A multi-parameter real-time monitoring system for a large indoor ventilation system

Abstract

The present application relates to the technical field of ventilation control, and particularly relates to a multi-parameter real-time monitoring system of an indoor large-scale ventilation system. The system uses a monitoring data acquisition module to acquire basic monitoring data. According to the collected temperature data, Kriging interpolation is performed by using a spherical semi-variogram function to obtain optimal unbiased estimation values under various coordinates, and an updraft tendency map is formed. The real-time updraft tendency map is segmented to determine a plurality of connected domains, the current indoor environment pollution plume mode index is evaluated by comparing particle concentration data and thermal energy data, and then a control strategy mode is determined. Under each control strategy mode, the regulation and control parameters can be determined according to the area of the connected domain or the pixel value in the real-time updraft tendency map. Through data monitoring and feature analysis, the present application constructs a real-time updraft tendency map, and realizes dynamic, low-disturbance and low-delay ventilation control in a large-scale commercial indoor environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ventilation control, and in particular to a multi-parameter real-time monitoring system for a large indoor ventilation system. BACKGROUND

[0002] In a large commercial complex or other high and open space, the hot plume carrying oil smoke and peculiar smell generated in the bottom dining area will naturally rise due to the buoyancy effect and spread to the upper retail area, seriously affecting the air quality. The existing ventilation mostly adopts global large air volume ventilation, which is high in energy consumption and poor in effect; for the one using the feedback control of traditional concentration sensor, due to the bottom logic of global ventilation, the response lag is obvious, and preventive interception cannot be realized. The existing one generally lacks real-time modeling and prediction ability for key physical processes such as rising kinetic energy and propagation path of pollutants, and it is difficult to develop efficient, low-disturbance and rapid control strategies. SUMMARY

[0003] In order to solve the technical problem that the prior art has strong decision lag for large indoor ventilation and cannot make efficient and rapid control strategies for the whole indoor space, the purpose of the present application is to provide a multi-parameter real-time monitoring system for a large indoor ventilation system, and the technical scheme adopted is as follows:

[0004] The present application provides a multi-parameter real-time monitoring system for a large indoor ventilation system, which comprises:

[0005] A monitoring data acquisition module is configured to acquire particle concentration data at a particle source in the bottom dining area, thermal energy data at a heat flux source, and temperature data at each height coordinate in the vertical direction.

[0006] A monitoring data analysis module is configured to perform Kriging interpolation on the temperature data using a spherical semi-variogram function to obtain the best unbiased estimate value at each coordinate and form a real-time updraft tendency graph; and compare the particle concentration data and the thermal energy data to obtain a pollution plume mode index.

[0007] A ventilation decision module is configured to segment the real-time updraft tendency graph to obtain a plurality of connected domains; control a supply air unit to execute a control strategy for each connected domain; and determine the control strategy according to the pollution plume mode index, wherein the control strategy includes a balance mode, a temperature adjustment mode and a wind speed adjustment mode; in the temperature adjustment mode, the target temperature difference of the control strategy is determined according to the pixel value in the connected domain; and in the wind speed adjustment mode, the target wind speed of the control strategy is determined according to the area of the connected domain.

[0008] Further, the real-time updraft tendency graph comprises:

[0009] A two-dimensional coordinate system is constructed based on an indoor horizontal plane, and the two-dimensional coordinate system includes real observation points, physical constraint points, and unknown data points;

[0010] The coordinate value of the real observation point is that, at each time in a preset real-time time window, each height in the vertical direction is divided into three categories of middle height, upper height, and lower height, the average temperature data between the lower height and the upper height is taken as reference temperature data, and the difference between the temperature data of the middle height and the reference temperature data is taken as an instantaneous buoyancy signal value at each time; the maximum instantaneous buoyancy signal value in the real-time time window is selected as the coordinate value of the real observation point at the real-time time;

[0011] The physical constraint point is an indoor air conditioner air outlet position and an indoor boundary, and the coordinate value is a preset fixed value;

[0012] In the two-dimensional coordinate system, a Kriging interpolation is performed according to a learned spherical semivariogram function to obtain the unbiased estimation value as the coordinate value of the unknown data point; the real observation point, the physical constraint point, and the interpolated unknown data point constitute a real-time updraft tendency map.

[0013] Further, the model parameters of the spherical semivariogram function are learned by using a three-stage evolution process, the first stage is used to accumulate historical data by using preset model parameter values; the second stage is used to globally optimize the model parameters based on a genetic algorithm to determine optimal model parameters; and the third stage is used to perform online adjustment on the optimal model parameters by using a mode of periodic global optimization and online fine tuning of real-time data.

