Optimal control method and system for transmissive roof spray shading and evaporative cooling with multiple targets

Abstract

This invention relates to the technical field of controlling spraying to cool buildings using model learning methods. It solves the problem of high energy and water consumption in existing technologies for temperature control of translucent buildings. It provides a multi-objective optimization control method and system for spraying shading and evaporative cooling of translucent roofs. First, the predicted average vote and glare probability of the translucent building are calculated to obtain the first real-time energy consumption of the air conditioning system and the real-time water consumption and second real-time energy consumption of the spraying system. Then, a multi-objective optimization model is established based on the predicted average vote, glare probability, first real-time energy consumption, real-time water consumption, and second real-time energy consumption. The predicted average vote and glare probability are used as the first optimization objective, and the minimum value of the first real-time energy consumption, real-time water consumption, and second real-time energy consumption is used as the second optimization objective. The multi-objective optimization model is solved in a rolling manner to calculate the predicted values ​​of spray pressure and spray flow rate, which ensures indoor light and heat comfort while reducing water and energy consumption.

Description

Technical Field

[0001] This invention relates to the technical field of controlling spraying to cool buildings using model learning methods, and particularly to a multi-objective optimization control method and system for evaporative cooling of translucent roofs using spraying for shading. Background Technology

[0002] Modern buildings widely use large-area translucent roofs to enhance natural lighting and spatial aesthetics. However, the solar heat gain caused by translucent roofs can account for 30% to 50% of the air conditioning load, significantly increasing the energy consumption for building temperature control. Therefore, in addition to using air conditioning for temperature control, a misting cooling system is also used to reduce solar radiation.

[0003] Traditional spray cooling systems often operate on a timed or constant-pressure basis, leading to problems of over-spraying or under-spraying. This results in high water and electricity consumption and difficulty in ensuring indoor thermal comfort. For example, Chinese patent CN120400864A discloses a multi-source synergistic spray cooling concentrated photovoltaic hydrogen production system and its control method. It achieves precise separation of the solar spectrum through jacketed tube frequency division and reaction tubes. Visible light is converted into electrical energy via photovoltaic modules, while infrared light undergoes photothermal conversion via catalytic reaction tubes. Excess heat energy is directly used in the activation process of water electrolysis for hydrogen production, achieving cascaded utilization of solar energy. However, this patent aims to stabilize the solar cell operating temperature within its optimal efficiency range while converting the heat loss of traditional systems into effective hydrogen production energy. Its application scenario is a photovoltaic hydrogen production system, which differs significantly from building indoor environmental control. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a multi-objective optimization control method and system for mist shading and evaporative cooling of translucent roofs, in order to solve the problems of high energy and water consumption in the existing technology when controlling the temperature of translucent buildings.

[0005] In a first aspect, embodiments of the present invention provide a multi-objective optimization control method for evaporative cooling and misting of translucent roofs, applicable to translucent buildings equipped with air conditioning and misting systems, the method comprising: Based on the outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the translucent building, calculate the predicted average vote and glare probability of the translucent building; obtain the first real-time energy consumption of the air conditioning system, and obtain the real-time water consumption and second real-time energy consumption of the spray system; A multi-objective optimization model is established based on the predicted average vote, the glare probability, the first real-time energy consumption, the real-time water consumption, and the second real-time energy consumption. The predicted average vote and the glare probability are taken as the first optimization objective, and the minimum value of the first real-time energy consumption, the real-time water consumption and the second real-time energy consumption is taken as the second optimization objective. The multi-objective optimization model is solved by rolling using model predictive control or deep reinforcement learning algorithm to calculate the predicted value of spray pressure and the predicted value of spray flow. The spray system is controlled based on the predicted spray pressure and the predicted spray flow rate.

[0006] As an optional implementation, the establishment of a multi-objective optimization model based on the predicted average vote, the glare probability, the first real-time energy consumption, the real-time water consumption, and the second real-time energy consumption includes: A first dynamic weight is set for the first real-time energy consumption of the air conditioning system, a second dynamic weight is set for the real-time water consumption of the spray system, a third dynamic weight is set for the second real-time energy consumption of the spray system, and the predicted average vote and the glare probability are combined into a comprehensive comfort deviation index. A fourth dynamic weight is set for the comprehensive comfort deviation index to construct an initial optimization objective function; wherein, a glare penalty coefficient is preset for the glare probability. The current energy consumption cost and the preset baseline energy consumption cost are obtained in real time. The comfort priority index is calculated by combining the predicted average vote and the glare probability calculated at the current time. The first dynamic weight, the second dynamic weight, the third dynamic weight, and the fourth dynamic weight are adjusted in real time according to the comfort priority index to establish the multi-objective optimization model in order to meet the normalization condition.

[0007] As an optional implementation, the step of obtaining the current energy consumption cost and the preset baseline energy consumption cost in real time, and combining the predicted average vote calculated at the current time with the glare probability, to calculate the comfort priority index includes: The system acquires the real-time electricity price signal and / or real-time water price signal at the current moment, and calculates the real-time energy consumption cost, which represents the comprehensive operating cost of the current system, by weighting the energy type consumed by the first real-time energy consumption of the air conditioning system and the resource type consumed by the real-time water consumption and the second real-time energy consumption of the spray system. Wherein, if both the air conditioning system and the spray system consume electrical energy, the real-time energy consumption cost is taken as the time-of-use electricity price at the current moment; if multiple energy media with different billing standards are involved, the cost is obtained by weighted summation based on the consumption ratio of each energy media and its corresponding unit price. Obtain the historical average energy price of the location of the translucent building during the cooling season or typical operating cycle, or set the value according to the user's preset economic operating target, as the benchmark energy cost; Based on the predicted average vote and the glare probability, a comprehensive comfort deviation is constructed. The comprehensive comfort deviation is then correlated with the ratio of real-time energy consumption cost to baseline energy consumption cost to calculate the comfort priority index.

