculations are dimensionless). This represents the current actual sampled power of the motor; The regularization coefficient is used to penalize power estimation errors and prevent the model from overfitting under high loads. This is the predicted torque from the previous control cycle; The angular velocity of the previous control cycle. Based on aerodynamic work and similarity laws: the aerodynamic side satisfies volume work and pipeline characteristics, while the axial flow fan approximately satisfies the similarity law and mechanical work relationship. Combining formula (4), we have: and and (Right now ) is negatively correlated, where, For based on The predicted theoretical output power of the motor, = This is because the larger these two terms are in the denominator of equation (4), the higher the effective conversion capacity is under the same aerodynamic work demand. Increasing the model's uncertainty (needing for conservative contraction, which increases the bias term) will make the required shaft torque estimation more difficult. The corresponding decrease; while the molecule The physical meaning is the aerodynamic load torque scale obtained by combining the pipeline system curve and the similarity law of the fan, that is, the current predicted wind resistance. With rotational speed The following is the magnitude of the shaft-side torque required to overcome the aerodynamic resistance of the air duct, followed by motor-side and uncertainty corrections using efficiency correction and power consistency regularization of the denominator.

[0074] in, The initial values ​​(i.e., the values ​​before the first control cycle) are usually set to zero or estimated based on the idling state, i.e., the angular velocity and torque before the first control cycle. Furthermore, The physical meaning is the "aerodynamic load torque scale" obtained by combining the pipeline system curve and the similarity law of the fan, that is, the current predicted wind resistance. With rotational speed The following is the magnitude of the shaft-side torque required to overcome the aerodynamic resistance of the air duct, followed by motor-side and uncertainty corrections using efficiency correction and power consistency regularization of the denominator.

[0075] in, ;

[0076] in, The motor output torque of the previous cycle (calculated by inversely from the motor current and speed); Angular velocity; These are the bus voltage and current sampled in real time by the frequency converter; The power factor is given by the inverter phase measurement module.

[0077] This ratio, after being filtered by a first-order digital filter, is used as the current cycle efficiency input formula (4).

[0078] in, The actual sampled power of the motor is currently calculated directly by the frequency converter. And obtained after smoothing for 500 ms. Theoretical output power based on predicted torque (this calculation is dimensionless):

[0079] ;

[0080] in This is the current iteration value of formula (4). This is the average rotational speed of the previous cycle; this product reflects the mechanical power estimate of "predicted torque × real-time angular velocity", used in conjunction with... Apply deviation constraints.

[0081] (Here, I would like to explain the relationship between formulas (4) and (6): The controller first uses the angular velocity of the previous control cycle.) and preliminary torque estimates The calculation formula (6) yields the following results. , and then Substituting into formula (4) generates a new torque. Only 1–2 iterations are needed to make Converging to a minimum, then the convergence value is... and Output. Wherein, .

[0082] The load coordination regularization coefficient, given by system debugging, is used for amplification. Bias, suppressing the risk of overfitting in high-load scenarios.

[0083] This torque formula embodies a key innovation: it incorporates physical measurements (power) in the state space into the verification terms of wind resistance torque inference, thereby avoiding excessively high ineffective torque output in special scenarios where both particle size and pressure difference are large but suction and exhaust efficiency is low.

[0084] For example, when the cookware is blocking the air intake but the concentration of oil fume particles is extremely high, only looking at... It might be mistakenly identified as high load, but If the actual increase is not significant, it indicates that the suction and discharge resistances are not truly transmitted to the motor. In this case, the regularization term will increase the denominator, automatically reducing the resistance. This avoids the erroneous output of high-energy-consuming signals. This design precisely reflects the special characteristics of the suction and exhaust system scenario and is one of the most technically valuable highlights of this step.

[0085] The entire operational logic of this model can be deployed on the edge control panel, where... The parameters are updated online using the Recursive Least Squares (RLS) algorithm, with perturbation derivatives. The regularization term is obtained through sliding difference and can be updated once per cycle by the main control logic. The entire model can be iterated twice per second.