[0014] Further, the method for obtaining the pollution plume mode index includes:

[0015] After normalizing the particle concentration data and the thermal energy data in their respective dimensions, normalized particle concentration data and normalized thermal energy data are obtained; the difference between the normalized thermal energy data and the normalized particle concentration data is taken as the pollution plume mode index.

[0016] Further, the real-time updraft tendency map is segmented to obtain a plurality of connected domains, including:

[0017] In the real-time updraft tendency map, data less than a preset response threshold is set to 0 to obtain a denoising result map; the connected domain segmentation algorithm is used to segment the denoising result map to obtain the connected domain.

[0018] Further, the control strategy is determined according to the pollution plume mode index, including:

[0019] If the pollution plume mode index is greater than a preset first threshold value, the control strategy is a temperature adjustment mode; if the pollution plume mode index is less than a preset second threshold value, the control strategy is a wind speed adjustment mode; if the pollution plume mode is greater than or equal to the second threshold value and less than or equal to the first threshold value, the control strategy is a standard balance strategy; the first threshold value is greater than the second threshold value.

[0020] Further, the method for obtaining the target air supply temperature difference comprises:

[0021] The maximum pixel value in the connected domain is taken as a peak intensity, and the peak intensity is multiplied by a preset coefficient to obtain the target air supply temperature difference.

[0022] Further, the method for obtaining the target air supply wind speed comprises:

[0023] The area of the connected domain is normalized to obtain an area weight, and the product of the area weight and a preset maximum wind speed adjustment amount is multiplied to obtain a real-time wind speed adjustment amount; and the sum of a basic wind speed and the real-time wind speed adjustment amount is taken as the target air supply wind speed.

[0024] Further, a preset number of air supply units closest to the center coordinates of each connected domain are selected to execute the control strategy.

[0025] Further, the particle concentration data is the average concentration data of the total exhaust pipeline in a real-time time window; the thermal energy data is the average power of the total active power detected on the main circuit of the total power distribution box in the real-time time window; and the range of the real-time time window is one minute before a real-time time.

[0026] The present application has the following beneficial effects:

[0027] The present application performs data monitoring on a bottom dining area in a large indoor space, and uses a monitoring data acquisition module to acquire basic monitoring data. The monitoring data is further analyzed, and in order to obtain global information of the indoor space in limited data monitoring points, the present application uses a spherical semi-variogram function to perform Kriging interpolation according to the collected temperature data to obtain optimal unbiased estimation values at various coordinates, and forms an upward airflow tendency diagram. Because the data values at various positions in the upward airflow tendency diagram can represent the tendency strength of the upward movement of the airflow, the real-time upward airflow tendency diagram can be segmented to determine a plurality of connected domains, that is, each connected domain is an independent to-be-regulated region. In order to perform effective control, the pollution plume mode index of the current indoor environment is evaluated by comparing the particle concentration data and the thermal energy data, and then the control strategy mode is determined. Under each control strategy mode, the regulation and control parameters can be determined according to the area of the connected domain or the pixel value in the real-time upward airflow tendency diagram, so that adaptive and targeted efficient and rapid ventilation control is realized. BRIEF DESCRIPTION OF DRAWINGS

[0028] The technical solutions and advantages in the embodiments of the present application or the prior art will be described below more clearly. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort based on these drawings.

[0029] Figure 1 A block diagram of a multi-parameter real-time monitoring system of an indoor large ventilation system according to an embodiment of the present application. DETAILED DESCRIPTION

[0030] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined purposes, the specific embodiments, structures, features and effects of a multi-parameter real-time monitoring system of an indoor large ventilation system according to the present application are described in detail as follows in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0032] The specific scheme of the multi-parameter real-time monitoring system of an indoor large ventilation system provided by the present application is described in detail below in combination with the drawings.

[0033] Please refer to Figure 1 which shows a block diagram of a multi-parameter real-time monitoring system of an indoor large ventilation system according to an embodiment of the present application. The system includes a monitoring data acquisition module 101, a monitoring data analysis module 102, and a ventilation decision module 103.