[0008] As an optional implementation, the step of adjusting the first dynamic weight, the second dynamic weight, the third dynamic weight, and the fourth dynamic weight in real time according to the comfort priority index includes: Compare the relationship between the currently calculated comfort priority index and the preset index threshold range; If the comfort priority index rises and exceeds the first preset threshold, it is determined that the current indoor light and heat comfort is deteriorating or the real-time energy consumption cost is low. Then, the value of the fourth dynamic weight is increased, and the values ​​of the first dynamic weight, the second dynamic weight, and the third dynamic weight are decreased accordingly. If the comfort priority index decreases and falls below the second preset threshold, it is determined that the current indoor light and heat comfort is good or the real-time energy consumption cost is high. Then, the values ​​of the first dynamic weight, the second dynamic weight, and the third dynamic weight are increased, and the value of the fourth dynamic weight is decreased accordingly. After the real-time adjustment is completed, the first dynamic weight, the second dynamic weight, the third dynamic weight, and the fourth dynamic weight are constrained to satisfy the normalization condition.

[0009] As an optional implementation, if model predictive control is used to solve the multi-objective optimization model in a rolling manner, after the prediction time domain and control time domain of the model predictive control are set, a multi-objective evolutionary algorithm is used to solve for the optimal solution set, and a knee point is selected from the optimal solution set by the fuzzy comprehensive evaluation method to calculate the predicted spray pressure value and the predicted spray flow rate value.

[0010] As an optional implementation, the step of using model predictive control to solve the multi-objective optimization model in a rolling manner specifically includes: The prediction time domain and control time domain of the model predictive control are set. At each rolling optimization time, based on the currently obtained outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, indoor thermal environment parameters, first real-time energy consumption, real-time water consumption and second real-time energy consumption, the multi-objective optimization model is used as the objective function, and the spray pressure and spray flow rate are used as decision variables to construct a constrained optimization problem with the prediction time domain as the span. The constrained optimization problem is solved by a multi-objective evolutionary algorithm to obtain a Pareto optimal solution set consisting of multiple non-dominated solutions. Each solution in the Pareto optimal solution set corresponds to a set of candidate control sequences for the spray pressure and the spray flow rate. The fuzzy comprehensive evaluation method is used to evaluate the overall satisfaction of each candidate solution in the Pareto optimal solution set, including: defining fuzzy membership functions for the first optimization objective and the second optimization objective respectively, calculating the weighted comprehensive satisfaction of each candidate solution, and selecting the solution with the highest weighted comprehensive satisfaction and the largest trade-off cost between the first optimization objective and the second optimization objective from the Pareto optimal solution set as the knee point; Extract the control increments of the spray pressure and spray flow rate corresponding to the knee point at the current control moment to generate the predicted values ​​of the spray pressure and spray flow rate.

[0011] As an optional implementation, if a deep reinforcement learning algorithm is used to solve the multi-objective optimization model in a rolling manner, the state space of the deep reinforcement learning algorithm includes the outdoor temperature of the light-transmitting building, the outdoor relative humidity, the outdoor solar radiation, the predicted average vote, the glare probability, and the real-time energy consumption cost, and the action space of the deep reinforcement learning algorithm includes the spray pressure increment and the spray flow rate increment. A first reward weight is set for the first real-time energy consumption, a second reward weight is set for the real-time water consumption, a third reward weight is set for the second real-time energy consumption, a fourth reward weight is set for the predicted average vote, and a fifth reward weight is set for the glare probability, so as to establish the reward function of the deep reinforcement learning algorithm. Among them, the first reward weight, the second reward weight, the third reward weight, the fourth reward weight, and the fifth reward weight satisfy the normalization condition.

[0012] As an optional implementation, the multi-objective optimization control method for light-transmitting roof spray shading evaporation cooling further includes: when the intensity of transmitted solar radiation is detected to be lower than a preset radiation intensity threshold, controlling the spray system based on the intermittent spray duty cycle, wherein the intermittent spray duty cycle is obtained by using the ratio of the spray on duration to the spray off duration.

[0013] As an optional implementation, the first real-time energy consumption is predicted based on a preset proxy model, the input of which includes the outdoor meteorological parameters and water mist parameters.

[0014] Secondly, embodiments of the present invention provide a multi-objective optimization control system for translucent roof spray shading evaporative cooling, applied to translucent buildings equipped with air conditioning and spray systems, the system comprising: The data acquisition device is used to acquire outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the light-transmitting building; to acquire the first real-time energy consumption of the air conditioning system; and to acquire the real-time water consumption and second real-time energy consumption of the spray system. The data processing device is used to calculate the predicted average vote and glare probability of the translucent building based on the outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the translucent building, and to establish a multi-objective optimization model based on the predicted average vote, the glare probability, the first real-time energy consumption, the real-time water consumption, and the second real-time energy consumption. The data processing device is further configured to take the predicted average vote and the glare probability as the first optimization objective, take the minimum value of the first real-time energy consumption, the real-time water consumption and the second real-time energy consumption as the second optimization objective, and use model predictive control or deep reinforcement learning algorithm to perform rolling solution on the multi-objective optimization model to calculate the predicted value of spray pressure and the predicted value of spray flow. An execution control device is used to control the spray system based on the predicted spray pressure and the predicted spray flow rate.

