Air conditioning load prediction control method and system for library distribution

Abstract

The application provides a library distribution-oriented air conditioning load prediction control method and system, which firstly acquires the three-dimensional coordinates of each functional area space of the library and the thermal physical property parameters of the envelope structure, and constructs a thermal inertia quantization index by partition; collects the reader flow statistical data during the opening period, generates a personnel residence time integral sequence, and multiplies the thermal inertia quantization index to obtain a load influence base; acquires the outdoor meteorological time series data in the prediction period, stacks the load influence base, the hourly outdoor comprehensive temperature sequence and the hourly solar radiation intensity sequence to generate a load prediction base tensor in three dimensions; performs time dimension sliding window energy accumulation on the tensor to obtain a predicted time load cumulative amount, and then carries out spatial filtering interpolation according to the regional space adjacency relationship by using a dynamic power exponential inverse distance weight method to obtain a library global air conditioning load prediction distribution field. The application integrates the building thermal inertia and the dynamic distribution of personnel, realizes multi-factor space-time accumulation and continuous load distribution prediction.

Description

Technical Field

[0001] This invention relates to the field of air conditioning load forecasting, and more specifically to an air conditioning load forecasting and control method and system for libraries. Background Technology

[0002] As a typical public building with high personnel mobility and diverse functional areas, libraries are internally divided into various functional areas such as reading areas, study areas, book stacks, and public corridors. The building envelope, window conditions, and spatial volume of these different areas vary significantly. The air conditioning load is simultaneously influenced by multiple factors, including the thermal characteristics of the building envelope, outdoor meteorological conditions, and the dynamic behavior of readers. With the increasing demands for energy conservation and indoor thermal comfort in public buildings, the need for refined zoning prediction and control of library air conditioning is becoming increasingly urgent.

[0003] Existing methods for predicting air conditioning loads mostly employ lumped parameter models or statistical regression models based on historical data, and are generally used to estimate loads for the entire building or large areas after consolidation. This approach ignores the spatial distribution of thermal inertia caused by differences in the thermal properties of the building envelope across different functional areas within the library, failing to accurately characterize the attenuation and delayed thermal response effects of outdoor temperature fluctuations in different areas. Furthermore, readers frequently move between reading areas, study areas, and stacks, and the number of people in each area changes dynamically over time. Current technologies often simplify personnel load to a fixed personnel density or the product of the average number of people and a static coefficient, without considering the coupling effect between personnel dwell time and building thermal inertia. This leads to a disconnect between the calculated regional load and the actual thermal environment response of the building, resulting in significant load prediction errors and failing to meet the accuracy requirements of zoning calculations.

[0004] In addition, most existing forecasting methods can only output load values ​​for a few discrete spatial points. They lack a technical approach to accumulate multi-dimensional influencing factors such as outdoor temperature, solar radiation, and internal occupancy load over time and then perform spatial continuity processing. This makes it difficult to obtain a smooth, spatially continuous load distribution field across the entire area. Relying solely on discrete zoned loads for air conditioning control makes it difficult to reflect the spatial transitions in load within the building, severely restricting the implementation of precise, zoned control of the air conditioning system. Summary of the Invention

[0005] To address the problems of existing technologies, the present invention aims to provide a method and system for predicting and controlling air conditioning loads in a library setting. By constructing a quantitative index of thermal inertia for each functional area and coupling it with the integral sequence of personnel residence time to obtain the load impact base, and then integrating outdoor meteorological time series data to construct a load prediction base tensor, a spatially continuous global air conditioning load prediction distribution field is obtained by using dynamic interpolation of time-dimensional sliding window energy accumulation and restricted adjacency relationships, thereby improving prediction accuracy and supporting refined zoning control of air conditioning.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] On the one hand, this invention provides an air conditioning load forecasting and control method for library distribution, including:

[0008] Step S1: Obtain the spatial three-dimensional coordinates of each functional area of ​​the library and the thermal properties of the building envelope, and construct the characteristic value of thermal inertia per unit volume based on the spatial three-dimensional coordinates and the thermal properties of the building envelope.

[0009] Step S2: Collect reader flow statistics during library opening hours and generate a sequence of personnel dwell time in each functional area reflecting changes in personnel stay.

[0010] Step S3: Couple the characteristic value of thermal inertia per unit volume with the integral sequence of personnel residence time to obtain the load influence base sequence of each functional area as a function changes over time;

[0011] Step S4: Obtain outdoor meteorological time series data for the predicted period of the library's location. The outdoor meteorological time series data shall include at least hourly outdoor comprehensive temperature sequence and hourly solar radiation illuminance sequence.

[0012] Step S5: Stack the load influence base, the hourly outdoor comprehensive temperature sequence, and the hourly solar radiation irradiance sequence into three-dimensional tensors according to each functional region to generate the load prediction base tensor;

[0013] Step S6: Perform a sliding window energy accumulation operation along the time dimension on the load prediction base tensor to obtain the load accumulation amount at the prediction time of each functional area;

[0014] Step S7: Spatial filtering and interpolation are performed on the cumulative load of each functional area at the predicted time according to the spatial adjacency relationship of each functional area to obtain the predicted distribution field of air conditioning load in the entire library.

[0015] Furthermore, step S1 includes:

[0016] The length, width, and height of each functional area are extracted based on the three-dimensional spatial coordinates of each area, and the volume of each functional area is calculated.

[0017] Extract the area and heat transfer coefficient of the exterior walls, roof and windows of each functional area from the thermal properties of the building envelope, and multiply them to calculate the thermal admittance of the exterior walls, roof and windows respectively.

[0018] The thermal admittance of the exterior wall, roof, and windows is summed, and the sum is divided by the volume of the corresponding functional area to obtain the characteristic value of thermal inertia per unit volume.

[0019] Furthermore, step S2 includes:

[0020] The library's opening hours are divided into multiple discrete time slots according to a preset time step, and the number of readers entering and leaving each functional area in each discrete time slot is collected.

[0021] For each functional area, starting from the first discrete time slice of the open period, the number of readers entering is added up and the number of readers leaving is subtracted to obtain the real-time number of readers in the room for each discrete time slice;

[0022] Multiply the real-time number of readers in the room for each discrete time slice by the preset time step to obtain the resident time value for the corresponding discrete time slice;

[0023] Sort all the time values ​​of people staying in the same functional area in chronological order to generate an integral sequence of people's stay time in that functional area.

[0024] Furthermore, step S3 includes:

[0025] Multiply the dwell time value in the personnel dwell time integral sequence by the preset single-person calorific value and the preset single-person moisture dissipation to obtain the median value of sensible heat load and the median value of latent heat load for each discrete time segment; the preset single-person calorific value is taken according to the sensible heat load and the latent heat load respectively.