[0086] This step outputs two physical predictions:

[0087] Predicting wind resistance This will be used as the basis for judging the thermal zone model and airflow configuration in step three;

[0088] Predicting motor output torque It will be used as one of the target variables for control optimization and participate in the design of energy consumption regulation and motor response strategies.

[0089] S3. Combining the current predicted wind resistance, actual motor output torque, and temperature array, generate a two-dimensional thermal map and dynamically mark the locations of the thermal zones that need intervention as control sensitive areas.

[0090] Specifically, this step is based on the motor-drag coupling model. Output results and Combined with temperature array Constructing a two-dimensional spatial thermal map and extract control sensitive areas .

[0091] Input: Predicted wind resistance : From the model , representing the total wind resistance response at the current moment; electromagnetic torque : From the model Reflects the load response caused by wind resistance; temperature array : Collects data through an array of infrared thermal sensors positioned above the stove, inside the range hood, and near the air duct inlet, forming a... The two-dimensional temperature field matrix is ​​updated at a frequency of 200ms, and the resolution can be 4×4 or 6×6.

[0092] In addition, to incorporate the thermal accumulation trend over the time scale, this step also introduces the temperature moving average in the state vector. It serves as a global thermal benchmark for constructing local relative thermal risks.

[0093] The goal of this step is to generate This is used to reflect the degree of thermal risk at different spatial locations under the current wind resistance background. Traditional temperature distribution judgments are based only on absolute values ​​or local differences, which are insufficient to reflect the dynamic characteristics of the "hot zone driven by wind resistance changes" in the suction and exhaust system. This scheme constructs a spectral function with temperature difference, wind resistance response, torque load, and local heat dissipation capacity as joint inputs to achieve spatial structure mapping across physical fields. Specifically:

[0094] First, spatial local difference and time derivative processing are performed on the temperature array to obtain the degree of heat accumulation per unit time in each region (this calculation is dimensionless):

[0095] ;

[0096] in: The global average temperature represents the environmental thermal reference. The time-dependent weight reflects the proportion of the rate of temperature rise in the assessment of thermal risk; Indicates the first Line 1 The thermal sensor at the measuring point at time Temperature; It is the infrared thermal array in the first Line 1 The measurement points were in the previous control cycle (compared to the current time). Separated Historical temperature samples were collected and cached after undergoing the same calibration / smoothing process;

[0097] The first term captures the relative heating location in space, and the second term captures the local heating rate over time.

[0098] Then based on the wind resistance prediction value With torque A nonlinear wind resistance amplification factor is introduced to reflect the coupling trend that "the risk of hot zones increases sharply with the increase of wind resistance pressure" (this calculation is dimensionless):

[0099] ;

[0100] in: This is a two-dimensional thermal map, representing a thermal risk map that covers a two-dimensional coordinate plane; The amplification effect of wind resistance changes on the rate of hot zone expansion; The influence of electromagnetic load on the "extrapolation zone" of heat accumulation is controlled by using a logarithmic function to balance the nonlinear growth trend under high load response.

[0101] Formula (8) is derived based on energy balance and the similarity law of wind turbines: the shaft work of the motor is When duct resistance is high or local backflow exists, a larger proportion of shaft work is consumed as turbulent and wall friction losses and locally converted into heat. These losses change in the same direction as the load torque, therefore the risk of hot spots increases with... Increase and improve;

[0102] Equation (8) adopts The reason is to maintain the effect of "monotonically increasing but gradually saturating" - to avoid excessive amplification in the high load area, so that the thermal risk amplification is attenuated to extreme torque, which not only conforms to the actual saturation characteristics, but also improves numerical stability.

[0103] Indicating in the prediction step For grid points The thermal risk intensity (scalar) calculated according to equation (8) is determined by the local temperature rise at that point. Determined together with the drag / torque amplification term, it reflects the relative risk of heat buildup and backflow at that point.

[0104] This formula represents a key innovation in the context of this invention: by structurally mapping wind resistance load to temperature change weighting logic, the same temperature rise is assigned a higher thermal risk score under conditions of sudden increase in wind resistance or increased motor load. For example, when... An increase of 2°C would normally be considered a moderate risk, but at this time... =1.9 (significantly higher than the reference value of 1.0). =0.8 N·m, then It will be nonlinearly amplified and incorporated into the thermal control sensitive region.