[0015] In summary, the beneficial effects of the present invention are as follows: The multi-objective optimization control method and system for light-transmitting roof spray shading evaporative cooling provided in this invention establishes a multi-objective optimization model, takes the predicted average vote and glare probability as the first optimization objective, and takes the minimum value of the first real-time energy consumption, real-time water consumption and the second real-time energy consumption as the second optimization objective. The multi-objective optimization model is solved by rolling model predictive control or deep reinforcement learning algorithm, which can calculate the predicted value of spray pressure and the predicted value of spray flow.

[0016] The spray system is controlled based on predicted spray pressure and flow rates, causing the variable frequency pumps and proportional valves to perform corresponding actions. This ensures indoor light and heat comfort while minimizing water and energy consumption. Furthermore, because the spray system dynamically alters the heat gain and thermal humidity environment within the translucent building, it indirectly controls the operating load and energy consumption of the air conditioning system, ultimately achieving synergistic energy savings between the spray and air conditioning systems. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, and these are all within the protection scope of the present invention.

[0018] Figure 1This is a flowchart illustrating the multi-objective optimization control method for evaporative cooling and shading of a light-transmitting roof using a spray system according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the adjustment strategy for the comfort priority index in an embodiment of the present invention; Figure 3 This is a timing diagram of the rolling optimization of model predictive control in an embodiment of the present invention; Figure 4 This is a schematic diagram of the process of training a multi-objective optimization model using a deep reinforcement learning algorithm in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of the multi-objective optimization control system for light-transmitting roof spray shading evaporative cooling in one embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of the multi-objective optimization control system for light-transmitting roof spray shading evaporative cooling in another embodiment of the present invention. Detailed Implementation

[0019] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only configured to explain the present invention and are not configured to limit the present invention. For those skilled in the art, the present invention can be practiced without some of these specific details. The following description of the embodiments is merely intended to provide a better understanding of the present invention by illustrating examples of the invention.

[0020] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0021] Because translucent buildings have translucent roofs, air conditioning systems need to work extra to handle the solar heat gain inside the building. Spraying systems, which spray the translucent building, can effectively reduce the workload of the air conditioning system, thereby reducing its energy consumption. However, existing spraying systems operate in a single mode and are not linked to the air conditioning system for control, resulting in high energy and water consumption in current temperature control solutions for translucent buildings.

[0022] Firstly, the multi-objective optimization control method for evaporative cooling and shading of translucent roofs provided in this invention can be applied to translucent buildings equipped with air conditioning and misting systems. By linking the misting and air conditioning systems for control, the energy and water consumption of the misting system are effectively reduced, and the energy consumption of the air conditioning system is indirectly reduced, achieving the beneficial effects of energy conservation and emission reduction. Please refer to [example missing]. Figure 1 As shown, the linkage control method includes the following steps: Step S101: Obtain and calculate the predicted average vote and glare probability of the translucent building based on the outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the translucent building; obtain the first real-time energy consumption of the air conditioning system, and obtain the real-time water consumption and second real-time energy consumption of the spray system.

[0023] Specifically, an outdoor weather station can be set up on the target light-transmitting building. The outdoor weather station can collect outdoor meteorological parameters in real time through meteorological data acquisition equipment. Outdoor meteorological parameters can include outdoor temperature, outdoor humidity, outdoor radiation, and outdoor wind speed.

[0024] The intensity of transmitted solar radiation is determined by the total outdoor solar radiation intensity and the solar transmittance of the translucent roof material, which in turn is determined by the properties of the translucent material itself. Indoor illuminance distribution can be determined by setting up multiple measuring points indoors according to certain rules, measuring each point with a lux meter, and then plotting isolux curves or generating an indoor illuminance distribution map through interpolation. Alternatively, illuminance sensors can be directly installed at key locations indoors, and the indoor illuminance distribution can be determined by combining this with the room's shape and layout.

[0025] Indoor thermal environment parameters can be obtained using instruments such as thermometers, hygrometers, and anemometers installed indoors. These parameters may include indoor temperature, indoor humidity, indoor wind speed, and black sphere temperature, which is used to calculate the average radiant temperature.

[0026] After obtaining outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters, the predicted mean vote (PMV) and daylight glare probability (DGP) of the translucent building are calculated based on these data.

[0027] Specifically, the predicted average vote is based on the human body thermal balance equation and a statistical model of thermal sensation voting, establishing a quantitative relationship between human thermal sensation and six physical and physiological parameters under steady-state conditions. The predicted average vote comprehensively considers temperature, humidity, wind speed, radiant temperature, human metabolic rate, and clothing thermal resistance, reflecting human thermal sensation. Temperature, radiant temperature, wind speed, and humidity are the aforementioned indoor thermal environment parameters, while human metabolic rate and clothing thermal resistance can be input as preset typical values. For example, in office activities, the human metabolic rate can be preset to 1.2 MET, and the thermal resistance of summer clothing can be preset to 0.5 CL0. Specifically, the human body thermal balance equation is:

[0028] Where M is the human metabolic rate, W is the work done by the human body, H is the sensible heat loss from respiration, E is the latent heat loss from respiration and heat dissipation through skin evaporation, R is the radiative heat transfer, C is the convective heat transfer, and S is the human body heat storage rate.