[0034] The embodiment of the present application is directed to a ventilation control system in a large commercial indoor environment. In the commercial indoor environment, the hot plume of oil fume and odor is generated in the bottom dining area due to the influence of food serving and people flow. Therefore, the monitoring data in the embodiment of the present application is also collected based on the bottom dining area. The monitoring data acquisition module 101 is used to acquire the particle concentration data at the particle source in the bottom dining area, the thermal energy data at the heat flux source, and the temperature data at each height coordinate in the vertical direction. It should be noted that the movement of the hot plume is affected by the temperature. If the indoor vertical space presents a feature of warm on the top and cool on the bottom, the rising of the hot plume formed in the bottom dining area will be inhibited due to the temperature propagation characteristics. Similarly, if it presents a feature of cool on the top and warm on the bottom, it will be promoted. Therefore, the temperature data at each height coordinate in the vertical direction of the indoor can represent the moving tendency of the hot plume.

[0035] In the embodiment of the present application, an optical scattering particle concentration meter is installed in the central exhaust duct of the B1 layer dining area before the oil fume purification equipment. The device measures the particle concentration of PM2.5 level in the duct with a sampling period of 10 seconds, and the unit is micrograms per cubic meter (μg / m³). The particle concentration can represent the mass intensity of the oil fume aerosol generated from the source area in real time, and indirectly represents the amount of particles that cannot be captured by the exhaust. Further, in other embodiments of the present application, the implementer can also select other particle concentration sensors according to the site conditions, or use other measurement points to avoid overrange. It should be noted that, in order to avoid the influence of accidental data on real-time monitoring, the average concentration data collected in the real-time time window of the real-time moment is taken as the particle concentration data in the monitoring data acquisition module 101 for each real-time moment. The range of the real-time time window is one minute before the real-time moment. This technical feature stably estimates the average mass flow of the oil fume aerosol generated from the source area in the past one minute, which is equivalent to the concentration of pollutants that cannot be captured by the exhaust, resulting in the equivalent load released in the atrium environment.

[0036] In the embodiment of the present application, an intelligent electric meter with a standard communication interface is installed on the main circuit of the total distribution box of the B1 layer dining area. The device also measures the total active power of the entire dining area with a sampling period of 10 seconds, and the unit is kilowatt (kW). This feature represents the total heat energy released by the source area to the environment in real time. Under the premise of constant power operation of the exhaust, the total heat energy can represent the self-brought heat when the food is served in the guest area, and the heat that cannot be captured by the exhaust, which is the macroscopic power source driving the pollution plume to rise. Similarly, in the embodiment of the present application, the thermal energy data is the average power of the total active power detected on the main circuit of the total distribution box in the real-time time window. This technical feature represents the driving force of the plume rising.

[0037] In the embodiment of the present application, in the atrium, in order to obtain temperature data with spatial distribution information, a sensor array is arranged in the vertical direction, and the sensors in a group of sensor arrays represent the temperature data collected at different heights in the vertical direction. The embodiment of the present application deploys three sensor arrays, that is, collects temperature data in three vertical directions, and these arrays are arranged at key nodes of the potential pollution plume path, for example, one is located directly above the center of the catering area, and the other two are located at the edge of the corridor or atrium that may be affected by crosswinds. Each sensor array is composed of three high-precision digital temperature sensors fixed along the vertical axis (for example, at a height of 3.5 meters, 4.0 meters, and 4.5 meters). All sensors in all arrays measure temperature synchronously at a sampling period of 1 second, and the unit is Celsius (°C).

[0038] In the embodiment of the present application, the monitoring data acquisition module 101 can deliver the obtained parameters to the monitoring data analysis module for data analysis in combination with the time stamp of the real time.

[0039] The monitoring data analysis module 102 is used for further analyzing the monitoring data and quantifying the data characteristics. Because the acquisition of the monitoring data cannot obtain the global information of the indoor space, it is necessary to calculate the monitoring data at other positions in the space based on the obtained monitoring data. Because the particle concentration data and the thermal energy data obtained by the embodiment of the present application are data at the source of the catering area, such data do not need to be interpolated. The embodiment of the present application adopts a spherical semi-variogram function for Kriging interpolation of the temperature data to obtain the best unbiased estimate value of each coordinate in the indoor space, and further to construct a real-time updraft tendency map. Because the embodiment of the present application mainly collects temperature data at each height coordinate in the vertical direction, the real-time updraft tendency map constructed is an indoor plan view, and the coordinate value information represents the updraft tendency of the thermal plume in the vertical direction corresponding to the coordinate, which can be used for subsequent determination of the to-be-controlled area, control type, control strength, etc.

