Intelligent venue air conditioner regulation and control method and system based on distributed sensor network

Abstract

The invention provides a stadium air conditioner intelligent regulation and control method and system based on a distributed sensor network. Comprising the following steps: acquiring multi-source monitoring data and three-dimensional coordinates of fixed and mobile acquisition terminals in real time; a stadium three-dimensional obstacle model is constructed, a three-dimensional dynamic environment field is generated based on obstacle constraints, and the shortest path length of bypassing an obstacle is adopted as distance measurement of spatial correlation to eliminate physical penetration artifacts; based on the initial field, a virtual extreme value injection method and a total variation regularization strategy are adopted to analyze structure mutation features, the optimal sampling position of the mobile terminal is calculated, and scheduling collection is carried out to update the environment field; and finally, a prediction model is constructed based on the updated environment field, rolling time domain optimization is executed by adopting a model prediction control strategy, and an air conditioner power distribution instruction is generated. The problems that a large venue environment monitoring blind area is large, model physical distortion exists, and control lags are solved, and accurate reconstruction of a three-dimensional environment field and energy-saving and comfortable regulation and control of an air conditioning system are achieved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent control and Internet of Things, and particularly relates to a venue air conditioner intelligent regulation and control method and system based on a distributed sensing network. BACKGROUND

[0002] The spatial structure of large venues (such as stadiums, exhibition halls, factory workshops, shopping malls, etc.) is huge, and the internal air flow and heat exchange process is extremely complex, with obvious non-uniform phenomena of temperature field and humidity field. At present, the venue air conditioning system mostly adopts simple feedback control or preset program control based on a few fixed point sensors, which is difficult to truly and comprehensively reflect the real-time environmental state of the entire space. This extensive control method is easy to cause local environment to be too cold, too hot or too humid, and at the same time, in order to cover the edge area, it often causes excessive consumption of energy, which does not conform to the green and energy-saving development trend.

[0003] The existing large venue air conditioning control technology, such as Chinese patent application CN118532799A, discloses a large venue central air conditioning system variable static pressure variable control object variable air volume control method, which usually adopts a variable air volume system, adjusts the frequency of the air supply fan and the opening degree of the air valve by monitoring the deviation of the temperature in the return air branch pipe from the set value, and uses a PID algorithm. However, such a scheme has a serious lag in the data collected by the sensor fixedly installed in the high return air pipe, which is far behind the real temperature of the ground personnel activity area; and the single fixed measuring point cannot sense the local heat island or cold spot caused by the uneven distribution of heat sources in the venue interior due to shielding, resulting in a passive response state of the control system, which makes it difficult to achieve precise comfort management. On the other hand, although mobile inspection robots have been introduced into dynamic environment monitoring, such as Chinese patent application CN120779748A, which discloses an intelligent decision-making and control method and system for dynamic environment inspection robots, it realizes the task response and obstacle avoidance of the robot in a complex environment by constructing a patrol domain geometry and executing multi-threaded decision-making. However, the path planning of the existing robot is mostly based on geometric structure constraints or preset task logic, and lacks the ability to deeply model continuous environmental scalar fields, resulting in path redundancy or missing of key data when collecting environmental data, which cannot directly support the high-precision air conditioning group control demand.

[0004] In summary, there is a lack of an environmental regulation and control method in the prior art that can break through the limitations of fixed sensor positions, overcome the lag of return air detection, and actively eliminate three-dimensional monitoring blind spots through intelligent mobile sampling strategies. SUMMARY

[0005] In order to solve the technical problems of large venue environment monitoring blind area, model physical distortion and control lag in the prior art, a low-cost, high-precision and easy-to-deploy venue air conditioning intelligent regulation and control system and method are provided. The system collects environmental data in real time through widely distributed micro-sensing terminals, constructs a dynamic environment field, thereby realizing accurate on-demand regulation and control of the air conditioning system, while ensuring human comfort and significantly reducing energy consumption.

[0006] The application adopts the following technical solutions.

[0007] In a first aspect, a venue air conditioning intelligent regulation and control method based on a distributed sensing network comprises: Step 1: Obtain fixed monitoring data uploaded by a plurality of fixed collection terminals distributed in the venue and corresponding fixed three-dimensional coordinates; Step 2: Construct a three-dimensional obstacle model containing the physical structure of the venue; based on the fixed monitoring data and the fixed three-dimensional coordinates, and using the shortest path length around the three-dimensional obstacle model as the distance measure for spatial correlation calculation, generate a three-dimensional dynamic environment field; Step 3: Based on the current three-dimensional dynamic environment field, calculate the optimal sampling position of the mobile collection terminal; dispatch the mobile collection terminal to the optimal sampling position for collection, and update the three-dimensional dynamic environment field according to the collected mobile monitoring data and corresponding mobile three-dimensional coordinates; Step 4: Based on the updated three-dimensional dynamic environment field, use model predictive control strategy to perform rolling horizon optimization, generate power distribution instructions for air conditioning equipment and execute.