[0029] Based on the above human body thermal balance equation and thermal sensation voting regression analysis, the predicted average vote can be calculated and solved according to the algorithm specified in ISO 7730:2025 "Ergonomics of thermal environments - Calculation of thermal comfort using PMV and PPD indices and analysis, determination and interpretation of thermal comfort using local thermal comfort criteria" or ASHRAE Standard 55-2020.

[0030] Alternatively, by calling Python's pythermalcomfort library and inputting indoor temperature and humidity, wind speed, black ball temperature, people's clothing, and human metabolic level, the predicted average vote value can be output. The predicted average vote value can determine whether the people indoors are in the comfort zone. When the predicted average vote value is greater than or equal to -0.5 and less than or equal to 0.5, it is usually determined that they are in the comfort zone.

[0031] Specifically, the glare probability is an assessment of human visual discomfort based on indoor illuminance distribution and the location of sunlight sources. The higher the glare probability value, the greater the risk of glare.

[0032] Calculating glare probability requires first obtaining the vertical illuminance of the target translucent building, the luminance of each glare source, and the solid angle. Vertical illuminance (Ev) represents the illuminance on the vertical plane along the line of sight at the observer's eye position; it represents the overall background brightness level at the eye's location. The luminance (Ls) of each glare source represents the luminance of all areas within the field of view that could potentially cause glare. The solid angle (ωs) of each glare source represents the size of the solid angle subtended by that source at the observer's eye, reflecting the apparent size of the source within the field of view.

[0033] Professional lighting simulation software, such as Radiance or DAYSIM, can be used to build a model of the target translucent building. The precise vertical illuminance Ev, the brightness Ls of each glare source, and the solid angle ωs can be obtained through ray tracing calculations. Then, the glare probability for the whole year or a typical day can be calculated by combining the position index Pi of each light source. The position index Pi of each light source represents a correction factor related to the angle of the light source's deviation from the line of sight.

[0034] In practical implementation, since the positions and line of sight of people inside the target translucent building are generally preset, the main variables are solar radiation intensity and indoor illuminance distribution. The vertical illuminance Ev at key locations can be obtained through an indoor illuminance sensor network, and the luminance distribution map within the field of view can be captured by a high dynamic range imaging device to determine the luminance Ls and solid angle ωs of each glare source, thereby enabling real-time calculation of glare probability.

[0035] Glare probability can be calculated using the following formula:

[0036] Where DGP is the glare probability value, Ev is the vertical illuminance, Ls is the luminance of each glare source, and ωs is the solid angle of each glare source.

[0037] When the calculated glare probability value is less than 0.35, it means there is no significant glare and it is acceptable for people indoors; when the calculated glare probability value is greater than or equal to 0.35 and less than or equal to 0.4, it means that people indoors can perceive glare; when the calculated glare probability value is greater than or equal to 0.4, it means that people indoors can perceive strong glare and it is unacceptable.

[0038] The first real-time energy consumption of the air conditioning system can be directly obtained using the corresponding air conditioning management system, such as a Building Automation System (BAS) or a Building Management System (BMS). The real-time water consumption of the spray system can be obtained using an online water flow meter, while the second real-time energy consumption can be obtained using a real-time electricity meter.

[0039] In order to reduce the delay in obtaining the first real-time energy consumption and thus enable more timely linkage control of spraying and air conditioning, as an optional implementation method, the first real-time energy consumption can be predicted based on a preset proxy model, wherein the input of the proxy model includes outdoor meteorological parameters and water mist parameters.

[0040] Specifically, historical data over a certain period of time is collected in advance. This historical data may include outdoor meteorological parameters, air conditioning setting parameters, and measured instantaneous power or short-term energy consumption of the air conditioning system. Among these, air conditioning setting parameters may include indoor set temperature, supply air temperature, etc.

[0041] Using outdoor meteorological parameters and water mist parameters as input features, and the measured instantaneous power or short-term energy consumption of the air conditioning system as the target output, a surrogate model is established based on the complex nonlinear mapping relationship between the input features and the target output.

[0042] After obtaining all the historical data mentioned above, the data can be cleaned, normalized, or standardized, and then divided into training, validation, and test sets according to time sequence. Support Vector Regression (SVR) or Long Short-Term Memory (LSTM) networks can be used for corresponding training to obtain the surrogate model mentioned above.

[0043] The SVR-based proxy model's main tuning parameters include the penalty coefficient, kernel function parameters, and insensitive loss parameters. The LSTM-based proxy model requires designing network structures such as the number of layers and neurons, and its main tuning parameters include the learning rate, training epochs, and batch size. The trained proxy model is then evaluated using the aforementioned validation set. Evaluation metrics can include the root mean square error (RMSE) and mean absolute percentage error (MAPE). Once the relevant evaluation metrics are met, the first real-time energy consumption of the air conditioning system can be predicted based on the trained proxy model.