[0040] Meanwhile, the monitoring data analysis module 102 further compares the particle concentration data and the thermal energy data. The relative intensity difference between heat generation and oil smoke particle generation can represent the shape feature of the thermal plume formed by the current catering area, which is convenient for executing the optimal control strategy, and the shape feature may include two extreme intermediate states of strong particles and weak heat and strong heat and weak particles.

[0041] Preferably, in the embodiments of the present application, considering that the real-time updraft tendency map should be a floor plan of the indoor space, and the thermal plume updraft tendency in the vertical direction under the coordinate system needs to be represented, a two-dimensional coordinate system is constructed based on the indoor horizontal plane. In order to facilitate the construction of the coordinate system, the embodiments of the present application can construct a grid map using a fixed resolution in a grid format. The grid division in the embodiments of the present application is 0.5 m x 0.5 m under the world coordinate system, and finally a two-dimensional coordinate system is obtained.

[0042] In the two-dimensional coordinate system, there are real observation points, physical constraint points and unknown data points. The real observation points are the data actually collected by the sensors obtained by the monitoring data acquisition module 101. The sensor position is the coordinate position of the real observation point. Because the sensor collects temperature data at each height, that is, there is a set of temperature data at each time at one coordinate position. Therefore, the coordinate value of the real observation point is: in each time under the preset real-time time window, each height in the vertical direction of the coordinate is divided into three types of middle height, upper height and lower height. The average temperature data between the lower height and the upper height is taken as the reference temperature data, and the difference between the temperature data of the middle height and the reference temperature data is taken as the instantaneous buoyancy signal value at each time. The maximum instantaneous buoyancy signal value in the real-time time window is selected as the coordinate value of the real observation point at the real-time time. It should be noted that the instantaneous buoyancy signal value can represent the temperature distribution in the vertical direction of the corresponding coordinate. In the embodiments of the present application, each height in the vertical direction is divided into three parts on average, and the average of the temperature data collected in each part is the temperature data used to calculate the instantaneous buoyancy signal value. The instantaneous buoyancy signal value can represent the updraft tendency of the thermal plume. A positive value and a larger value indicate that the temperature at the middle height is higher, and the overall indoor environment presents a feature of warm in the upper part and cool in the lower part, which leads to a certain inhibitory effect on the updraft of the thermal plume. A negative value indicates that the temperature at the middle height is lower, and the overall indoor environment presents a feature of cool in the upper part and warm in the lower part, which leads to a certain aggravating effect on the updraft of the thermal plume.

[0043] The physical constraint points are positions where no sensors are deployed, but the thermal plume updraft tendency at this position is a static boundary condition set based on the recognized principles of building environment science and pre-input building information. Specifically, it includes the indoor air supply outlet position and the indoor boundary. The air supply outlet in the embodiments of the present application is the central position of the air supply outlet located at or capable of significantly affecting the height of 4 meters. The cold air of the air supply outlet at this position must produce a downward airflow, so its fixed value should be a negative number, -0.5 degrees Celsius, as a forced inhibition zone for the updraft. The indoor boundary is the wall, glass curtain wall and other solid boundaries around the atrium. A series of points are uniformly selected in such solid boundaries, and according to the no-slip boundary condition in fluid mechanics, the observation value is set to 0, as a forced vertical airflow-free zone.

[0044] In a two-dimensional coordinate system, in addition to the real observation points and the physical constraint points, all other coordinate points are unknown data points, and the unknown data points need to be interpolated and filled, and the learned spherical semi-variation function is used for Kriging interpolation to obtain the unbiased estimation value as the coordinate value of the unknown data point. Finally, the real-time updraft tendency diagram is formed.

[0045] Further, the learned spherical semi-variation function in the embodiment of the present application is a model parameter determined based on real monitoring data through online learning. In order to enable the model to continuously adapt to environmental changes, the learning is performed by using a three-stage evolution process. The first stage is used to accumulate historical data by using preset model parameter values; the second stage is used to globally optimize the model parameters based on a genetic algorithm to determine optimal model parameters; and the third stage is used to periodically globally optimize and online fine-tune the optimal model parameters by using real-time data. Specifically, the three-stage specific implementation process of the embodiment of the present application includes:

[0046] (1) The first stage:

[0047] The first stage in the embodiment of the present application starts at the time of 7 days before the system in the embodiment of the present application is put into operation. The semi-variation function is forcibly set as an isotropic standard parameter, wherein the model parameters mainly include five parameters of a, b, C0, C, and Range. The five parameters are all set as empirical values, and the purpose is to accumulate a sufficient set of historical observation point data to provide initial samples for subsequent learning.