[0105] Next, according to Extract control sensitive areas This region is defined as exceeding the dynamic threshold. The set of all points (this calculation is dimensionless):

[0106] ;

[0107] in, Represents the global temperature reference: at time [time missing] For infrared thermal array First, perform a spatial average (all) The temperature obtained by averaging the values ​​over a short window is used as a threshold. Adaptive baseline adjustment; It is not a fixed constant, but depends on the current situation. and Adaptive adjustment, for example:

[0108] If the wind resistance is 30% higher than the reference value, the threshold decreases by 10%, thus increasing sensitivity.

[0109] If the global temperature rises slowly, the threshold is increased to avoid false alarms;

[0110] This adjustment strategy is executed online via table lookup, without introducing complex calculations.

[0111] In practical control strategies, The corresponding areas of the stove layout will be mapped (e.g., left stove, middle stove, right stove) and converted into suggestions for adjusting the airflow angle of the secondary duct or areas to enhance the target wind speed. For example, if Including coordinates If the heat buildup is severe in the upper right corner of the stove, the system should be adjusted to lower the main fan speed on the right side or increase the jet angle of the central auxiliary air duct.

[0112] Output: Space thermal map :one A floating-point matrix, used to describe the heat risk weights at each point, is given by equation (8) at time [time value missing]. For grid points The calculated thermal risk intensity (scalar) is directly used as the weight of that point in the summation;

[0113] Control sensitive areas : Point set array, representing the location of the area that requires key risk control intervention, providing regional weight support for the next step of generating the MPC objective function.

[0114] S4. Based on the two-dimensional thermal map and the control sensitive area, construct an optimization objective function with the goals of minimizing thermal risk, minimizing energy consumption, and ensuring control stability, and coordinate the output of control commands for fan speed, torque, and air guide angle.

[0115] Specifically, the core task of this step is to construct a multi-objective model predictive controller (MPC) to control the sensitive areas identified in the previous step. and thermal map Compared to previously predicted wind resistance and electromagnetic torque They are all incorporated into an optimized control framework to achieve variable frequency output from the wind turbine. Motor response torque and the angle of the secondary air duct Dynamic coordination control. The entire MPC structure needs to accomplish the following three tasks within a short cycle: predict the evolution trend of the target variable, evaluate the hot zone and energy consumption objective function, and output a real-time executable wind turbine speed and wind direction control strategy.

[0116] The controller objective is based on the future window [t, t+ The prediction of ], where t+ This represents the current time plus the prediction window length;

[0117] Under constraints, the control variable group The objectives are to achieve the following: (1) minimize total energy consumption, (2) minimize hot zone risk, and (3) achieve smooth controller response. To this end, the following objective function is designed for this step (this calculation is dimensionless):

[0118] ;

[0119] in: To predict the electromagnetic torque value within the period, from supply; The intensity of thermal risk is represented and used as the thermal risk weight for each point in the calculation, which is given by the thermal zone map in step three. The set of coordinates of sensitive points is directly derived from the previous step; This indicates the rate of change of the fan speed, controlling airflow fluctuations. This indicates the rate of change of the angle of the secondary wind channel, which affects the local wind direction. These are the weighting coefficients of the objective function, corresponding to the three aspects of energy consumption, thermal balance, and control smoothness, respectively.

[0120] The objective function structure embodies three distinct innovative characteristics specific to the invention scenario of suction and discharge systems:

[0121] For the first time, the "weight of spatially sensitive points in the hot zone" was directly embedded into the control function as a control objective, realizing a direct response to spatially non-uniform heat flow disturbances;

[0122] Introducing electromagnetic torque prediction values ​​instead of traditional power indicators is more in line with the physical model of the motor and can better control energy consumption trends.

[0123] A smoothing control term for the rate of change of wind speed / direction has been added, enabling the system to maintain stability under high-speed disturbance scenarios, such as the opening and closing of pot lids or sudden bursts of food.