[0008] Preferably, in step 1, the fixed collection terminal comprises a microcontroller integrated with wireless communication function, a temperature and humidity sensing module, a rechargeable lithium polymer battery and a shell; the microcontroller, the temperature and humidity sensing module and the rechargeable lithium polymer battery are arranged in the shell; the microcontroller is electrically connected with the temperature and humidity sensing module and the rechargeable lithium polymer battery respectively; the bottom surface of the shell is provided with a magnetic patch; Obtaining the fixed three-dimensional coordinates corresponding to each fixed collection terminal comprises: obtaining scanning information of a device identification code located on the surface of the shell, determining an installation position point according to input operation on a visual management interface map, binding coordinate data of the installation position point with the device identification code to obtain the fixed three-dimensional coordinates.

[0009] Preferably, in step 2, the three-dimensional dynamic environment field is generated by using a reproducing kernel function interpolation method based on three-dimensional obstacle model constraints, which specifically comprises: Discretize the venue space into a three-dimensional voxel grid; identifying voxels occupied by the three-dimensional obstacle model in the three-dimensional voxel grid and marking as non-traversable voxels, and marking the remaining voxels as traversable voxels; For any one to be interpolated voxel in the three-dimensional voxel grid, a graph search algorithm is used to search a connected path between it and the voxels where each fixed acquisition terminal is located, the connected path only includes traversable voxels, and the physical length of the connected path is determined as the shortest path length; A spatial correlation model is constructed using a Gaussian kernel function, which is used to calculate the spatial correlation value by taking a natural constant as the base number, taking the ratio of the negative of the square of the shortest path length to twice the square of the preset bandwidth parameter as the exponent, and taking the power value; the interpolation weight is determined based on the spatial correlation value, and the fixed monitoring data uploaded by each fixed acquisition terminal is weighted and summed to obtain the temperature and humidity estimate value of the to-be-interpolated voxel; The three-dimensional voxel grid is traversed, and the three-dimensional dynamic environment field is generated based on the set of temperature and humidity estimate values of all to-be-interpolated voxels.

[0010] Preferably, in step 3, the mobile acquisition terminal is arranged on an unmanned mobile platform, and includes an indoor positioning module; the mobile monitoring data and the corresponding mobile three-dimensional coordinates when the mobile acquisition terminal runs in the venue are obtained, including: calculating the three-dimensional coordinate data of the mobile acquisition terminal in the venue coordinate system by using the indoor positioning module and uploading.

[0011] Preferably, in step 3, calculating the optimal sampling position of the mobile acquisition terminal specifically includes: A virtual detection value is set, and the value of the virtual detection value is set to be greater than the maximum value in the fixed monitoring data or less than the minimum value in the fixed monitoring data; For any one candidate sampling position in the venue space, an incremental update field function is constructed, the incremental update field function is composed of the three-dimensional dynamic environment field and a correction term; the correction term is the product of a base function update term and a prediction residual, wherein the prediction residual is the difference between the virtual detection value and the interpolation value of the three-dimensional dynamic environment field at the candidate sampling position; the base function update term is a spatial distribution function composed of a kernel function for the candidate sampling position and a linear combination of kernel functions for each fixed acquisition terminal; The total variation norm of the incremental update field function is calculated, and the total variation norm is the integral of the modulus of the gradient of the incremental update field function in the venue space; The candidate sampling positions in the venue space are traversed, and the candidate sampling position that makes the total variation norm reach the minimum value is selected as the optimal sampling position.

[0012] Preferably, in step 4, the model predictive control strategy is used to perform rolling horizon optimization, specifically including: A non-steady-state heat conduction equation and a fluid dynamics equation are used to establish a venue temperature and humidity prediction model with the updated three-dimensional dynamic environment field as an initial condition; An optimal objective function is constructed, which includes a square term of a comprehensive control deviation and an air conditioning equipment energy consumption term. Under the condition of meeting the physical constraints of the air conditioning equipment, a future control input sequence is solved based on the temperature and humidity prediction model to minimize the optimal objective function; The first instruction in the future control input sequence is issued as the power distribution instruction to the air conditioning equipment for execution, and in the next control period, the temperature and humidity prediction model is feedback corrected using newly acquired fixed monitoring data and mobile monitoring data, and rolling horizon optimization is performed again.

[0013] Preferably, the calculation method of the comprehensive control deviation specifically includes: The PMV value of each voxel in the three-dimensional voxel grid is calculated using a predicted mean vote model; For any air conditioning control domain, the comprehensive control deviation in the prediction time domain is calculated, which is the weighted sum of the average of the PMV values of all three-dimensional voxel grids in the air conditioning control domain, the maximum of the absolute values of the PMV values, and the personnel density factor.