[0026] The median value of latent heat load is converted into a latent heat load value, and the median value of sensible heat load is directly used as the sensible heat load value. The sensible heat load value of each discrete time slice is multiplied by the characteristic value of thermal inertia per unit volume, and then summed with the latent heat load value to obtain the load influence base of each discrete time slice.

[0027] The load impact base values ​​of all discrete time slices in each functional area are arranged in chronological order to generate a load impact base value sequence for each functional area.

[0028] Furthermore, step S4 includes:

[0029] Obtain hourly outdoor dry-bulb temperature, hourly outdoor relative humidity, and hourly wind speed for the future forecast period from the meteorological data interface;

[0030] Obtain hourly total solar irradiance and hourly diffuse solar irradiance on the horizontal surface during the future forecast period from the meteorological data interface;

[0031] The hourly outdoor air enthalpy is calculated based on the hourly outdoor dry-bulb temperature, hourly outdoor relative humidity, and hourly wind speed. The hourly outdoor air enthalpy is then converted into an hourly comprehensive outdoor temperature. The above conversion operation is performed for each prediction hour. The hourly comprehensive outdoor temperatures of all prediction hours are arranged in chronological order of the prediction hours to generate an hourly comprehensive outdoor temperature sequence.

[0032] The hourly solar irradiance of each orientation vertical plane is calculated based on the hourly total solar irradiance and the hourly diffuse solar irradiance of the horizontal plane. The hourly solar irradiance of each orientation vertical plane in all predicted hours is arranged by orientation and time to generate an hourly solar irradiance sequence.

[0033] Furthermore, step S5 includes:

[0034] The load impact base sequence for each functional area is used as the first dimension vector;

[0035] The hourly outdoor comprehensive temperature sequence corresponding to each functional area is used as the second dimension vector;

[0036] The hourly solar radiation irradiance sequence corresponding to each functional area is used as the third dimension vector;

[0037] Stack the first dimension vector, the second dimension vector, and the third dimension vector of the same functional region along the third dimension to form a two-dimensional feature matrix of each functional region.

[0038] Stack the two-dimensional feature matrices of all functional regions along the first dimension to generate a three-dimensional load prediction fundamental tensor.

[0039] Furthermore, step S6 includes:

[0040] A sliding window is set on the time dimension of the load prediction base tensor, and the window width of the sliding window is the ratio of the preset cumulative duration to the preset time step.

[0041] The sliding window is slid along the time dimension, one discrete time slice at a time.

[0042] At each sliding position, the load impact base of all loads within the discrete time slice covered by the sliding window is summed to obtain the cumulative load at the predicted time for each functional area;

[0043] The predicted cumulative load values ​​obtained at each sliding position for each functional area are arranged into a sequence of predicted cumulative load values ​​for each functional area.

[0044] Furthermore, step S7 includes:

[0045] Based on the three-dimensional spatial coordinates of each functional area, determine the center point coordinates of each functional area and the adjacency relationship of adjacent functional areas;

[0046] Using the center point coordinates of each functional area as interpolation nodes, the cumulative load of each functional area at the same prediction time is used as the node value.

[0047] Using the inverse distance weighting method, the interpolation range is limited by the adjacency relationship, and the load values ​​are interpolated for the spatial locations between each functional area; the continuous spatial distribution data obtained after interpolation is used as the predicted distribution field of the library's overall air conditioning load.

[0048] Furthermore, the power exponent of the inverse distance weighting method is dynamically determined based on the spatial distance between each functional area and the load difference rate. When the load difference rate is greater than a preset difference threshold, the power exponent is increased; when the load difference rate is less than the preset difference threshold, the power exponent is decreased.

[0049] According to another aspect of the present invention, an air conditioning load forecasting and control system for library distribution is also provided, applied to the above-described air conditioning load forecasting and control method for library distribution, comprising:

[0050] The parameter acquisition and thermal inertia construction module is used to acquire the spatial three-dimensional coordinates of each functional area of ​​the library and the thermal properties of the building envelope, and to construct the characteristic value of thermal inertia per unit volume based on the spatial three-dimensional coordinates and the thermal properties of the building envelope.

[0051] The personnel time sequence generation module is used to collect reader flow statistics during library opening hours and generate a personnel dwell time integral sequence reflecting changes in personnel stay in each functional area;

[0052] The load coupling calculation module is used to couple the characteristic value of thermal inertia per unit volume with the integral sequence of personnel residence time to calculate the load influence base sequence of each functional area over time.

[0053] The meteorological parameter acquisition module is used to acquire outdoor meteorological time-series data for the predicted period of the library's location. The outdoor meteorological time-series data includes at least hourly outdoor comprehensive temperature sequence and hourly solar radiation illuminance sequence.

[0054] The three-dimensional tensor construction module is used to stack the load influence base, the hourly outdoor comprehensive temperature sequence and the hourly solar radiation irradiance sequence into three-dimensional tensors according to each functional region to generate the load prediction base tensor.

[0055] The time-dimensional thermal inertia accumulation module is used to perform a sliding window energy accumulation operation along the time dimension on the load prediction base tensor to obtain the load accumulation amount at the prediction time of each functional area.

[0056] The spatial interpolation generation module is used to perform spatial filtering and interpolation on the cumulative load of each functional area at the predicted time based on the spatial adjacency relationship of each functional area, so as to obtain the predicted distribution field of air conditioning load of the entire library and output it to the outside world.

[0057] The beneficial effects of this invention are as follows:

[0058] This invention constructs a quantitative index of thermal inertia per unit volume, using three-dimensional spatial information of each functional area and thermal property parameters of the building envelope to characterize the building's thermal response characteristics in different areas, accurately representing the attenuation and delay effects of outdoor temperature fluctuations in each area. Simultaneously, based on passenger flow statistics to obtain the integral sequence of personnel dwell time, it couples personnel thermal and humidity load with the thermal inertia index to obtain the load impact base, abandoning the simplified model of personnel load equal to a fixed number of people multiplied by a static coefficient. It fully integrates the coupling relationship between personnel dwell time history and building thermal inertia, effectively reducing the prediction deviation of regional air conditioning load.

[0059] This invention organizes the load influence base, hourly outdoor comprehensive temperature sequence, and hourly solar radiation illuminance sequence into a basic tensor for load prediction, realizing a structured expression of multi-dimensional temporal influencing factors such as building envelope, personnel, and meteorology. Furthermore, it performs sliding window energy accumulation calculations on the time dimension of the tensor to simulate the thermal inertia accumulation process of the building and air conditioning system, obtaining the predicted load accumulation for each functional area, thus overcoming the shortcomings of existing technologies that lack multi-dimensional data fusion and time accumulation processing methods. Based on this, using the center point of each functional area as an interpolation node and combining regional adjacency relationships, a power-law dynamically adjustable inverse distance weighting method is used for spatial interpolation, generating a spatially continuous predicted distribution field of the library's overall air conditioning load from the discrete partitioned load results.