[0124] For example, if the current Including the upper left corner of the stove and central The heat intensities are respectively , And at this time It rose significantly to 1.8. If the value increases from 0.5 in the previous cycle to 0.9, the controller will increase the weight of the second objective function during the prediction process, guide the angle of the secondary air duct to deflect towards this area, and control the main wind speed to rise slowly within an acceptable range, rather than increasing the air volume significantly and causing energy waste.

[0125] Meanwhile, this controller sets the following physical constraints on the control variables, reflecting the actual hardware boundaries of the device (this calculation is dimensionless):

[0126] ;

[0127] in, These are the upper and lower limits of the safe wind speed for the equipment. This refers to the maximum rate of change of the secondary air duct angle, which is generally set by the factory parameters to not exceed a certain value per cycle. .

[0128] The controller runs an MPC solver every 250ms and updates in real time via a scrolling window. The optimization process is implemented using a single-layer QP structure and can be implemented by an open-source embedded MPC solver (such as ACADO or FORCES). The computation time is less than 50ms and it can be deployed on platforms such as STM32F4 and TI C2000 series with floating-point support.

[0129] The controller outputs the following two variables per cycle to directly drive the physical actuator response: fan speed. With electromagnetic torque : As the direct control input of the variable frequency drive module; secondary air duct angle Output to the wind vane servo mechanism to control wind direction deflection.

[0130] This invention also provides a real-time optimization system for intake and exhaust energy efficiency based on multivariable frequency conversion technology, the system comprising:

[0131] The sensor acquisition module is used to collect data in real time from multiple sources, including the average speed of the fan, the average pressure difference in the duct, the sliding average of the motor power, the total concentration of the particle size distribution, the sliding average of the oil fume concentration, and the average temperature of the duct. After synchronous processing, a state vector and a wind resistance particle index are generated.

[0132] The online modeling module is used to establish a wind resistance prediction model based on the state vector and wind resistance particle index, so as to express the nonlinear relationship between electromagnetic torque and wind resistance in the suction and exhaust system in real time, and output the current predicted wind resistance and the actual motor output torque.

[0133] The thermal zone analysis module is used to combine the current predicted wind resistance, actual motor output torque and temperature array to generate a two-dimensional thermal zone map and dynamically mark the locations of the thermal zones that need intervention as control sensitive zones.

[0134] The predictive control module is used to construct an optimization objective function based on the two-dimensional thermal map and the control sensitive area, with the goals of minimizing thermal risk, minimizing energy consumption, and ensuring control stability, and to coordinate the output of control commands for fan speed, torque, and air guide angle.

[0135] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0136] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of apparatuses or units may be electrical, mechanical, or other forms.

[0137] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0138] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for real-time optimization of intake and exhaust energy efficiency based on multivariable frequency conversion technology, characterized in that, The method includes the following steps: The average fan speed, average duct pressure difference, motor power sliding average, total particle size distribution, oil fume concentration sliding average, and duct temperature average are collected in real time by multi-source sensors. The data are then processed synchronously to generate a state vector and wind resistance particle index. Based on the state vector and the wind resistance particle index, a wind resistance prediction model is established to express the nonlinear relationship between electromagnetic torque and wind resistance in the suction and exhaust system in real time, and output the current predicted wind resistance and the actual motor output torque. Combining the current predicted wind resistance, actual motor output torque, and temperature array, a two-dimensional thermal map is generated and the locations of the thermal zones requiring intervention are dynamically marked as control sensitive areas; wherein, the temperature array is acquired by an infrared thermal sensor. Based on the two-dimensional thermal zone map and the control sensitive zone, an optimization objective function is constructed with the goals of minimizing the risk of the two-dimensional thermal zone map, minimizing energy consumption, and achieving control stability. This function coordinates the output of control commands for fan speed, torque, and air guide angle. The drag particle index represents the drag particle index per unit speed, which is calculated based on the average pressure difference in the duct, the total concentration of particle size distribution, and the average speed of the fan. It is used to reflect the combined effect of pressure difference and particle size concentration on wind resistance. The wind resistance prediction model calculates the current predicted wind resistance by weighted summation based on the preset resistance, combined with the average pressure difference in the duct, the total concentration of the particle size distribution, the average temperature of the duct, and the average speed of the fan, and the square of the first derivative of the wind resistance particle index. The actual motor output torque is calculated by combining the current predicted wind resistance, average fan speed, sliding average motor power, and predicted theoretical motor output power. The entire operating logic of the wind resistance prediction model is deployed in the edge control board. The model parameters are updated online using a recursive least squares algorithm. The perturbation derivative based on the wind resistance particle exponent is obtained through sliding difference. The regularization term for predicting the theoretical output power of the motor is updated once per cycle by the main control logic. The wind resistance prediction model can iterate twice per second. The perturbation derivative is the first derivative of the wind resistance particle exponent. The process of combining the current predicted wind resistance, actual motor output torque, and temperature array to generate a two-dimensional thermal map and dynamically marking the locations of the thermal zones requiring intervention as control sensitive areas specifically includes: Spatial local difference and time derivative processing is performed on the temperature array to obtain the degree of heat accumulation per unit time in each region; Based on the current predicted wind resistance and the actual motor output torque, a nonlinear wind resistance amplification factor is introduced to calculate the two-dimensional thermal map. The control sensitive area is extracted based on the two-dimensional thermal map; wherein, the sensitive area is the set of all points exceeding the dynamic threshold; The control sensitive area represents the location of the area that currently requires key risk control intervention.