[0014] In a second aspect, a venue air conditioning intelligent regulation and control system based on a distributed sensing network, which runs the venue air conditioning intelligent regulation and control method based on the distributed sensing network, includes: A multi-source data aggregation module is configured to acquire fixed monitoring data uploaded by a plurality of fixed collection terminals distributed in the venue and corresponding fixed three-dimensional coordinates; An environment field reconstruction module is configured to construct a three-dimensional obstacle model including a venue entity structure, generate a three-dimensional dynamic environment field based on the fixed monitoring data and the fixed three-dimensional coordinates, and use the shortest path length around the three-dimensional obstacle model as a distance measure for spatial correlation calculation; A mobile enhanced perception module is configured to calculate an optimal sampling position of a mobile collection terminal based on the current three-dimensional dynamic environment field, dispatch the mobile collection terminal to the optimal sampling position for collection, and update the three-dimensional dynamic environment field based on the collected mobile monitoring data and corresponding mobile three-dimensional coordinates; A collaborative regulation and control decision module is configured to perform rolling horizon optimization using a model predictive control strategy based on the updated three-dimensional dynamic environment field, generate a power distribution instruction for the air conditioning equipment, and issue the instruction for execution.

[0015] In a third aspect, a terminal includes a processor and a storage medium; The storage medium is configured to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method.

[0016] Fourthly, a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0017] The beneficial effects of this invention are as follows: 1. High Precision and Realism: By collaboratively collecting multi-source environmental data through distributed fixed and mobile acquisition terminals, and combining this with an obstacle-constrained regenerative kernel function interpolation method, a dynamic environmental field that realistically reflects the three-dimensional spatial distribution of the venue is constructed. This effectively overcomes the "wall artifact" problem caused by neglecting physical obstacles in traditional models, thus improving the accuracy of environmental perception. Fully utilizing the "fill-in" function of mobile data acquisition terminals, data is collected from locations with high uncertainty, further reducing the discrete error of the interpolated dynamic environmental field.

[0018] 2. High efficiency and energy saving: Based on data, a three-dimensional dynamic environmental field and fluid dynamics mechanism model are established to form a predictive model that evolves over time. Then, a model predictive control strategy is adopted to perform rolling time-domain optimization, so as to achieve precise allocation of air conditioning power, avoid energy waste caused by monitoring blind spots or lag in traditional control methods, and significantly improve energy efficiency.

[0019] 3. Low cost and easy deployment: The data acquisition terminal uses common, low-cost components and batteries for power, has a compact structure, is easy to install, and requires no complex wiring, making it ideal for large-scale, rapid deployment in existing venues. The mobile terminal reuses existing unmanned platforms, reducing the overall deployment and maintenance costs of the system.

[0020] 4. Intelligent and Adaptive: By actively identifying the most likely abrupt change regions in the environmental structure through the virtual extreme value injection method and scheduling mobile terminals for enhanced sampling, the system has the ability to actively perceive and adaptively optimize. It can dynamically respond to environmental changes such as personnel flow, changes in sunlight, and equipment heat dissipation, and achieve adaptive and dynamic intelligent control to always maintain the environment in the optimal comfort range. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the architecture of the intelligent control system for venue air conditioning based on a distributed sensor network provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0023] Example 1: This invention provides a method for intelligent control of venue air conditioning based on distributed sensor networks, comprising: Step 1: Obtain fixed monitoring data and corresponding fixed three-dimensional coordinates uploaded by multiple fixed acquisition terminals distributed within the venue; wherein, the fixed monitoring data includes real-time temperature data and humidity data.

[0024] In step 1, the fixed acquisition terminal includes a microcontroller with integrated wireless communication function, a temperature and humidity sensing module, a rechargeable lithium polymer battery, and a housing; the microcontroller, the temperature and humidity sensing module, and the rechargeable lithium polymer battery are disposed inside the housing; the microcontroller is electrically connected to the temperature and humidity sensing module and the rechargeable lithium polymer battery respectively; a magnetic patch is provided on the bottom surface of the housing; Obtaining the fixed three-dimensional coordinates corresponding to each fixed acquisition terminal includes: obtaining the scanned information of the device identification code located on the surface of the housing; determining the installation location point based on the input operation on the map of the visual management interface; binding the coordinate data of the installation location point with the device identification code to obtain the fixed three-dimensional coordinates.

[0025] Specifically, in this embodiment, to adapt to the complex installation environment of large venues, the fixed data acquisition terminal adopts a miniaturized integrated hardware design. The terminal uses a low-power IoT microcontroller as its core (e.g., an ESP8266 or ESP32 module with integrated Wi-Fi / Bluetooth functionality), connected to a temperature and humidity sensing module. This sensing module can use a Pt100 platinum resistance thermometer specifically for high-precision temperature acquisition, and a DHT22 (or SHT30, etc., industrial-grade) sensor specifically for humidity acquisition, reducing the error impact of a single sensor through the combination of heterogeneous sensors. The Pt100 platinum resistance thermometer is connected to the microcontroller's communication interface via a signal conditioning circuit (e.g., a MAX31865 conversion module) to achieve high-precision analog signal digitization. To support long-term wireless operation, the power supply unit uses a 3.7V, 200mAh rechargeable lithium polymer battery. All components are encapsulated in a portable ABS plastic shell, preferably with a miniaturized size of 60mm × 40mm × 20mm. The bottom of the box is fitted with a strong magnetic patch, allowing it to be directly attached to the metal beams, columns, supports, or wall surfaces of the venue without the need for drilling; the casing can also be pre-drilled with hanging holes or 3M adhesive mounting points to accommodate installation on non-metallic surfaces.