[0060] The spatiotemporal continuous global air conditioning load prediction distribution field obtained by this invention breaks through the limitation of traditional schemes that can only output discrete spatial point loads. It can provide complete data basis for the fine control of library air conditioning system on demand and zoned, support the differentiated adjustment of terminal equipment in different functional areas, take into account the thermal comfort level of each area in the library, and thus achieve energy-saving operation of library public building air conditioning system. Attached Figure Description

[0061] Figure 1 This is a flowchart of an air conditioning load forecasting and control method for libraries.

[0062] Figure 2 This is a flowchart for calculating the characteristic value of thermal inertia per unit volume of each functional area;

[0063] Figure 3 This is a flowchart illustrating the method for generating the base sequence of load impacts on various functional areas of the library;

[0064] Figure 4 These are the probability density distribution curves of the thermal admittance of the building envelope in each functional area of ​​the library.

[0065] Figure 5 These are comparison curves of the energy accumulation results of the load prediction baseline tensor sliding window under different preset accumulation durations;

[0066] Figure 6This is a predicted distribution field of air conditioning load in various functional areas of the library;

[0067] Figure 7 This is a logical architecture diagram of an air conditioning load forecasting and control system for a library distribution system. Detailed Implementation

[0068] In the description of this embodiment, it should be noted that the terms "upper", "lower", "center", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of the invention is usually placed in when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.

[0069] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0070] like Figure 1 As shown in the figure, this application provides an air conditioning load forecasting and control method for library distribution, including:

[0071] Step S1: Obtain the spatial three-dimensional coordinates of each functional area of ​​the library and the thermal properties of the building envelope, and construct the characteristic value of thermal inertia per unit volume based on the spatial three-dimensional coordinates and the thermal properties of the building envelope.

[0072] Step S2: Collect reader flow statistics during library opening hours and generate a sequence of personnel dwell time in each functional area reflecting changes in personnel stay.

[0073] Step S3: Couple the characteristic value of thermal inertia per unit volume with the integral sequence of personnel residence time to obtain the load influence base sequence of each functional area as a function changes over time;

[0074] Step S4: Obtain outdoor meteorological time series data for the predicted period of the library's location. The outdoor meteorological time series data shall include at least hourly outdoor comprehensive temperature sequence and hourly solar radiation illuminance sequence.

[0075] Step S5: Stack the load influence base, the hourly outdoor comprehensive temperature sequence, and the hourly solar radiation irradiance sequence into three-dimensional tensors according to each functional region to generate the load prediction base tensor;

[0076] Step S6: Perform a sliding window energy accumulation operation along the time dimension on the load prediction base tensor to obtain the load accumulation amount at the prediction time of each functional area;

[0077] Step S7: Spatial filtering and interpolation are performed on the cumulative load of each functional area at the predicted time according to the spatial adjacency relationship of each functional area to obtain the predicted distribution field of air conditioning load in the entire library.

[0078] like Figure 2 As shown, in step S1, the spatial three-dimensional coordinates of each functional area are derived from the library's architectural information model or 3D point cloud scan data. These coordinates include the three-dimensional spatial positions of the boundary corners of each functional area. The length, width, and height of each functional area are extracted from these spatial three-dimensional coordinates as follows: Two boundary corners are selected along the length direction of each functional area, and the Euclidean distance between these two corners is calculated as the length of the functional area; two boundary corners are selected along the width direction of each functional area, and the Euclidean distance between these two corners is calculated as the width of the functional area; a top corner and a bottom corner are selected along the height direction of each functional area, and the vertical distance between these two corners is calculated as the height of the functional area. The volume of each functional area is obtained by multiplying its length, width, and height together.

[0079] The thermal properties of the building envelope can be obtained from building design documents or on-site thermal testing reports. These parameters cover the exterior wall area, exterior wall heat transfer coefficient, roof area, roof heat transfer coefficient, exterior window area, and exterior window heat transfer coefficient for each functional area. The above parameters for each functional area are extracted separately, and thermal admittance is calculated: exterior wall thermal admittance is obtained by multiplying the exterior wall area by the exterior wall heat transfer coefficient; roof thermal admittance is obtained by multiplying the roof area by the roof heat transfer coefficient; and exterior window thermal admittance is obtained by multiplying the exterior window area by the exterior window heat transfer coefficient. All three types of thermal admittance are scalar quantities, with units of W / K.

[0080] The thermal admittance of the exterior walls, roof, and windows is summed, and the result is divided by the volume of each functional area to obtain the characteristic value of the thermal inertia per unit volume for each functional area. This value serves as a quantitative index of thermal inertia characterizing the building's thermal response characteristics for each functional area. The characteristic value of the thermal inertia per unit volume for each functional area is expressed by the following formula:

[0081] in, This represents the characteristic value of thermal inertia per unit volume for each functional region. This indicates the exterior wall area of ​​each functional area. This indicates the heat transfer coefficient of the exterior walls in each functional area. This indicates the roof area of ​​each functional area. This indicates the roof heat transfer coefficient for each functional area. This indicates the area of ​​the exterior windows in each functional area. This indicates the heat transfer coefficient of the exterior windows in each functional area. This represents the volume of each functional area. External wall heat transfer coefficient. Roof heat transfer coefficient and the heat transfer coefficient of the external window All values ​​are known values ​​obtained through thermal performance testing of the building envelope materials; the volume of each functional area. The thermal inertia per unit volume of each functional area is calculated based on its length, width, and height. The calculated characteristic value of thermal inertia per unit volume for each functional area is used in the calculation of the load influence base for subsequent functional areas.

[0082] The probability density distributions of thermal admittance of the exterior walls, roof, and windows in various functional areas of the library are as follows: Figure 4 As shown in the figure. The horizontal axis represents the thermal admittance value, in watts per Kelvin (W / K), and the vertical axis represents the probability density of the corresponding thermal admittance value. The figure contains three curves, corresponding to the thermal admittance of the exterior wall (solid blue line), the thermal admittance of the roof (dashed orange line), and the thermal admittance of the exterior window (dotted green line).