[0044] Step S102: Based on the predicted average vote, glare probability, first real-time energy consumption, real-time water consumption, and second real-time energy consumption, establish a multi-objective optimization model; Specifically, a first dynamic weight can be set for the first real-time energy consumption of the air conditioning system, a second dynamic weight can be set for the real-time water consumption of the spray system, a third dynamic weight can be set for the second real-time energy consumption of the spray system, and the predicted average vote and glare probability can be combined into a comprehensive comfort deviation index. A fourth dynamic weight can be set for the comprehensive comfort deviation index to construct an initial optimization objective function; where a glare penalty coefficient is preset for the glare probability. The current energy consumption cost and the preset baseline energy consumption cost are obtained in real time. Combined with the predicted average vote and glare probability calculated at the current moment, the comfort priority index is calculated. The first, second, third, and fourth dynamic weights are adjusted in real time based on the comfort priority index to establish a multi-objective optimization model that satisfies the normalization condition.

[0045] In one alternative implementation, the multi-objective optimization model can refer to the function shown below: J=w1·E ac +w2·W+w3·P+w4·(PMV²+λ·DGP) Where w1 is the first dynamic weight, w2 is the second dynamic weight, w3 is the third dynamic weight, w4 is the fourth dynamic weight, and E ac denoted as the first real-time energy consumption, W as the real-time water consumption, P as the second real-time energy consumption, PMV as the predicted average vote, DGP as the glare probability, and λ as the glare penalty coefficient, with λ ranging from 1 to 10.

[0046] Specifically, the first, second, and third dynamic weights are all energy consumption weights, while the fourth dynamic weight is a comfort weight. By dynamically adjusting the energy consumption weight and comfort weight, the corresponding optimization objectives can be changed, which determines whether the subsequent linkage control of the spray and air conditioning is biased towards indoor comfort or system energy saving.

[0047] Therefore, it is necessary to dynamically adjust the energy consumption weight and comfort weight. Specifically, the first dynamic weight, the second dynamic weight, the third dynamic weight and the fourth dynamic weight can be adjusted in real time through the preset comfort priority index. The first dynamic weight, the second dynamic weight, the third dynamic weight and the fourth dynamic weight satisfy the normalization condition, that is, w1+w2+w3+w4=1.

[0048] The comfort priority index is determined based on real-time energy consumption cost and baseline energy consumption cost. Specifically, the comfort priority index can be determined through the following steps A1~A3: A1: Obtain the real-time electricity price signal and / or real-time water price signal at the current moment, and calculate the real-time energy consumption cost that represents the current comprehensive operating cost of the system by weighting the energy type consumed by the first real-time energy consumption of the air conditioning system and the resource type consumed by the real-time water consumption and the second real-time energy consumption of the spray system. If both the air conditioning system and the spray system consume electricity, the real-time energy cost is taken as the time-of-use electricity price at the current moment; if multiple energy media with different billing standards are involved, the cost is obtained by weighted summation based on the consumption ratio of each energy media and its corresponding unit price. A2: Obtain the historical average energy price of the location of the translucent building during the cooling season or typical operating cycle, or set the value according to the user's preset economic operating target, as the benchmark energy cost; A3: Based on the predicted average vote and glare probability, a comprehensive comfort deviation is constructed. The comprehensive comfort deviation is then correlated with the ratio of real-time energy consumption cost to benchmark energy consumption cost to calculate the comfort priority index.

[0049] When comfort levels decrease or real-time energy costs decrease, the comfort priority index increases the weight of comfort, thus prioritizing indoor comfort; conversely, when comfort levels improve and real-time energy costs increase, the comfort priority index decreases the weight of comfort, thus prioritizing energy conservation. In one optional implementation, the comfort priority index can be calculated using the following formula: CPI = (PMV² + k·DGP) / ( E price / E price,ref ) In the formula, CPI is the comfort priority index. E price For real-time energy consumption costs, E price,ref The base energy cost is k, which is a constant. The real-time energy cost is the real-time cost of energy consumed by the air conditioning system and / or the misting system. For example, if the air conditioning system and / or the misting system consumes natural gas, the real-time energy cost is the real-time natural gas price; if the air conditioning system and / or the misting system consumes electricity, the real-time energy cost is the real-time electricity price.

[0050] like Figure 2 As shown, the real-time adjustment strategy for the comfort priority index includes: when the comfort priority index value rises and exceeds the first preset threshold, indicating that the current indoor thermal environment or visual comfort is deteriorating, and / or the real-time energy consumption cost is low, the fourth dynamic weight is automatically increased, while the first, second, and third dynamic weights are correspondingly decreased, so that the optimization objective is more inclined to improve comfort. Conversely, when the comfort priority index value decreases and falls below the second preset threshold, the first, second, and third dynamic weights are automatically increased, while the fourth dynamic weight is decreased, so that the optimization objective is more inclined to save energy and water.

[0051] The first and second preset thresholds can be set according to the actual application scenario. Furthermore, during the adjustment process, it is ensured that w1+w2+w3+w4=1. The adjustment of the comfort priority index on the weights of energy consumption and comfort can dynamically prioritize the comfort of indoor personnel or reduce the overall operating cost of the target translucent building under complex and ever-changing thermal environments and energy cost changes.