[0026] During the deployment phase, to establish the correspondence between the physical location of each fixed data acquisition terminal and the virtual data, a visual interactive calibration method is used to obtain its fixed three-dimensional coordinates. The specific operation process is as follows: At the installation site, staff use a handheld device (such as a PDA or smartphone) to scan the device identification code located on the surface of the fixed data acquisition terminal's casing; then, they open the venue's visual two-dimensional map or three-dimensional BIM model in the handheld device's APP and click on the physical installation location point corresponding to the terminal on the map; when using a two-dimensional map, the planar coordinates (a, b) are obtained based on the clicked location, and combined with the preset installation height or the height value input on site, the height coordinates (c) are generated; the three-dimensional coordinate data (a, b, c) of the location point are automatically bound to the device identification code and uploaded to the server database, thereby completing the initial configuration.

[0027] Through the aforementioned fixed data acquisition network, the system can aggregate multi-source heterogeneous environmental data covering the entire venue in real time, providing basic data support for subsequent environmental field reconstruction.

[0028] Step 2: Construct a three-dimensional obstacle model containing the physical structure of the venue; based on the fixed monitoring data and the fixed three-dimensional coordinates, and using the shortest path length around the three-dimensional obstacle model as the distance metric for spatial correlation calculation, generate a three-dimensional dynamic environment field; In step 2, generating the three-dimensional dynamic environment field specifically includes: Discretize the venue space into a three-dimensional voxel grid; Identify the voxels in the three-dimensional voxel mesh that are occupied by the three-dimensional obstacle model and mark them as impassable voxels, and mark the remaining voxels as passable voxels; For any voxel to be interpolated in the three-dimensional voxel grid, a graph search algorithm is used to search for a connected path between it and the voxels where each fixed acquisition terminal is located. The connected path only includes passable voxels, and the physical length of the connected path is determined as the shortest path length. A Gaussian kernel function is used to construct a spatial correlation model, which is used to calculate the power value of the spatial correlation value by using the natural constant as the base and the ratio of the negative square of the shortest path length to twice the square of the preset bandwidth parameter as the exponent. Based on the spatial correlation value, the interpolation weight is determined, and the fixed monitoring data uploaded by each fixed acquisition terminal is weighted and summed to obtain the estimated temperature and humidity value of the voxel to be interpolated. The three-dimensional voxel mesh is traversed, and the three-dimensional dynamic environment field is generated based on the set of temperature and humidity estimates of all voxels to be interpolated.

[0029] Specifically, in this embodiment, the server receives monitoring datasets uploaded by N fixed acquisition terminals in real time, performs spatiotemporal alignment and cleaning, and forms a standardized discrete observation dataset.

[0030] To address the impact of vertical thermal stratification and physical partitions in large venues, a 3D obstacle model is constructed by pre-importing the venue's BIM model or architectural drawings, including solid structures such as walls, large equipment, stands, and columns. Based on this, the continuous airspace within the venue is discretized into a 3D voxel mesh. For example, the mesh resolution is set to 0.5m × 0.5m × 1.0m (this size can be adaptively adjusted according to the venue's spatial scale). During the meshing process, binarization is performed to identify voxels occupied by solid structures and mark them as "impassable voxels"; the remaining free space voxels are marked as "passable voxels."

[0031] Next, a three-dimensional dynamic environment field is generated using the regenerative kernel interpolation method. Unlike traditional algorithms that use Euclidean straight-line distance, this embodiment uses the shortest path length around obstacles as the distance metric. Specifically, for any voxel to be interpolated in the mesh, the system uses a graph search algorithm (such as Dijkstra's algorithm or A* algorithm, etc., shortest path search algorithms) to search for a connected path between the voxel to be interpolated and the voxels where each fixed acquisition terminal is located in the three-dimensional voxel mesh. During the search, the path is limited to passing only through "accessible voxels". The sum of the physical distances between the center points of all adjacent voxels (including face-adjacent or corner-adjacent) in this connected path is calculated and determined as the shortest path length between the two points. If two points are completely blocked by an obstacle (no connecting path), then the distance is set to infinity.

[0032] A spatial correlation model is constructed using the Gaussian kernel function, and is expressed as follows:

[0033] in, The preset bandwidth parameters, This represents the spatial correlation value. It is an exponential function with the natural constant e as the base. Spatial correlation within the formula decreases exponentially with the square of the distance. This is especially true when there are obstacles between two points causing detours. When increasing, the calculated The value will approach 0, thus mathematically severing the thermal correlation between the two sides of the obstacle and achieving the effect of preventing wall penetration.