[0048] a and b are anisotropy ratios, and the two parameters directly describe the directional characteristics of the atrium space. For example, if the atrium is a long and narrow rectangle, the airflow correlation along the long axis direction (assuming the X axis) may be much greater than that along the short axis direction. In the genetic algorithm, an excellent individual may find a larger a value and a smaller b value, which physically means that the tendency propagates farther and decays more slowly along the X axis.

[0049] Range is the range, and the parameter defines the maximum influence range of an isolated thermal plume. A smaller Range value means that the influence of the thermal plume is very localized and rapidly decays, and a larger Range value means that the influence can spread far away. The value is related to the overall size of the atrium and the ventilation condition.

[0050] C0 is the nugget value, and the parameter represents the inherent randomness and microscale variation in the atrium. In our scenario, it mainly quantifies two parts of noise: one is the measurement error of the sensor itself, and the other is the random temperature fluctuation caused by turbulence and the like in a very small spatial scale (for example, within a few centimeters) that cannot be captured by the model.

[0051] C is the difference between the base value and the nugget value, which represents the maximum spatial variation of the upward airflow tendency in the entire space. A large C value means that there can be very strong upward airflow areas and very stable areas in the space at the same time, that is, the physical field contrast is high.

[0052] At the end of the 7th day, before entering the second stage of global optimization, the initial empirical values need to be determined, including: 1. Calculate the experimental semi-variogram function, summarize all the real observation point data in the past 7 days, calculate the average semi-variogram value at different distance intervals, and obtain an experimental semi-variogram function graph; 2. Fit the theoretical model, use the least squares method to fit a standard spherical model (C0, C, Range as to be determined parameters) to the experimental semi-variogram function graph; 3. Determine the empirical value, the best parameters C0, C, Range obtained by fitting are used as an initial population individual or center reference of the search range in the subsequent genetic algorithm optimization process.

[0053] (2) The second stage:

[0054] After accumulating 7 days of data, a global optimization based on genetic algorithm is performed. The goal of this process is to automatically find and determine a set of semi-variogram function parameters that best describe the spatial correlation of the upward airflow in the current atrium space from a large number of possibilities.

[0055] In order to evaluate the pros and cons of each individual (i.e. each physical model assumption), an objective evaluation standard is needed, and the embodiment of the present application proposes a fitness function defined as a quantitative evaluation of the prediction ability for historical real data. Specifically, for any individual in the genetic algorithm population (i.e. a given {α, β, C0, C, Range} parameter), the following operations are performed to calculate its fitness:

[0056] 1. Load historical data, randomly extract a copy from the minute-level observation point data set accumulated in the past 7 days.

[0057] 2. Perform leave-one-out cross-validation: from the data set, temporarily remove each real observation point, such as the observation data of the array located in the center of the dining area, as the real answer to be verified. Using the extracted parameters, combining the remaining real observation points in the data set and all physical constraint points, the prediction value of the removed real observation point position is calculated by Kriging interpolation. Calculate the square error between the predicted value and the real answer as the prediction error, and the average of all prediction errors is the total mean square error MSE, which directly measures the prediction accuracy of the parameter model when facing real world data.

[0058] 3. Fitness calculation: Fitness is defined as 1 / (1+MSE). The advantage of this is that the smaller the MSE, the more accurate the model, and the higher the fitness value, with a maximum of 1, which is in line with the intuitive logic of the survival of the fittest of genetic algorithms.

[0059] After determining the fitness, the genetic algorithm and particle swarm algorithm are used to evolve and globally optimize the model, and through the simulation of biological evolution, a global search is performed in the parameter space, and finally the optimal model parameters are evolved.

[0060] First, the population needs to be initialized. An initial population containing hundreds of different individuals (i.e. hundreds of different {α, β, C0, C, Range} parameter combinations) is randomly generated. This represents hundreds of initial guesses of the physical laws of the atrium space. A multi-generation (e.g. 100 generations) evolution cycle is performed. Fitness evaluation is required in each generation. According to strategies such as roulette selection, individuals with higher fitness are selected to enter the next generation, and model assumptions that can more accurately predict the historical true thermal plume position and intensity have a greater probability of being retained.

[0061] Further, the selected excellent individuals will exchange chromosome fragments to produce new offspring. For example, a model that performs well in directionality (α, β) may be combined with another model that performs well in the influence range (Range), thereby producing a new model that may be more accurate in all aspects of the atrium characteristics.