[0052] Step S103: Take the predicted average vote and glare probability as the first optimization objective, and take the minimum of the first real-time energy consumption, real-time water consumption and the second real-time energy consumption as the second optimization objective. Use model predictive control or deep reinforcement learning algorithm to solve the multi-objective optimization model in a rolling manner, and calculate the predicted spray pressure and spray flow rate. As an optional implementation method, if Model Predictive Control (MPC) is used to solve the multi-objective optimization model in a rolling manner, after the prediction time domain and control time domain of the model predictive control are set, the optimal solution set is solved by the multi-objective evolutionary algorithm, and the knee point is selected from the optimal solution set by the fuzzy comprehensive evaluation method to calculate the predicted values ​​of spray pressure and spray flow.

[0053] In specific implementation, such as Figure 3 As shown, the prediction time domain and control time domain of the model predictive control can be set. At each rolling optimization time, based on the currently obtained outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, indoor thermal environment parameters, first real-time energy consumption, real-time water consumption and second real-time energy consumption, the multi-objective optimization model is used as the objective function, and the spray pressure and spray flow rate are used as decision variables to construct a constrained optimization problem with the prediction time domain as the span.

[0054] Next, a multi-objective evolutionary algorithm is used to solve the constrained optimization problem, obtaining a Pareto optimal solution set consisting of multiple non-dominated solutions. Each solution in the Pareto optimal solution set corresponds to a set of candidate control sequences for spray pressure and spray flow rate. The fuzzy comprehensive evaluation method is used to evaluate the comprehensive satisfaction of each candidate solution in the Pareto optimal solution set, including: defining fuzzy membership functions for the first optimization objective and the second optimization objective respectively, calculating the weighted comprehensive satisfaction of each candidate solution, and selecting the solution with the highest weighted comprehensive satisfaction and the largest trade-off cost between the first optimization objective and the second optimization objective from the Pareto optimal solution set as the knee point.

[0055] Finally, by extracting the control increments of the spray pressure and spray flow rate corresponding to the knee point at the current control moment, the predicted values ​​of spray pressure and spray flow rate are generated.

[0056] For example, the prediction time domain of model predictive control can be set to 10–30 minutes, and its control time domain can be set to 3–5 minutes. Multi-objective evolutionary algorithms, such as NSGA-II or MOEA / D, can be used to search and generate a series of Pareto optimal solutions that achieve different trade-offs between the first and second optimization objectives. Subsequently, a fuzzy decision-making method is used to set a satisfaction function for each optimization objective, and the knee point with the highest overall satisfaction and the largest trade-off cost among the objectives is calculated and selected as the final optimal solution, thereby achieving the prediction of spray pressure and spray flow rate.

[0057] For example, consider the glass skylight of a university library, with an area of ​​approximately 320m², made of three layers of ultra-clear tempered laminated glass. Its visible light transmittance τ_vis=0.78, solar heat gain coefficient SHGC=0.62, and peak summer transmitted radiation of approximately 650~850W / m².

[0058] A solar intensity meter and a small weather station are installed on the outer surface of the glass skylight; a nine-point illuminance meter and a black ball thermometer are arranged on the indoor work surface; an electromagnetic flow meter and a pressure transmitter are installed on the main spray pipe; and the air conditioner and spray pump are connected to a power meter. All sensors are connected to the edge computing gateway via an RS485 interface, and the sampling period can be set to 10 seconds.

[0059] Based on the above embodiment, PMV and DGP are calculated. The glare penalty coefficient λ is set to 5. Then the multi-objective optimization model is: J=w1·Eac+w2·W+w3·P+w4·(PMV²+5·DGP) The input constraints for the multi-objective optimization model include: -0.5≤PMV≤+0.5, DGP≤0.35, 0.2MPa≤spray pressure p≤0.8MPa, and 0≤spray flow rate Q≤120L / min.

[0060] Set the dynamic weights for the multi-objective optimization model, including: CPI = (PMV² + 2·DGP) / ( E price / 0.8), when CPI>1.2, w4=0.6, w1=w2=w3=0.133; when CPI≤1.2, w4=0.3, w1=w2=w3=0.233.

[0061] MPC was used to solve the multi-objective optimization model in a rolling manner. The prediction time domain was set to 20 minutes and the control time domain to 4 minutes. The meteorological forecast was based on Python-pvlib+LSTM and input of the past 24 hours of data and numerical weather prediction.

[0062] The optimization is performed once per minute using NSGA-II from the pymoo library, with a population size of 200 and a generation number of 50. The Pareto front is calculated, and then the TOPSIS method is used to select the inflection point solution. With the help of an accelerator, the solution time can be less than 0.8 seconds.

[0063] The optimal spray pressure p obtained from the solution is sent to the frequency converter of the spray system, and the spray flow rate Q is sent to the proportional valve of the spray system, thereby achieving precise control of the spray pressure and flow rate.

[0064] Assuming a traditional timed and pressure-controlled spraying system operates at a fixed rate from 10:00 to 16:00 daily, with a spray pressure of 0.50 MPa, a spray flow rate of 90 L / min, and a constant pump power of 2.2 kW, the embodiment of this invention, through the above-mentioned dynamic solution for optimal spray pressure and optimal spray flow rate, achieves a total air conditioning power saving of 877 kWh, approximately 29.4% lower than the traditional system. Spray water consumption is reduced by 192 m³, 53.2% lower than the traditional system, and the percentage of time with PMV in the [-0.5, +0.5] range increases from 73% to 96%. Specific comparisons can be found in Table 1 below.