[0034] Finally, the server uses the correlation values ​​between each fixed terminal and the interpolation point calculated above. Calculate the normalized interpolation weights. Specifically, let the spatial correlation value between the i-th fixed acquisition terminal and the current voxel to be interpolated be denoted as . Then the normalized interpolation weight corresponding to the i-th fixed acquisition terminal The calculation is as follows:

[0035] Where N is the total number of fixed data acquisition terminals. This is the sum of the correlation values ​​for this voxel from all fixed acquisition terminals. If the voxel has no connection path to any terminal (denominator is 0), the temperature at that point is marked as invalid or uses the default value. Subsequently, the monitoring data from each fixed acquisition terminal are weighted and summed to obtain the estimated temperature and humidity value of the voxel to be interpolated. :

[0036] in, This refers to the real-time temperature and humidity monitoring data uploaded by the i-th fixed acquisition terminal. By traversing all voxels in the 3D voxel grid and repeating the above process, a 3D dynamic environmental field covering the entire venue can be generated.

[0037] This embodiment, by adding a weight to the detour distance, essentially focuses on environmental field reconstruction based on air connectivity. This effectively avoids unreasonable smoothing interference between data from different functional areas (such as enclosed boxes and open stands) at the algorithm level, and outputs the data to the venue's air conditioning control system in real time with a high response speed, serving as a direct basis for environmental adjustment; it also serves as the basic data source for in-depth optimization analysis in step 3.

[0038] Step 3: Based on the current 3D dynamic environment field, the optimal sampling position of the mobile acquisition terminal is calculated; the mobile acquisition terminal is then dispatched to the optimal sampling position to collect data. After collection, the mobile acquisition terminal transmits the mobile monitoring data, including real-time coordinates, back to the server. The server performs spatiotemporal alignment and fusion processing on the newly received incremental mobile data and the existing fixed monitoring data to form a high-confidence fused dataset. Finally, this fused dataset is used to correct and update the 3D dynamic environment field.

[0039] The mobile data acquisition terminal is installed on an unmanned mobile platform and includes an indoor positioning module; acquiring the mobile monitoring data and corresponding three-dimensional coordinates of the mobile data acquisition terminal while it is running in the venue includes: using the indoor positioning module to calculate the three-dimensional coordinate data of the mobile data acquisition terminal in the venue coordinate system and uploading it.

[0040] In this embodiment, the mobile terminal is mounted on a mobile platform such as an automated guided vehicle (AGV) or an automated inspection robot, and uses a LiDAR SLAM module or an ultra-wideband (UWB) positioning module to calculate and feed back its dynamic three-dimensional position coordinates in the venue coordinate system in real time during operation within the venue. The mobile data acquisition terminal can be equipped with an independent high-capacity battery pack to support the high power consumption operation of the positioning module, or it can be directly connected to the power system of the mobile platform through a power interface to ensure power endurance during long-term inspection tasks.

[0041] Step 3, calculating the optimal sampling position of the mobile acquisition terminal specifically includes: Although step 2 has generated a discrete voxel mesh field, it is necessary to transform the discrete field into a continuous analytical model in order to perform accurate gradient calculations and variational analysis to identify environmental blind spots. This step directly uses the detour distance metric defined in step 2. The environmental field is characterized using a regenerative kernel interpolation model (such as Gaussian kernel RBF) that includes a linear trend term. In this case, the three-dimensional dynamic environmental field is based on data from N fixed acquisition terminals. Represented as:

[0042] in, For any coordinate within the venue space, Let i be the coordinates of the i-th fixed acquisition terminal. These are the interpolation weighting coefficients. The Gaussian kernel function defined in step 2 is used, and its independent variable is kept as the shortest path distance around the obstacle to inherit the anti-wall-penetration property; is a linear trend term used to capture the global gradient changes in temperature and humidity within the venue, where b is the trend coefficient vector and a is the intercept.

[0043] To identify blind spots in the environment, an extreme potential anomaly is simulated. Virtual detection values ​​are set. The value is set to be greater than the maximum value or less than the minimum value among the fixed monitoring data. In this embodiment, the fixed monitoring data set is statistically analyzed. Set as: or

[0044] in, The preset bias (e.g., 3 times the standard deviation of the current data) is designed to simulate an extreme potential outlier.

[0045] It should be noted that, although It was initially defined based on discrete voxels, but in the construction For any continuous coordinate x in space that is not at the center of a voxel, its corresponding coordinate can be obtained through trilinear interpolation or nearest-neighbor voxel mapping. Numerical values, or by mapping spatial coordinates x to their corresponding voxel indices for calculation.