[0083] As shown in the figure, the probability density curve of the thermal admittance of exterior windows is mainly concentrated in the lower thermal admittance range, with the peak appearing around 200 W / K. The probability density is higher in the range of 100 W / K to 300 W / K, indicating that the thermal admittance of exterior windows in most functional areas is relatively small and concentrated. The peak probability density of roof thermal admittance is located at approximately 400 W / K, and its distribution is relatively wide, covering the range of approximately 200 W / K to 600 W / K. This indicates that there are some differences in roof thermal admittance among functional areas, but it is generally concentrated in the medium value range. The peak probability density curve of exterior wall thermal admittance appears at approximately 600 W / K, and its distribution is wide, covering the range of approximately 300 W / K to 900 W / K. This indicates that the thermal admittance of exterior walls is generally large, and there are significant differences among functional areas.

[0084] like Figure 3 As shown, in step S2, the library's daily opening hours refer to the continuous time interval between the library's opening and closing times. The preset time step is a pre-defined time interval length, and its value is determined based on the sampling frequency of the library's reader entry and exit statistics equipment. The preset time step can be 5 minutes, 10 minutes, or 15 minutes. The library's daily opening hours are divided into multiple discrete time slices according to the preset time step. These discrete time slices are connected end-to-end on the time axis and do not overlap. The duration of each discrete time slice is equal to the preset time step.

[0085] Within each discrete time slice, the number of readers entering and leaving each functional area is recorded using infrared passenger flow counting devices or video passenger flow statistics devices located at the entrances of each functional area. The number of readers entering represents the total number of people entering a functional area from the outside within that discrete time slice, and the number of readers leaving represents the total number of people leaving a functional area from the outside within that discrete time slice. The number of readers entering and leaving within each discrete time slice is stored as a numerical sequence according to the functional area.

[0086] For each functional area, starting from the first discrete time slice arranged chronologically within the library's opening hours for the day, the number of readers is accumulated and deducted for each discrete time slice. Initially, the real-time number of readers in each functional area is set to zero. After each discrete time slice, the number of readers entering the functional area is added to the real-time number of readers in that functional area, while the number of readers leaving the functional area is deducted from the real-time number of readers in that functional area. After the accumulation and deduction operations are completed, the real-time number of readers at the end of that discrete time slice is obtained. All discrete time slices are processed sequentially to obtain the real-time number of readers corresponding to each discrete time slice.

[0087] After obtaining the real-time number of readers in the room for each discrete time slice, the number of readers in the room for each discrete time slice is multiplied by a preset time step. The product is used as the dwell time value for that discrete time slice. The dwell time value represents the cumulative dwell time of all readers in that functional area within that discrete time slice, in seconds or minutes, consistent with the unit of the preset time step. All dwell time values ​​for each functional area, arranged in chronological order of the discrete time slices, are compiled into a sequence to generate the integral sequence of dwell time for each functional area. The integral sequence of dwell time for each functional area is a one-dimensional numerical sequence, where each element corresponds to the dwell time value for a discrete time slice.

[0088] Step S3 couples the thermal inertia index with the personnel dwelling sequence to solve for the load influence base of each discrete time slice. The dwelling time value corresponding to each discrete time slice in the integral sequence of personnel dwelling time in each functional area is extracted. This dwelling time value is multiplied by a preset single-person heat generation to obtain the median sensible heat load of that discrete time slice. Simultaneously, this dwelling time value is multiplied by a preset single-person moisture dissipation to obtain the median latent heat load of that discrete time slice.

[0089] The preset single-person calorific value is set according to the sensible calorific value and the latent calorific value. The sensible calorific value represents the heat dissipated by a single reader to the environment in a unit of time through convection and radiation. The latent calorific value represents the latent heat corresponding to the amount of moisture dissipated by a single reader to the environment in a unit of time through respiration and skin evaporation.

[0090] The sensible heat calorific value was determined based on a survey of library readers' activity types, with a range of 60 watts to 75 watts; the latent heat calorific value ranged from 35 watts to 50 watts.

[0091] The preset single-person moisture dissipation capacity represents the mass of water vapor released into the environment by a single reader per unit time. The preset single-person moisture dissipation capacity is derived from the standard human body moisture dissipation model, and the preset single-person moisture dissipation capacity ranges from 80 grams per hour to 120 grams per hour.

[0092] The median latent heat load is converted into a latent heat load value using the latent heat constant of vaporization of water vapor. The latent heat constant is taken as 2450 kJ per kilogram at room temperature. The latent heat load value is the product of the median latent heat load and the latent heat constant.

[0093] The personnel heat and humidity load is the sum of the sensible heat load and the latent heat load. The sensible heat load is the intermediate value of the sensible heat load and participates in the subsequent multiplication of the characteristic value of thermal inertia per unit volume. The latent heat load is added to the calculation result as an independent additional term.

[0094] The sensible heat load value in the personnel thermal and moisture load values ​​of each discrete time slice is multiplied by the characteristic value of the unit volume thermal inertia of each functional area. The result of the multiplication is added to the latent heat load value to obtain the load influence base for each discrete time slice. The formula for calculating the load influence base is expressed as follows:

[0095] in, This represents the load impact base of the c-th functional region in the t-th discrete time slice; This represents the sensible heat load value of the c-th functional region in the t-th discrete time slice. It is obtained by multiplying the dwell time value of the c-th functional region in the t-th discrete time slice by the sensible heat generation; This represents the characteristic value of the thermal inertia per unit volume of the c-th functional region. The result is obtained by summing the thermal admittance of the exterior walls, roof, and exterior windows, and then dividing by the volume of the c-th functional area. This represents the latent heat load value of the c-th functional region in the t-th discrete time slice. It is obtained by multiplying the dwell time value of the c-th functional region in the t-th discrete time slice by the preset single-person moisture dissipation amount, and then by the latent heat of vaporization constant.

[0096] The load impact base values ​​for all discrete time slices in each functional area are arranged in chronological order according to the discrete time slices, generating a load impact base value sequence for each functional area. The load impact base value sequence is a one-dimensional numerical sequence, and the sequence length is equal to the total number of discrete time slices into which the library's opening hours are divided on that day.

[0097] Step S4: Obtain meteorological time-series data for the forecast period. The forecast period is a time extending forward from the current moment, and the time resolution of the meteorological data is 1 hour.

[0098] The system obtains hourly outdoor dry-bulb temperature, hourly outdoor relative humidity, and hourly wind speed for the forecast period from the meteorological data interface. The meteorological data interface is a data communication interface established with the public meteorological service system or local meteorological station. Through the meteorological data interface, it obtains continuous hourly meteorological element data in a standard data format. The unit for hourly outdoor dry-bulb temperature is degrees Celsius, the unit for hourly outdoor relative humidity is percentage, and the unit for hourly wind speed is meters per second.

[0099] Meanwhile, the hourly total solar irradiance and hourly diffuse solar irradiance on the horizontal surface are obtained from the meteorological data interface for the future forecast period. The units of the hourly total solar irradiance and hourly diffuse solar irradiance on the horizontal surface are both watts per square meter.