[0062] Then the concept of mutation is introduced, and a certain parameter value on the chromosome of the offspring is randomly changed with a small probability. This corresponds to a small range of mutation exploration of the existing model assumptions, which helps to jump out of the local optimal solution and increases the possibility of finding the global optimal solution.

[0063] After hundreds of generations of evolution, most individuals in the population will converge to a state with very high fitness. At this time, the individual with the highest fitness in the algorithm population, which carries the {α, β, C0, C, Range} parameter combination, is considered to be the optimal global solution describing the spatial correlation of the rising air flow in the current building environment, i.e. the optimal model parameters. This set of parameters will be adopted as the initial optimal baseline model.

[0064] (3) The third stage:

[0065] The third stage is the stage that the system provided by the embodiment of the present application enters after being used, and its purpose is to adjust the optimal model parameters online through periodic global optimization and online fine-tuning mode using real-time data.

[0066] Wherein the purpose of online fine-tuning mode is to achieve high-frequency real-time adaptation, taking into account the dynamic changes of passenger flow and business mode, at each moment, the above-identified optimal baseline model parameters are taken as the starting point, and the data set (for example, 72 hours) in the current period is taken as the batch input to perform one-step simple stochastic gradient descent (SGD) optimization. The goal of this step is to make small adjustments to the five parameters according to the data of the last few minutes to quickly respond to real-time environmental fluctuations. This process has very small calculation amount, ensuring real-time adaptability.

[0067] Wherein the purpose of periodic global optimization is low-frequency baseline calibration, and the second stage global model identification based on genetic algorithm is repeated once every long period (for example, 72 hours). This process uses the latest large amount of historical data to perform a thorough recalibration of the optimal baseline model to respond to possible structural changes in the environment (such as seasonal changes, special events), and prevent model drift caused by noise accumulation in the online fine-tuning process.

[0068] Preferably, in the embodiments of the present application, considering that the particle concentration data and thermal energy data have their own dimensional problems, therefore, the particle concentration data and thermal energy data are normalized in their respective dimensions to obtain normalized particle concentration data and normalized thermal energy data. That is, the value range of the normalized data in the respective dimensions is between 0 and 1, without the influence of the dimension, and can be compared. The difference between the normalized thermal energy data and the normalized particle concentration data is taken as the pollution plume mode index.

[0069] The normalization method adopted in the embodiments of the present application is range standardization, that is, the maximum and minimum values in the respective dimensions are counted, and then normalized.

[0070] The features quantified by the detection data analysis module 102 can be used to specify the ventilation decision in the ventilation decision module 103. In the embodiment of the present application, the control strategy is executed by a plurality of independently controllable air supply units installed under the ceiling of the first floor and arranged along the edge of the atrium, each unit having the ability to independently adjust the air supply speed and air supply temperature. The real-time rising air flow tendency map is divided to obtain a plurality of connected domains. Each connected domain can be regarded as a local area that needs to be ventilated. The larger the pixel value in the local area, the greater the temperature in the upper part relative to the bottom dining area, which inhibits the upward movement of the hot plume and requires an increase in the control of the air supply unit on the temperature difference. The larger the area of the connected domain, the larger the scale of the local area, which requires a larger air speed to quickly solve the ventilation problem. Therefore, the air supply unit executes the control strategy for each connected domain; the control strategy is determined according to the pollution plume mode index, and the control strategy includes a balance mode, a temperature adjustment mode, and a wind speed adjustment mode. In the temperature adjustment mode, the target air supply temperature difference in the control strategy is determined according to the pixel value in the connected domain; in the wind speed adjustment mode, the target air supply speed is determined according to the area of the connected domain. Wherein, the larger the pollution plume mode index, the more the indoor belongs to the characteristics of strong heat and weak particles at this time, such as steam and other forms, which need to be cooled by refrigeration, so the control strategy needs to be adjusted to the temperature adjustment mode, and the larger the pixel value of the connected domain, the greater the need to increase the size of the target air supply temperature difference; the smaller the pollution plume mode index, the more it belongs to weak heat and strong particles at this time, such as frying oil fume, which indicates that the buoyancy of the hot plume is not strong, and a large number of oil smoke particles or odor molecules are carried, which need to be ventilated as soon as possible, so the control strategy should start the wind speed adjustment mode, horizontally capture and push to the exhaust port, preferentially increase the air supply speed to provide enough momentum, form an air flow barrier, and the larger the area of the connected domain, the larger the required wind speed; if the pollution plume mode index is moderate, there is no temperature difference priority or wind speed priority at this time, and the temperature difference and the wind speed can be adjusted at the same time, and the balance mode is used for ventilation treatment.