[0046] For any candidate sampling location y within the venue space, construct an incremental update field function. To avoid repeatedly solving the entire system of equations containing N+1 nodes (the original N fixed points + 1 virtual point), this embodiment employs an incremental update strategy. The incrementally updated field function is derived from the three-dimensional dynamic environment field. The superimposed correction term is constituted as follows:

[0047] Wherein, the correction term is the basis function update term. Compared with the predicted residual The product of the virtual probe values, the prediction residual is the virtual probe value. The interpolated value of the three-dimensional dynamic environment field at the candidate sampling position y difference; The estimated temperature and humidity at candidate location y (i.e., the calculation result of step 2); the basis function update term The influence weighting function of the new observation data introduced at position y on the overall spatial distribution is a spatial distribution function composed of a linear combination of the kernel function for the candidate sampling position and the kernel function for each of the fixed acquisition terminals, expressed as:

[0048] in, These are polynomial basis functions (the linear combination form corresponds to the trend term part of the three-dimensional dynamic environment field). This represents the number of basis functions for the trend term (e.g., for a linear trend, the value is 4, corresponding to the spatial coordinate basis). , , All are linear combination coefficients (Lagrange multipliers) that depend only on position y, and their values ​​are obtained by solving the incremental interpolation equations, ensuring that... .

[0049] To determine the optimal position, the incremental update field function is calculated. The total variation norm, which is the integral of the magnitude of the gradient of the incremental update field function in the venue space; Traverse the candidate sampling locations within the venue space, and select the candidate sampling location that minimizes the total variation norm as the optimal sampling location, expressed as:

[0050] If the virtual detection value fluctuates wildly Injecting into a smooth, connected region will cause a drastic artificial distortion of the entire field function, resulting in an increase in the total variation. If the virtual probe value... When injected into structural boundary regions of the venue environment (such as the sides of walls and barriers, the edges of equipment heat dissipation, etc.), the overall variation of the function is minimized because these regions physically allow for large gradient abrupt changes. Therefore, the location that minimizes the total variation means that this location is the area within the venue most capable of accommodating gradient abrupt changes, which is also the environmental blind spot most easily overlooked by fixed sensors.

[0051] Finally, the scheduling system controls the mobile acquisition terminal to move to the optimal sampling position. And conduct on-site data collection to obtain real mobile monitoring data. Subsequently, using real... Replace the virtual probe value in the above formula And substitute it again into the incremental update field function. This allows us to obtain the final three-dimensional dynamic environmental field that integrates fixed and mobile monitoring data.

[0052] Step 4: Based on the updated three-dimensional dynamic environment field, a rolling time-domain optimization is performed using the Model Predictive Control (MPC) strategy to generate power allocation instructions for the air conditioning equipment and issue them for execution.

[0053] Step 4, which involves employing a model predictive control strategy to perform rolling time-domain optimization, specifically includes: By simplifying the equations using the unsteady heat conduction equation and the fluid dynamics (CFD) equation, and taking the updated three-dimensional dynamic environmental field as the initial condition, a prediction model describing the temperature and humidity inside the venue is established:

[0054] in, These are state variables, corresponding to the temperature and humidity values ​​of each voxel mesh (or aggregated control sub-region) in the updated three-dimensional dynamic environment field; To control the input variables, the corresponding operating state sequence of the air conditioning equipment (including fan speed, chilled water valve opening, and compressor frequency) is used. Here, A represents the disturbance variable (such as changes in outdoor temperature or fluctuations in human heat load); B and D are coefficient matrices determined through system identification or physical modeling. This model can predict the temperature and humidity change trajectories of various areas within the venue in the future control time domain under a given control input, with low computational cost.

[0055] Before optimization, the Predicted Average Voting (PMV) model was used as the core evaluation index. The PMV model comprehensively considers six factors: ambient temperature, relative humidity, air velocity, mean radiant temperature, intensity of human activity, and thermal resistance of clothing. In practice, ambient temperature and relative humidity are directly obtained from state variables. Air velocity is estimated based on the fan speed model of the air conditioning equipment; the average radiation temperature is approximately taken from the wall temperature or average air temperature around the voxel; the intensity of personnel activity and the thermal resistance of clothing are preset parameters according to the functional attributes of different areas of the venue (such as the audience seating area and the competition area).

[0056] Construct an optimization objective function that includes the square of the comprehensive control deviation and the energy consumption term of the air conditioning equipment. The calculation method for the comprehensive control deviation specifically includes: calculating the PMV value of each voxel in the three-dimensional voxel grid using the PMV model. For any air conditioning control domain, the comprehensive control deviation in the prediction time domain is calculated. The comprehensive control deviation is the weighted sum of the average PMV value of all three-dimensional voxel grids in the air conditioning control domain, the maximum absolute value of the PMV value, and the personnel density factor.

[0057] In this embodiment, it is represented as:

[0058] Where P is the prediction time domain, and k represents the index of the discrete time in the prediction time domain; This is the comfort index vector for the predicted time. The desired comfort level (typically 0); Let Q represent the square of the weighted Euclidean norm, where Q is the state weight matrix (usually a diagonal matrix) used to distinguish the importance of different areas within the venue. Specifically, to prioritize the experience in the core areas, the diagonal elements of matrix Q are dynamically assigned based on the real-time population density of the voxel's area: for high-density areas such as full stands, a larger weight coefficient is set, so that its PMV deviation dominates the objective function; Reflects air conditioning energy consumption; R represents the square of the weighted Euclidean norm, where R is the control weight matrix used to balance the energy consumption costs of different devices. Reflects the range of change in control actions (to avoid frequent equipment oscillations); , , These are the weighting coefficients.