[0100] After obtaining the above meteorological data, the hourly outdoor enthalpy is calculated based on the hourly outdoor dry-bulb temperature, hourly outdoor relative humidity, and hourly wind speed. The hourly outdoor enthalpy is calculated using the following formula:

[0101] in, This represents the hourly outdoor air enthalpy value at the p-th prediction hour, expressed in kilojoules per kilogram of dry air. This represents the hourly outdoor dry-bulb temperature in the p-th forecast hour, in degrees Celsius. This represents the hourly moisture content at the p-th prediction hour, expressed in kilograms of water vapor per kilogram of dry air. From hourly outdoor dry bulb temperature Hourly outdoor relative humidity The hourly atmospheric pressure was calculated using the standard wet air thermodynamic relationship, with the hourly atmospheric pressure taken as the standard atmospheric pressure value corresponding to the altitude of the library's location; constant 1.006 is the isobaric specific heat capacity of dry air, in kilojoules per kilogram of Kelvin; constant 2501 is the latent heat of vaporization of water at zero degrees Celsius, a universal constant for calculating wet air enthalpy, in kilojoules per kilogram; constant 1.86 is the isobaric specific heat capacity of water vapor, in kilojoules per kilogram of Kelvin.

[0102] The hourly outdoor air enthalpy value for each predicted hour was calculated. Next, the hourly outdoor air enthalpy is converted into hourly outdoor composite temperature. The conversion method is as follows: divide the hourly outdoor air enthalpy by the specific heat capacity conversion factor of moist air, with the conversion factor set at a fixed value of 1.02 kJ / kg Kelvin. The resulting quotient is the hourly outdoor composite temperature, and the unit of the hourly outdoor composite temperature is degrees Celsius. The above conversion operation is performed for each forecast hour, and the hourly outdoor composite temperatures of all forecast hours are arranged in chronological order of the forecast hours to generate an hourly outdoor composite temperature sequence.

[0103] The hourly solar irradiance of each oriented vertical plane is calculated based on the hourly total solar irradiance and hourly diffuse solar irradiance on the horizontal plane. Each oriented vertical plane includes the east-facing, south-facing, west-facing, and north-facing vertical planes. The hourly solar irradiance of each oriented vertical plane is calculated using the Perez sky model and the solar incidence angle for each direction. The Perez sky model divides sky diffuse radiation into three components: circumpolar diffuse radiation, celestial diffuse radiation, and isotropic diffuse radiation. Combined with the geometric parameters of each oriented vertical plane, the hourly solar irradiance received by each oriented vertical plane is calculated. During the calculation, the hourly total solar irradiance, hourly diffuse solar irradiance, solar geometric position parameters, and the azimuth and tilt angles of each oriented vertical plane are required input, with the tilt angle set to 90 degrees. The hourly solar irradiance of the east-facing, south-facing, west-facing, and north-facing vertical planes is calculated sequentially for each forecast hour within the future forecast period. The hourly solar irradiance of each facing vertical plane is then arranged by orientation and time for all forecast hours to generate an hourly solar irradiance sequence.

[0104] After obtaining the load impact base sequence, hourly comprehensive outdoor temperature sequence, and hourly solar radiation illuminance sequence, since the original meteorological data has an hourly resolution, it is necessary to perform linear interpolation on the hourly comprehensive outdoor temperature sequence and hourly solar radiation illuminance sequence to align them to the discrete time step of the personnel load.

[0105] The load impact baseline sequence for each functional area is used as the first dimension vector, with its length equal to the total number of discrete time slices in the load impact baseline sequence. The hourly outdoor comprehensive temperature sequence corresponding to each functional area is used as the second dimension vector. To align the second dimension vector with the first dimension vector on the same time scale, the hourly outdoor comprehensive temperature sequence is linearly interpolated according to a preset time step to obtain an hourly outdoor comprehensive temperature sequence with the same time resolution as the load impact baseline sequence. The length of the second dimension vector is equal to the length of the first dimension vector. The hourly solar irradiance sequence corresponding to each functional area is used as the third dimension vector. Originally, the hourly solar irradiance sequence is multi-dimensional data arranged by orientation and time. For each functional area, the hourly solar irradiance sequence corresponding to the orientation of the external window of the functional area is extracted, and a third dimension vector with the same length as the first dimension vector is obtained through linear interpolation.

[0106] Stacking the first-dimensional vector, second-dimensional vector, and third-dimensional vector of the same functional region along the third-dimensional direction involves combining three vectors of equal length according to their element positions to form a shape... A two-dimensional feature matrix is ​​obtained, where L represents the vector length. The same stacking operation is performed on all functional regions to obtain the two-dimensional feature matrix of each functional region. The two-dimensional feature matrices of all functional regions are then stacked along the first dimension to generate a matrix of shape [formula missing]. The three-dimensional load prediction base tensor is defined as follows: C represents the total number of functional regions, L represents the number of discrete time slices, and 3 represents the number of feature dimensions. The three feature dimensions correspond to the load influence base, hourly outdoor comprehensive temperature, and hourly solar radiation illuminance, respectively.

[0107] Step S6 performs time-dimensional sliding window energy accumulation calculation. The sliding window width W is obtained by dividing the preset accumulation time by the preset time step; the preset accumulation time matches the thermal response characteristics of the air conditioning terminal: 30 minutes for variable air volume terminals, 45 minutes for fan coil terminals, and 60 minutes for radiant air conditioning terminals. Taking a time step of 5 minutes as an example, the window width W is equal to 6, 9, and 12 for the above three scenarios, respectively.

[0108] In practice, a sliding window energy accumulation operation along the time dimension is performed on the load forecasting base tensor to obtain the cumulative load at the predicted time for each functional area. The time dimension of the load forecasting base tensor corresponds to the discrete time slice sequence direction of the load influence base sequence, and the indices of the time dimension are arranged in ascending order of the discrete time slices. A sliding window is set on the time dimension of the load forecasting base tensor. The window width of the sliding window is preset to the ratio of the preset accumulation duration to the preset time step, and this ratio is represented by the symbol W. The preset accumulation duration is a time length pre-set based on the thermal response time characteristics of the terminal actuators of the library air conditioning system. The preset accumulation duration ranges from 30 minutes to 60 minutes. When the library air conditioning system uses variable air volume (VAV) terminals, the preset accumulation duration is 30 minutes; when the library air conditioning system uses fan coil terminals, the preset accumulation duration is 45 minutes; and when the library air conditioning system uses radiant terminals, the preset accumulation duration is 60 minutes. The preset time step is set to 5 minutes as described in the above embodiment. Under this setting, when the preset cumulative time is 30 minutes, the window width W of the sliding window is equal to 6; when the preset cumulative time is 45 minutes, the value of W is equal to 9; and when the preset cumulative time is 60 minutes, the value of W is equal to 12.