[0071] Preferably, in the embodiment of the present application, in order to eliminate weak background noise, only significant features with actual physical meaning are concerned, in the real-time rising air flow trend diagram, data less than a preset response threshold is set to 0 to obtain a denoising result diagram; the denoising result diagram is segmented by using a connected domain marking algorithm to obtain the connected domain. In the embodiment of the present application, the response threshold is set to 0.2 degrees Celsius. It should be noted that, in the embodiment of the present application, negative values and smaller values in the real-time rising air flow trend diagram are set to 0, only features with obvious suppression of the rising thermal plume are reserved, and the local area ventilation problem is more targetedly processed, that is, the obtained connected domain is a region composed of other non-0 pixel points, and a region with a pixel value of 0 is considered to be normally ventilated without intervention of the air supply unit. The obtained connected domain is quantified to record its centroid coordinates, peak pixel value, and area, which are used for ventilation control decision-making.

[0072] Preferably, in the embodiment of the present application, the control strategy is determined according to the pollution plume mode index, including:

[0073] If the pollution plume mode index is greater than a preset first threshold, the control strategy is a temperature adjustment mode; if the pollution plume mode index is less than a preset second threshold, the control strategy is a wind speed adjustment mode; if the pollution plume mode is greater than or equal to the second threshold and less than or equal to the first threshold, the control strategy is a standard balance strategy; and the first threshold is greater than the second threshold. In the embodiment of the present application, the first threshold is set to 0.5, and the second threshold is set to -0.5.

[0074] Preferably, in the embodiment of the present application, the method for obtaining the air supply target temperature difference includes:

[0075] The maximum pixel value in the connected domain is taken as a peak intensity, the peak intensity is multiplied by a preset coefficient to obtain the air supply target temperature difference. The preset coefficient can be set according to the cooling capacity of the air supply unit, and the air supply target temperature difference finally determines the opening degree of the air supply unit refrigerant valve. Therefore, the setting of the coefficient needs to ensure that the obtained maximum value cannot exceed the cooling capacity of the air supply unit. The specific case can be set, and the embodiment of the present application can set the maximum value in the allowed range. In other embodiments, a standard value can also be taken, and details are not described herein.

[0076] Preferably, in the embodiment of the present application, the method for obtaining the air supply target wind speed includes:

[0077] The area of the connected domain is normalized to obtain an area weight, and the product of the area weight and a preset maximum wind speed adjustment amount is multiplied to obtain a real-time wind speed adjustment amount; and the sum of the basic wind speed and the real-time wind speed adjustment amount is taken as the target air supply wind speed. Similarly, the preset maximum wind speed adjustment amount needs to be specifically set according to the specific capacity of the air supply unit, and will not be repeated here.

[0078] Preferably, in the embodiment of the present application, the control strategy can be executed by selecting a preset number of air supply units closest to the centroid coordinates of each connected domain. Each control strategy can be set to execute for two minutes; or the total energy of the hot spot is dynamically associated, that is, the total energy is the product of the peak intensity and the area of the connected domain, and the greater the total energy, the longer the control strategy is executed. During the execution period, the selected single or multiple air supply units are forced to execute the set control strategy. The preset number can be specifically set according to the distribution of the air supply unit according to the actual implementation, which is not limited here.

[0079] In summary, the present application uses a monitoring data acquisition module to obtain basic monitoring data. According to the collected temperature data, the best unbiased estimate value under each coordinate is obtained by using the spherical semi-variogram function for Kriging interpolation, and an updraft tendency map is formed. The real-time updraft tendency map is segmented to determine a plurality of connected domains, the current indoor environment pollution plume mode index is evaluated by comparing the particle concentration data and thermal energy data, and then the control strategy mode is determined. Under each control strategy mode, the regulation and control parameters can be determined according to the area of the connected domain or the pixel value in the real-time updraft tendency map, to realize adaptive and targeted efficient and rapid ventilation control. The embodiment of the present application realizes dynamic, low disturbance and low delay ventilation control in large commercial indoor environment through data monitoring and feature analysis, and constructs a real-time updraft tendency map.

[0080] It should be noted that the above-mentioned embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or can be advantageous.