[0059] In each control cycle, rolling time-domain optimization is performed. Under the physical constraints of the air conditioning equipment (such as damper opening degree 0-100%, inverter frequency upper limit, etc.), the future control input sequence that minimizes the optimization objective function is solved based on the temperature and humidity prediction model.

[0060] Following the MPC mechanism, the first instruction in the future control input sequence is sent to the air conditioning equipment as the power allocation instruction for execution, and then transmitted to the air conditioning control terminal via the Internet of Things (IoT) protocol. The control terminal then drives the air valve, fan, or refrigeration unit accordingly. In the next control cycle, the newly acquired fixed and mobile monitoring data are used to refine the temperature and humidity prediction model. Perform feedback correction and re-optimize the rolling time domain.

[0061] Example 2: like Figure 1 As shown, this invention provides a venue air conditioning intelligent control system based on a distributed sensor network, and the method for running the venue air conditioning intelligent control system based on a distributed sensor network includes: The multi-source data aggregation module is used to acquire fixed monitoring data and corresponding fixed three-dimensional coordinates uploaded by multiple fixed acquisition terminals distributed within the venue; The environmental field reconstruction module is used to construct a three-dimensional obstacle model containing the physical structure of the venue; based on the fixed monitoring data and the fixed three-dimensional coordinates, and using the shortest path length around the three-dimensional obstacle model as the distance metric for spatial correlation calculation, a three-dimensional dynamic environmental field is generated. The motion-enhanced sensing module calculates the optimal sampling position of the mobile acquisition terminal based on the current 3D dynamic environment field and schedules the mobile acquisition terminal to move to the optimal sampling position for acquisition. After acquisition, the mobile acquisition terminal transmits motion monitoring data containing real-time coordinates back to the multi-source data aggregation module on the server via a wireless network. This module performs spatiotemporal alignment and fusion processing on the newly received incremental motion data and the original fixed monitoring data, and inputs it into the environment field reconstruction module. Finally, the environment field reconstruction module corrects and updates the 3D dynamic environment field based on these high-confidence motion monitoring data and the corresponding motion 3D coordinates.

[0062] The collaborative control decision module is used to perform rolling time-domain optimization based on the updated three-dimensional dynamic environment field and the model predictive control strategy, generate power allocation instructions for air conditioning equipment and issue them for execution.

[0063] Example 3: A terminal, comprising a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method.

[0064] Example 4: A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0065] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0066] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0067] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0068] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for intelligent control of stadium air conditioning based on distributed sensor networks, characterized in that, include: Step 1: Obtain fixed monitoring data and corresponding fixed 3D coordinates uploaded by multiple fixed acquisition terminals distributed within the venue; Step 2: Construct a three-dimensional obstacle model containing the physical structure of the venue; based on the fixed monitoring data and the fixed three-dimensional coordinates, and using the shortest path length around the three-dimensional obstacle model as the distance metric for spatial correlation calculation, generate a three-dimensional dynamic environment field; Step 3: Based on the current three-dimensional dynamic environment field, calculate the optimal sampling position of the mobile acquisition terminal; The mobile acquisition terminal is dispatched to the optimal sampling location to collect data, and the three-dimensional dynamic environment field is updated based on the collected mobile monitoring data and the corresponding mobile three-dimensional coordinates. Step 4: Based on the updated three-dimensional dynamic environment field, a model predictive control strategy is used to perform rolling time-domain optimization, generate power allocation instructions for the air conditioning equipment, and issue them for execution.

2. The intelligent control method for venue air conditioning based on distributed sensor networks according to claim 1, characterized in that, In step 1, the fixed acquisition terminal includes a microcontroller with integrated wireless communication function, a temperature and humidity sensing module, a rechargeable lithium polymer battery, and a housing; the microcontroller, the temperature and humidity sensing module, and the rechargeable lithium polymer battery are disposed inside the housing; the microcontroller is electrically connected to the temperature and humidity sensing module and the rechargeable lithium polymer battery respectively; a magnetic patch is provided on the bottom surface of the housing; Obtaining the fixed three-dimensional coordinates corresponding to each fixed acquisition terminal includes: obtaining the scanned information of the device identification code located on the surface of the housing; determining the installation location point based on the input operation on the map of the visual management interface; binding the coordinate data of the installation location point with the device identification code to obtain the fixed three-dimensional coordinates.