2. The method for real-time optimization of intake and exhaust energy efficiency based on multivariable frequency conversion technology according to claim 1, characterized in that, If the control sensitive area is actually used, the control sensitive area will be mapped to the area corresponding to the stove layout and transformed into a secondary air duct airflow angle adjustment suggestion or an auxiliary target wind speed enhancement area.

3. The method for real-time optimization of intake and exhaust energy efficiency based on multivariable frequency conversion technology according to claim 1, characterized in that, Based on the two-dimensional thermal map and the control sensitive area, an optimization objective function is constructed with the goals of minimizing the risk, minimizing energy consumption, and ensuring control stability of the two-dimensional thermal map. This function coordinates the output of control commands for fan speed, torque, and air guide angle. Specifically, this includes: For the prediction of future windows, the fan speed, motor output torque and auxiliary air duct angle are used as the control variables. The design optimizes the control variable set using the objective functions of minimizing total energy consumption, minimizing the risk of the two-dimensional thermal map, and smoothing the controller response. At the same time, the controller is designed with physical constraints to reflect the hardware boundaries of the actual equipment and outputs control commands for fan speed, torque, and air guide angle.

4. The method for real-time optimization of intake and exhaust energy efficiency based on multivariable frequency conversion technology according to claim 3, characterized in that, The physical constraints are that the fan speed is within the upper and lower limits of the equipment's safe wind speed, and the maximum response rate of change of the secondary air duct angle cannot exceed a preset threshold.

5. The method for real-time optimization of intake and exhaust energy efficiency based on multivariable frequency conversion technology according to claim 3, characterized in that, The controller runs an MPC solution once every 250ms and updates in real time using a scrolling window.

6. A system for implementing the real-time optimization method for intake and exhaust energy efficiency based on multivariable frequency conversion technology as described in claim 1, characterized in that, The system includes: The sensor acquisition module is used to collect data in real time from multiple sources, including the average speed of the fan, the average pressure difference in the duct, the sliding average of the motor power, the total concentration of the particle size distribution, the sliding average of the oil fume concentration, and the average temperature of the duct. After synchronous processing, a state vector and a wind resistance particle index are generated. The online modeling module is used to establish a wind resistance prediction model based on the state vector and wind resistance particle index, so as to express the nonlinear relationship between electromagnetic torque and wind resistance in the suction and exhaust system in real time, and output the current predicted wind resistance and the actual motor output torque. The thermal analysis module is used to combine the current predicted wind resistance, actual motor output torque, and temperature array to generate a two-dimensional thermal map and dynamically mark the locations of the thermal zones that need intervention as control sensitive areas; wherein, the temperature array is acquired by an infrared thermal sensor. The predictive control module is used to construct an optimization objective function based on the two-dimensional thermal map and the control sensitive area, with the goals of minimizing the risk of the two-dimensional thermal map, minimizing energy consumption, and achieving control stability, and to coordinate the output of control commands for fan speed, torque, and air guide angle.

Images