[0081] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

Claims

1. A multi-parameter real-time monitoring system for large indoor ventilation systems, characterized in that, The system comprises: a monitoring data acquisition module for acquiring particle concentration data at a particle source in a lower-level restaurant, heat flux data at a heat source, and temperature data at various height coordinates in the vertical direction; a monitoring data analysis module for performing Kriging interpolation on the temperature data using a spherical semi-variogram function to obtain optimal unbiased estimates at various coordinates and form a real-time updraft tendency map; comparing the particle concentration data and the heat flux data to obtain a pollution plume mode index; a ventilation decision module for segmenting the real-time updraft tendency map to obtain multiple connected domains; controlling a supply air unit to execute a control strategy for each connected domain; determining the control strategy according to the pollution plume mode index, wherein the control strategy includes a balance mode, a temperature adjustment mode, and a wind speed adjustment mode; in the temperature adjustment mode, determining a target temperature difference in the control strategy according to the pixel value in the connected domain; in the wind speed adjustment mode, determining a target wind speed according to the area of the connected domain; the method for obtaining the pollution plume mode index comprises: normalizing the particle concentration data and the heat flux data in their respective dimensions to obtain normalized particle concentration data and normalized heat flux data; and taking the difference between the normalized heat flux data and the normalized particle concentration data as the pollution plume mode index; determining the control strategy according to the pollution plume mode index comprises: if the pollution plume mode index is greater than a preset first threshold, the control strategy is the temperature adjustment mode; if the pollution plume mode index is less than a preset second threshold, the control strategy is the wind speed adjustment mode; if the pollution plume mode index is greater than or equal to the second threshold and less than or equal to the first threshold, the control strategy is a standard balance strategy; the first threshold is greater than the second threshold.

2. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 1, wherein, the real-time updraft tendency map comprises: constructing a two-dimensional coordinate system based on the indoor horizontal plane, wherein the two-dimensional coordinate system includes real observation points, physical constraint points, and unknown data points; the coordinate value of a real observation point is: at each time in a preset real-time time window, dividing the various heights in the vertical direction into three categories: middle height, upper height, and lower height; taking the average temperature data between the lower height and the upper height as reference temperature data, and taking the difference between the temperature data of the middle height and the reference temperature data as an instantaneous buoyancy signal value at each time; selecting the maximum instantaneous buoyancy signal value in the real-time time window as the coordinate value of the real observation point at the real-time time; the physical constraint points are indoor air supply outlets and indoor boundaries, and the coordinate values are preset fixed values; in the two-dimensional coordinate system, performing Kriging interpolation according to the spherical semi-variogram function that has been learned to obtain the unbiased estimates as the coordinate values of the unknown data points; the real observation points, the physical constraint points, and the interpolated unknown data points form the real-time updraft tendency map.

3. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 2, wherein, The model parameters of the spherical semi-variogram are learned by using a three-stage evolution process, the first stage is used to accumulate historical data by preset model parameter values; the second stage is used to globally optimize the model parameters based on a genetic algorithm to determine optimal model parameters; and the third stage is used to online adjust the optimal model parameters by periodic global optimization and online fine tuning using real-time data.

4. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 1, wherein, The real-time updraft tendency map is segmented to obtain a plurality of connected domains, including: In the real-time updraft tendency map, data less than a preset response threshold is set to 0 to obtain a denoising result map; and the connected domain segmentation is performed on the denoising result map by using a connected domain labeling algorithm to obtain the connected domain.

5. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 1, wherein, The method for obtaining the target air supply temperature difference includes: The maximum pixel value in the connected domain is taken as a peak intensity, and the peak intensity is multiplied by a preset coefficient to obtain the target air supply temperature difference.

6. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 1, wherein, The method for obtaining the target air supply wind speed includes: The area of the connected domain is normalized to obtain an area weight, and the product of the area weight and a preset maximum wind speed adjustment amount is multiplied to obtain a real-time wind speed adjustment amount; and the sum of a basic wind speed and the real-time wind speed adjustment amount is taken as the target air supply wind speed.

7. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 1, wherein, The control strategy is executed by selecting a preset number of air supply units closest to the centroid coordinates of each connected domain.

8. The multi-parameter real-time monitoring system of a large indoor ventilation system according to claim 1, wherein, The particle concentration data is the average concentration data of the exhaust main pipeline in a real-time time window; the thermal energy data is the average power of the total active power detected on the main circuit of the total distribution box in the real-time time window; and the range of the real-time time window is one minute before the real-time time.

Images