[0109] The sliding window is moved along the time dimension of the load prediction base tensor, one discrete time slice at a time. The starting position of the sliding window aligns its left boundary with the first discrete time slice of the time dimension, covering the range from the first to the Wth discrete time slice. After completing the calculation at this position, the left boundary of the sliding window moves one discrete time slice to the right, covering the range from the second to the (W+1)th discrete time slice, and so on, until the right boundary of the sliding window reaches the last discrete time slice of the time dimension, at which point the sliding process stops.

[0110] At each sliding position, the load impact bases of all loads within the discrete time slice covered by the sliding window are summed to obtain the cumulative load at the predicted time for each functional area. (The symbol is used to represent this.) This represents the cumulative load of the u-th functional area at the predicted time of the v-th sliding position. The calculation formula is: in, For functional area indexing, The value range is 1 to , represents the total number of functional areas in the library; v is the sliding position index, with a value ranging from 1 to... L represents the total number of discrete time slices in the load influence base sequence; This represents the load impact base of the u-th functional region in the s-th discrete time slice. The method described in the above embodiments is used to obtain the value; W represents the window width of the sliding window, which is the ratio of the preset cumulative duration to the preset time step.

[0111] Calculate the cumulative load of each functional area at each sliding position according to the above formula. The functional areas are at the first sliding position to the second. The cumulative load values ​​obtained from all predicted time points at each sliding position are arranged in ascending order of sliding position index to form a numerical sequence. This numerical sequence is the cumulative load value sequence for the u-th functional region. The same arrangement operation is performed on all functional regions to obtain the cumulative load value sequence for all functional regions.

[0112] like Figure 5 The figure shows the trend of the cumulative load at the predicted time as a function of the sliding position index, calculated by the energy accumulation of the sliding window under different preset accumulation durations. The horizontal axis represents the sliding position index, indicating the position of the sliding window in the time dimension of the load prediction tensor, and the vertical axis represents the cumulative load at the predicted time, in watts (W). The figure contains three curves, corresponding to the changes in the cumulative load when the preset accumulation duration is 30 minutes (window width W=6), 45 minutes (W=9), and 60 minutes (W=12), respectively.

[0113] The curves exhibit a periodic fluctuation characteristic, reflecting the time-varying nature of the library's air conditioning load and its cumulative effect over time. Specifically, the curve with a preset cumulative duration of 30 minutes represents the lowest load level, with the cumulative load generally maintained between 1100W and 1350W, exhibiting small fluctuations and gentle changes in peaks and troughs. The curve with a preset cumulative duration of 45 minutes shows a significantly higher load level, with the cumulative load mainly distributed between 1550W and 2050W, exhibiting a fluctuation trend synchronized with the 30-minute curve, but with increased amplitude. The curve with a preset cumulative duration of 60 minutes shows the highest cumulative load, ranging from approximately 2100W to 2700W, with more pronounced peaks and troughs, and the overall load value is about 1000W higher than the other two curves.

[0114] Step S7 generates a global air conditioning load prediction distribution field by spatial interpolating based on the discrete load of each zone. In specific implementation, the cumulative load at the predicted time of each functional area is spatially filtered and interpolated according to the spatial adjacency relationship of each functional area to obtain the global air conditioning load prediction distribution field of the library. Based on the spatial three-dimensional coordinates of each functional area, the center point coordinates of each functional area and the adjacency relationship of adjacent functional areas are determined. The spatial three-dimensional coordinates of each functional area are obtained from the library's building information model, and the spatial three-dimensional coordinates of each functional area include the position information of the boundary corner points of each functional area in three-dimensional space. The three-dimensional coordinates of all boundary corner points of each functional area are extracted, and the average value of the coordinates of all boundary corner points in the length direction is calculated as the center point length coordinate, the average value of the coordinates of all boundary corner points in the width direction is calculated as the center point width coordinate, and the average value of the coordinates of all boundary corner points in the height direction is calculated as the center point height coordinate. The center point coordinates of each functional area are composed of the center point length coordinate, center point width coordinate, and center point height coordinate.

[0115] The adjacency relationship between adjacent functional areas is determined based on the spatial distance between the center point coordinates of each functional area and whether the functional areas share walls or floors. All pairs of functional areas are traversed, and the Euclidean distance between the center point coordinates of each pair is calculated. When the Euclidean distance between the center point coordinates of two functional areas is less than the adjacency distance threshold, and the two functional areas are marked as having a shared enclosure structure in the Building Information Model (BIM), these two functional areas are determined to be adjacent, and this adjacency relationship is recorded in the adjacency table. The adjacency distance threshold is set based on the average depth of the library building's floor plan, and the threshold value ranges from 15 meters to 25 meters.

[0116] At the same forecast time, the cumulative load of each functional area at that forecast time is used as the interpolation node value. (The symbol is omitted as it is not part of the original text.) The cumulative load at the predicted time for the m-th functional area is represented by the symbol... Indicates the first The coordinates of the center points of each functional area are given, where m is the index of the functional area, ranging from 1 to M, and M is the total number of functional areas in the library. The coordinates of the center points of all functional areas and their corresponding cumulative load at the predicted time constitute the interpolation node set.

[0117] The inverse distance weighting method is used to interpolate the load values ​​of the entire library space. The inverse distance weighting method uses the inverse power of the spatial distance between the location to be interpolated and the center point of each interpolation node as the weight, and calculates a weighted average of the cumulative load at the predicted time for each interpolation node. The calculation formula for the inverse distance weighting method is: in, Indicates the spatial location coordinates as The interpolated load value at the point; This represents the cumulative load of the m-th functional area at the current forecast time. Extracted from the cumulative load sequence at predicted times for each functional area; This indicates the spatial location of the m-th functional area. The weighting coefficients, The calculation method is as follows: ,in Indicates spatial location The spatial Euclidean distance to the center point coordinates of the m-th functional area. is the power exponent of the inverse distance weighting method.

[0118] Power exponent of inverse distance weighting The load difference rate between adjacent functional areas is dynamically determined based on the spatial distance and load variation rate between each functional area. The calculation method for the load variation rate between adjacent functional areas is as follows: Extract a pair of adjacent functional areas from the adjacency table, obtain the cumulative load at the same forecast time for this pair of adjacent functional areas, calculate the absolute value of the difference between the cumulative load at the forecast time, divide the absolute value of the difference between the cumulative load at the forecast time by the larger value of the cumulative load at the forecast time for this pair of adjacent functional areas, and the quotient is the load variation rate for this pair of adjacent functional areas. This process is repeated for all pairs of adjacent functional areas in the adjacency table, calculating the load variation rate for each pair of adjacent functional areas, and the average load variation rate of all pairs of adjacent functional areas is taken as the average load variation rate.