[0065] Table 1. Comparison of measured effects of traditional timed and pressured spraying with the dynamic control effect of the present invention.

[0066] As another alternative implementation, if a deep reinforcement learning algorithm is used to solve the multi-objective optimization model in a rolling manner, refer to... Figure 4 As shown, the state space of the deep reinforcement learning algorithm includes the outdoor temperature of the translucent building, outdoor relative humidity, outdoor solar radiation, predicted average vote, glare probability, and real-time energy consumption cost. The action space of the deep reinforcement learning algorithm includes the spray pressure increment and spray flow rate increment. A reward function for the deep reinforcement learning algorithm is established by setting a first reward weight for the first real-time energy consumption, a second reward weight for the real-time water consumption, a third reward weight for the second real-time energy consumption, a fourth reward weight for the predicted average vote, and a fifth reward weight for the glare probability. The first, second, third, fourth, and fifth reward weights satisfy a normalization condition.

[0067] In practical implementation, deep reinforcement learning algorithms can employ either DDPG (Deep Deterministic Policy Gradient) or TQC (Tensor Quantization Config) algorithms. The reward function for these deep reinforcement learning algorithms can be referenced in the following formula: R=-(αEac +βW+γP)+δ / (|PMV|+ε)- η ·max(DGP-0.35,0) In the formula, α is the first reward weight, β is the second reward weight, γ is the third reward weight, δ is the fourth reward weight, η is the fifth reward weight, α+β+γ+δ+η=1, ε is a small positive number to prevent division by zero, and E ac W represents the first real-time energy consumption, P represents the real-time water consumption, P represents the second real-time energy consumption, PMV represents the predicted average vote, and DGP represents the glare probability.

[0068] In the rolling solution process using a deep reinforcement learning algorithm, the outdoor temperature, relative humidity, solar radiation, predicted average votes, glare probability, and real-time energy cost of the translucent building are continuously acquired. Adjustments to the spray pressure and flow rate increments are then output. After executing the corresponding spray adjustment actions on the target translucent building, the deep reinforcement learning algorithm obtains a reward signal that integrates energy cost and comfort penalty. Through extensive training and trial and error, an optimal deep reinforcement learning algorithm is trained. In subsequent applications, by acquiring the aforementioned parameters such as outdoor temperature, relative humidity, solar radiation, predicted average votes, glare probability, and real-time energy cost of the translucent building in real time, the optimal spray pressure and flow rate can be output in milliseconds.

[0069] As an optional implementation, when the intensity of transmitted solar radiation is detected to be lower than a preset radiation intensity threshold, the spray system can be controlled based on the intermittent spray duty cycle, where the intermittent spray duty cycle is obtained by the ratio of the spray on-time to the spray off-time. While meeting basic cooling and anti-glare requirements, the ineffective spraying time is significantly reduced by periodically starting and stopping the spray system, thereby significantly reducing the system's total water and energy consumption with almost no impact on comfort.

[0070] To achieve forward-looking control, meteorological parameters in a certain future time domain can also be predicted. Specifically, data-driven time-series forecasting methods can be used, and the forecasting model can be a Long Short-Term Memory (LSTM) network or an Autoregressive Integrated Moving Average (ARIMA) model.

[0071] The inputs to the prediction model can include historical meteorological data and numerical weather prediction (NWP). Historical meteorological data reflects the climate evolution patterns and short-term trends of the target translucent building's location and can be historical meteorological data collected by local sensors, including temperature, humidity, and solar radiation intensity. Numerical weather prediction data comes from external services and provides weather forecasts for the next few hours to days. By using high-precision numerical weather prediction data as input constraints for the prediction model, the prediction bias that can easily arise from relying solely on historical meteorological data for weather forecasting can be effectively overcome, significantly improving the accuracy of subsequent weather predictions.

[0072] The trained prediction model can output future weather forecasts in a rolling manner. By inputting the weather forecasts into the multi-objective optimization model, the evolution trajectory of indoor thermal environment and energy consumption under different spray strategies can be calculated in advance, thereby making more scientific and energy-saving spray system control decisions and enhancing the robustness of spray and air conditioning systems in response to weather changes.

[0073] Step S104: Control the spray system based on the predicted spray pressure and spray flow rate.

[0074] The variable frequency pump of the spray system is controlled based on the predicted spray pressure value, thereby changing the spray pressure of the spray system. Similarly, the proportional valve of the spray system is controlled based on the predicted spray flow value, thereby changing the spray flow of the spray system.

[0075] Secondly, embodiments of the present invention provide a multi-objective optimization control system for translucent roof spray shading evaporative cooling, applicable to translucent buildings equipped with air conditioning and spray systems, as shown in the following example. Figure 5 As shown, the system includes: The data acquisition device 201 is used to acquire outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the light-transmitting building, acquire the first real-time energy consumption of the air conditioning system, and acquire the real-time water consumption and second real-time energy consumption of the spray system. The data processing device 202 is used to calculate the predicted average vote and glare probability of the transparent building based on the outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the transparent building. Based on the predicted average vote, glare probability, first real-time energy consumption, real-time water consumption, and second real-time energy consumption, a multi-objective optimization model is established. The data processing device 202 is also used to take the predicted average vote and glare probability as the first optimization objective, take the minimum value of the first real-time energy consumption, real-time water consumption and the second real-time energy consumption as the second optimization objective, and use model predictive control or deep reinforcement learning algorithm to solve the multi-objective optimization model in a rolling manner to calculate the predicted value of spray pressure and the predicted value of spray flow. The execution control device 203 is used to control the spray system based on the predicted spray pressure and the predicted spray flow.