3. The intelligent control method for venue air conditioning based on distributed sensor networks according to claim 1, characterized in that, In step 2, generating the three-dimensional dynamic environment field specifically includes: Discretize the venue space into a three-dimensional voxel grid; Identify the voxels in the three-dimensional voxel mesh that are occupied by the three-dimensional obstacle model and mark them as impassable voxels, and mark the remaining voxels as passable voxels; For any voxel to be interpolated in the three-dimensional voxel grid, a graph search algorithm is used to search for a connected path between it and the voxels where each fixed acquisition terminal is located. The connected path only includes passable voxels, and the physical length of the connected path is determined as the shortest path length. A Gaussian kernel function is used to construct a spatial correlation model, which is used to calculate the power value of the spatial correlation value by using the natural constant as the base and the ratio of the negative square of the shortest path length to twice the square of the preset bandwidth parameter as the exponent. Based on the spatial correlation value, the interpolation weight is determined, and the fixed monitoring data uploaded by each fixed acquisition terminal is weighted and summed to obtain the estimated temperature and humidity value of the voxel to be interpolated. The three-dimensional voxel mesh is traversed, and the three-dimensional dynamic environment field is generated based on the set of temperature and humidity estimates of all voxels to be interpolated.

4. The intelligent control method for venue air conditioning based on distributed sensor networks according to claim 1, characterized in that, In step 3, the mobile data acquisition terminal is installed on an unmanned mobile platform and includes an indoor positioning module; Acquiring mobile monitoring data and corresponding three-dimensional coordinates of the mobile acquisition terminal while it is running in the venue includes: using the indoor positioning module to calculate the three-dimensional coordinate data of the mobile acquisition terminal in the venue coordinate system and uploading it.

5. The intelligent control method for venue air conditioning based on a distributed sensor network according to claim 1, characterized in that, Step 3, calculating the optimal sampling position of the mobile acquisition terminal specifically includes: A virtual detection value is set, wherein the value of the virtual detection value is set to be greater than the maximum value of the fixed monitoring data or less than the minimum value of the fixed monitoring data; For any candidate sampling location within the venue space, an incremental update field function is constructed. The incremental update field function is composed of the superimposed correction term of the three-dimensional dynamic environment field. The correction term is the product of the basis function update term and the prediction residual, where the prediction residual is the difference between the virtual detection value and the interpolated value of the three-dimensional dynamic environment field at the candidate sampling location. The basis function update term is a spatial distribution function composed of a linear combination of the kernel function for the candidate sampling location and the kernel function for each of the fixed acquisition terminals. Calculate the total variation norm of the incremental update field function, where the total variation norm is the integral of the magnitude of the gradient of the incremental update field function over the venue space; The candidate sampling locations within the venue space are traversed, and the candidate sampling location that minimizes the total variation norm is selected as the optimal sampling location.

6. The intelligent control method for venue air conditioning based on a distributed sensor network according to claim 1, characterized in that, Step 4, which involves employing a model predictive control strategy to perform rolling time-domain optimization, specifically includes: Using the unsteady heat conduction equation and fluid dynamics equation, and taking the updated three-dimensional dynamic environmental field as the initial condition, a prediction model describing the temperature and humidity inside the venue is established. An optimization objective function is constructed, which includes the square of the comprehensive control deviation and the energy consumption of the air conditioning equipment. Under the condition of satisfying the physical constraints of the air conditioning equipment, the future control input sequence that minimizes the optimization objective function is solved based on the temperature and humidity prediction model. The first instruction in the future control input sequence is sent to the air conditioning equipment as the power allocation instruction for execution. In the next control cycle, the temperature and humidity prediction model is corrected by feedback using the newly acquired fixed monitoring data and mobile monitoring data, and rolling time-domain optimization is performed again.

7. The intelligent control method for venue air conditioning based on a distributed sensor network according to claim 6, characterized in that, The calculation method for the comprehensive control deviation specifically includes: The PMV value of each voxel in the three-dimensional voxel grid is calculated using the predictive average voting model; For any air conditioning control domain, the comprehensive control deviation in the prediction time domain is calculated. The comprehensive control deviation is the average value of the PMV value of all three-dimensional voxel grids in the air conditioning control domain, the maximum value of the absolute value of the PMV value, and the weighted sum of the personnel density factor.

8. A venue air conditioning intelligent control system based on a distributed sensor network, operating the venue air conditioning intelligent control method based on a distributed sensor network as described in any one of claims 1-7, characterized in that, include: The multi-source data aggregation module is used to acquire fixed monitoring data and corresponding fixed three-dimensional coordinates uploaded by multiple fixed acquisition terminals distributed within the venue; The environmental field reconstruction module is used to construct a three-dimensional obstacle model containing the physical structure of the venue; based on the fixed monitoring data and the fixed three-dimensional coordinates, and using the shortest path length around the three-dimensional obstacle model as the distance metric for spatial correlation calculation, a three-dimensional dynamic environmental field is generated. The motion-enhanced sensing module is used to calculate the optimal sampling position of the mobile acquisition terminal based on the current three-dimensional dynamic environment field. The mobile acquisition terminal is dispatched to the optimal sampling location to collect data, and the three-dimensional dynamic environment field is updated based on the collected mobile monitoring data and the corresponding mobile three-dimensional coordinates. The collaborative control decision module is used to perform rolling time-domain optimization based on the updated three-dimensional dynamic environment field and the model predictive control strategy, generate power allocation instructions for air conditioning equipment and issue them for execution.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-7.

Images