[0119] The preset difference threshold is a pre-defined boundary for judging load differences. This threshold is set based on the typical difference in load characteristics between different functional areas of the library, and its value is 0.3. When the average load difference rate exceeds the preset difference threshold, it indicates a large gradient in load change between adjacent functional areas, and the power exponent should be increased. power exponent The value is 3 when the average load difference rate is greater than the preset difference threshold. When the average load difference rate is less than or equal to the preset difference threshold, it indicates that the load change gradient between adjacent functional areas is small, and the power exponent is reduced. power exponent The value is 1.5 when it is less than or equal to the preset difference threshold.

[0120] The interpolation range is limited by adjacency relationships. Interpolation operations are performed only within the spatial location of the convex hull formed by the coordinates of the center points of each functional area. Furthermore, for functional areas that are not adjacent to each other, their corresponding weight coefficients are directly set to zero and do not participate in the weighted summation of the inverse distance weighting method. Uniformly distributed spatial grid points are generated within the library space according to a preset spatial resolution of 0.5 meters. The interpolated load value is calculated for each spatial grid point using the inverse distance weighting method described above. The interpolated load values ​​of all spatial grid points at the same prediction time are organized according to spatial location to form continuous spatial distribution data for that prediction time. The above steps are repeated for each prediction time, and the continuous spatial distribution data for all prediction times are combined in chronological order to obtain the predicted distribution field of the library's overall air conditioning load.

[0121] like Figure 6 As shown, in this embodiment, the predicted air conditioning load distribution field of the entire library is obtained by interpolating the cumulative load of each functional area at the predicted time with spatial adjacency using the inverse distance weighting method. The horizontal axis represents the length coordinate within the library space, ranging from 0 to 50 meters, and the vertical axis represents the width coordinate, ranging from 0 to 30 meters. The color scale on the right side of the legend represents the load value of the predicted air conditioning load distribution field, in watts, with values ​​gradually changing from light yellow (low load) to dark red (high load), and the maximum predicted load value is approximately 1200 watts.

[0122] Figure 6 Several distinct load peak areas are visible, located approximately at coordinates (15,22), (24,15), (37,24), (8,8), and (40,10). Each peak area exhibits a relatively concentrated dark red to orange-red hot zone, indicating that the corresponding functional areas have a high cumulative air conditioning load at the time of prediction. The spatial distribution of the load peaks corresponds to the three-dimensional coordinates of the functional area center point, and the load values ​​around them show a clear decreasing trend. The load value decreases with increasing distance from the center point, which is consistent with the calculation characteristics of the inverse distance weighted interpolation method. The interpolation range is limited to within the convex hull formed by the functional area center points. The load value in the figure decreases significantly in the edge areas, reflecting the spatial constraints of the interpolation.

[0123] The load distribution field exhibits a complex spatial characteristic with multiple peaks and valleys, reflecting the significant spatial heterogeneity of the cumulative air conditioning load due to differences in the thermal and humidity load of occupants, the characteristic value of thermal inertia per unit volume, and the influence of external environmental loads in different functional areas. The color scale ranges from approximately 0 to 1200 watts, indicating a large dynamic range in the predicted load distribution, thus meeting the needs for detailed spatial prediction of the air conditioning load across the entire library area.

[0124] This application also provides an air conditioning load forecasting and control system for library distribution, such as... Figure 7As shown, it includes the following modules:

[0125] The parameter acquisition and thermal inertia construction module is used to acquire the spatial three-dimensional coordinates of each functional area of ​​the library and the thermal properties of the building envelope, and to construct the characteristic value of thermal inertia per unit volume based on the spatial three-dimensional coordinates and the thermal properties of the building envelope.

[0126] The personnel time sequence generation module is used to collect reader flow statistics during library opening hours and generate a personnel dwell time integral sequence reflecting changes in personnel stay in each functional area;

[0127] The load coupling calculation module is used to couple the characteristic value of thermal inertia per unit volume with the integral sequence of personnel residence time to calculate the load influence base sequence of each functional area over time.

[0128] The meteorological parameter acquisition module is used to acquire outdoor meteorological time-series data for the predicted period of the library's location. The outdoor meteorological time-series data includes at least hourly outdoor comprehensive temperature sequence and hourly solar radiation illuminance sequence.

[0129] The three-dimensional tensor construction module is used to stack the load influence base, the hourly outdoor comprehensive temperature sequence and the hourly solar radiation irradiance sequence into three-dimensional tensors according to each functional region to generate the load prediction base tensor.

[0130] The time-dimensional thermal inertia accumulation module is used to perform a sliding window energy accumulation operation along the time dimension on the load prediction base tensor to obtain the load accumulation amount at the prediction time of each functional area.

[0131] The spatial interpolation generation module is used to perform spatial filtering and interpolation on the cumulative load of each functional area at the predicted time based on the spatial adjacency relationship of each functional area, so as to obtain the predicted distribution field of air conditioning load of the entire library and output it to the outside world.

[0132] Although embodiments of the present invention have been described in the specification, these embodiments are merely illustrative and should not be construed as limiting the scope of protection of the present invention. Various omissions, substitutions, and modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predictive control of air conditioning loads in a library-oriented system, characterized in that: include: Step S1: Obtain the spatial three-dimensional coordinates of each functional area of ​​the library and the thermal properties of the building envelope, and construct the characteristic value of thermal inertia per unit volume based on the spatial three-dimensional coordinates and the thermal properties of the building envelope. Step S2: Collect reader flow statistics during library opening hours and generate a sequence of personnel dwell time in each functional area reflecting changes in personnel stay. Step S3: Couple the characteristic value of thermal inertia per unit volume with the integral sequence of personnel residence time to obtain the load influence base sequence of each functional area as a function changes over time; Step S4: Obtain outdoor meteorological time series data for the predicted period of the library's location. The outdoor meteorological time series data shall include at least hourly outdoor comprehensive temperature sequence and hourly solar radiation illuminance sequence. Step S5: Stack the load influence base, the hourly outdoor comprehensive temperature sequence, and the hourly solar radiation irradiance sequence into three-dimensional tensors according to each functional region to generate the load prediction base tensor; Step S6: Perform a sliding window energy accumulation operation along the time dimension on the load prediction base tensor to obtain the load accumulation at the prediction time of each functional area; Step S7: Spatial filtering and interpolation are performed on the cumulative load of each functional area at the predicted time according to the spatial adjacency relationship of each functional area to obtain the predicted distribution field of air conditioning load in the entire library.

2. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S1 includes: The length, width, and height of each functional area are extracted based on the three-dimensional spatial coordinates of each area, and the volume of each functional area is calculated. Extract the area and heat transfer coefficient of the exterior walls, roof and windows of each functional area from the thermal properties of the building envelope, and multiply them to calculate the thermal admittance of the exterior walls, roof and windows respectively. The thermal admittance of the exterior wall, roof, and windows is summed, and the sum is divided by the volume of the corresponding functional area to obtain the characteristic value of thermal inertia per unit volume.

3. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S2 includes: The library's opening hours are divided into multiple discrete time slots according to a preset time step, and the number of readers entering and leaving each functional area in each discrete time slot is collected. For each functional area, starting from the first discrete time slice of the open period, the number of readers entering is added up and the number of readers leaving is subtracted to obtain the real-time number of readers in the room for each discrete time slice; Multiply the real-time number of readers in the room for each discrete time slice by the preset time step to obtain the resident time value for the corresponding discrete time slice; Sort all the time values ​​of people staying in the same functional area in chronological order to generate an integral sequence of people's stay time in that functional area.

4. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S3 includes: Multiply the dwell time value in the personnel dwell time integral sequence by the preset single-person calorific value and the preset single-person moisture dissipation to obtain the median value of sensible heat load and the median value of latent heat load for each discrete time segment; the preset single-person calorific value is taken according to the sensible heat load and the latent heat load respectively. The median value of latent heat load is converted into a latent heat load value, and the median value of sensible heat load is directly used as the sensible heat load value. The sensible heat load value of each discrete time slice is multiplied by the characteristic value of thermal inertia per unit volume, and then summed with the latent heat load value to obtain the load influence base of each discrete time slice. The load impact base values ​​of all discrete time slices in each functional area are arranged in chronological order to generate a load impact base value sequence for each functional area.

5. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S4 includes: Obtain hourly outdoor dry-bulb temperature, hourly outdoor relative humidity, and hourly wind speed for the future forecast period from the meteorological data interface; Obtain hourly total solar irradiance and hourly diffuse solar irradiance on the horizontal surface during the future forecast period from the meteorological data interface; The hourly outdoor air enthalpy is calculated based on the hourly outdoor dry-bulb temperature, hourly outdoor relative humidity, and hourly wind speed. The hourly outdoor air enthalpy is then converted into an hourly comprehensive outdoor temperature. The above conversion operation is performed for each prediction hour. The hourly comprehensive outdoor temperatures of all prediction hours are arranged in chronological order of the prediction hours to generate an hourly comprehensive outdoor temperature sequence. The hourly solar irradiance of each orientation vertical plane is calculated based on the hourly total solar irradiance and the hourly diffuse solar irradiance of the horizontal plane. The hourly solar irradiance of each orientation vertical plane in all predicted hours is arranged by orientation and time to generate an hourly solar irradiance sequence.

6. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S5 includes: The load impact base sequence for each functional area is used as the first dimension vector; The hourly outdoor comprehensive temperature sequence corresponding to each functional area is used as the second dimension vector; The hourly solar radiation irradiance sequence corresponding to each functional area is used as the third dimension vector; Stack the first dimension vector, the second dimension vector, and the third dimension vector of the same functional region along the third dimension to form a two-dimensional feature matrix of each functional region. Stack the two-dimensional feature matrices of all functional regions along the first dimension to generate a three-dimensional load prediction base tensor.

7. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S6 includes: A sliding window is set on the time dimension of the load prediction base tensor, and the window width of the sliding window is the ratio of the preset cumulative duration to the preset time step. The sliding window is slid along the time dimension, one discrete time slice at a time. At each sliding position, the load impact base of all loads within the discrete time slice covered by the sliding window is summed to obtain the cumulative load at the predicted time for each functional area; The predicted cumulative load values ​​obtained at each sliding position for each functional area are arranged into a sequence of predicted cumulative load values ​​for each functional area.

8. The air conditioning load forecasting and control method for library distribution as described in claim 1, characterized in that, Step S7 includes: Based on the three-dimensional spatial coordinates of each functional area, determine the center point coordinates of each functional area and the adjacency relationship of adjacent functional areas; Using the center point coordinates of each functional area as interpolation nodes, the cumulative load of each functional area at the same prediction time is used as the node value. Using the inverse distance weighting method, the interpolation range is limited by the adjacency relationship, and the load values ​​are interpolated for the spatial locations between each functional area; the continuous spatial distribution data obtained after interpolation is used as the predicted distribution field of the library's overall air conditioning load.

9. The air conditioning load forecasting and control method for library distribution as described in claim 8, characterized in that, The power exponent of the inverse distance weighting method is dynamically determined based on the spatial distance between each functional area and the load difference rate. When the load difference rate is greater than a preset difference threshold, the power exponent is increased; when the load difference rate is less than the preset difference threshold, the power exponent is decreased.

10. A library-oriented air conditioning load forecasting and control system, applied to the library-oriented air conditioning load forecasting and control method according to any one of claims 1-9, characterized in that, include: The parameter acquisition and thermal inertia construction module is used to acquire the spatial three-dimensional coordinates of each functional area of ​​the library and the thermal properties of the building envelope, and to construct the characteristic value of thermal inertia per unit volume based on the spatial three-dimensional coordinates and the thermal properties of the building envelope. The personnel time sequence generation module is used to collect reader flow statistics during library opening hours and generate a personnel dwell time integral sequence reflecting changes in personnel stay in each functional area; The load coupling calculation module is used to couple the characteristic value of thermal inertia per unit volume with the integral sequence of personnel residence time to calculate the load influence base sequence of each functional area over time. The meteorological parameter acquisition module is used to acquire outdoor meteorological time-series data for the predicted period of the library's location. The outdoor meteorological time-series data includes at least hourly outdoor comprehensive temperature sequence and hourly solar radiation illuminance sequence. The three-dimensional tensor construction module is used to stack the load influence base, the hourly outdoor comprehensive temperature sequence and the hourly solar radiation irradiance sequence into three-dimensional tensors according to each functional region to generate the load prediction base tensor. The time-dimensional thermal inertia accumulation module is used to perform a sliding window energy accumulation operation along the time dimension on the load prediction base tensor to obtain the load accumulation amount at the prediction time of each functional area. The spatial interpolation generation module is used to perform spatial filtering and interpolation on the cumulative load of each functional area at the predicted time based on the spatial adjacency relationship of each functional area, so as to obtain the predicted distribution field of air conditioning load of the entire library and output it to the outside world.

Images