[0076] In practical implementation, the data processing device 202 can be a cloud-based digital twin platform. This cloud-based digital twin platform has sufficient computing power, supports the NWP interface, and can perform computationally intensive tasks such as offline training of agent models, multi-objective optimization models, and prediction models. The data acquisition device 201 can be an outdoor weather station, an indoor environmental sensor, and an energy consumption metering device, etc., and the execution control device 203 can be a variable frequency pump and a proportional valve in a spray system.

[0077] like Figure 6 As shown, an outdoor weather station 501, an indoor environmental sensor 502, and an energy consumption metering device 503, installed on the target translucent building, collect outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters, which are then sent to an edge computing gateway 504. The edge computing gateway 504 uses this data to calculate and predict the average vote and glare probability, and performs calculations using a built-in multi-objective optimization model to obtain predicted spray pressure and spray flow rates. The edge computing gateway 504 then sends these predicted values ​​to the spray system, where the variable frequency pump 505 adjusts the spray pressure, and the proportional valve 506 adjusts the spray flow rate, thus enabling intelligent spraying of the target translucent building. Additionally, a cloud-based digital twin platform 507 can be installed. This platform can periodically update the multi-objective optimization model in the edge computing gateway 504, ensuring the long-term accuracy and robustness of the spray control.

[0078] For example, to demonstrate the universality and predictability of this method's effectiveness, the applicant has established a high-fidelity digital twin platform using EnergyPlus (the name of an energy consumption simulation software) 23.2+Python MPC closed-loop interface. The simulation results of the high-fidelity digital twin platform deviate from the actual field measurements by less than 8%, demonstrating that reliable results can be obtained in advance through the digital twin platform, without relying on long-term field testing to verify the solution.

[0079] The simulation platform consists of the following components: EnergyPlus 23.2+Radiance is used for building and lighting; a calculation model for the photothermal performance of a water mist-transmitting roof is established based on Mie scattering theory and evaporative cooling theory, and is coupled into EnergyPlus 23.2 for calculation through the EnergyPlus Python plugin; the control algorithm uses real MPC code embedded in the BCVTB (Building Controls Virtual Test Bed) interface; numerical weather prediction can include typical meteorological years from previous years and high-frequency measured data; the cost is set according to the price of the actual application site, for example, the non-residential water price is set at 4.40 yuan / m³, and the electricity price is the time-of-use price.

[0080] Calibration using measured data showed that the simulation results of the digital twin platform deviated from the measured air conditioning power consumption by 3.4%, water consumption by 8%, and PMV compliance rate by 2.1%, with R² > 0.92, meeting the engineering accuracy requirements.

[0081] The dynamic simulation results for a certain region over an annual period of 8760 hours (focusing on the cooling season from May to October) are shown in Table 2 below. Using the MPC dynamic optimization method of this invention, while prioritizing water consumption reduction, air conditioning energy savings of approximately 28%–29% and a total operating cost reduction of approximately 51%–59% for the spray system can still be achieved, while significantly improving the indoor thermal comfort compliance rate. These effects can be reliably predicted using the aforementioned high-fidelity digital twin simulation platform, greatly reducing the technology verification cycle and cost.

[0082] Table 2. Annual Simulation Comparison of Control Effects between Traditional Timed and Pressure-Controlled Systems and the Invention

[0083] Since the multi-objective optimization control system for light-transmitting roof spray shading evaporative cooling described in this embodiment is the electronic equipment used to implement the multi-objective optimization control method for light-transmitting roof spray shading evaporative cooling in this embodiment of the invention, those skilled in the art can understand the specific implementation and various variations of the electronic equipment in this embodiment based on the multi-objective optimization control method for light-transmitting roof spray shading evaporative cooling described in this embodiment of the invention. Therefore, how the electronic equipment implements the method in this embodiment of the invention will not be described in detail here. Any electronic equipment used by those skilled in the art to implement the multi-objective optimization control method for light-transmitting roof spray shading evaporative cooling in this embodiment of the invention falls within the scope of protection of this invention.

[0084] In summary, the multi-objective optimization control method and system for evaporative cooling and shading of translucent roofs provided in this invention obtains and calculates the predicted average vote and glare probability of the translucent building based on outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters. It then obtains the first real-time energy consumption of the air conditioning system and the real-time water consumption and second real-time energy consumption of the spray system. Based on the predicted average vote, glare probability, first real-time energy consumption, real-time water consumption, and second real-time energy consumption, a multi-objective optimization model is established. The predicted average vote and glare probability are used as the first optimization objective, and the minimum value of the first real-time energy consumption, real-time water consumption, and second real-time energy consumption is used as the second optimization objective. Model predictive control or deep reinforcement learning algorithms are used to solve the multi-objective optimization model in a rolling manner, enabling the calculation of predicted spray pressure and predicted spray flow rate.

[0085] The spray system is controlled based on predicted spray pressure and flow rates, causing the variable frequency pumps and proportional valves to perform corresponding actions. This ensures indoor light and heat comfort while minimizing water and energy consumption. Furthermore, because the spray system dynamically alters the heat gain and thermal humidity environment inside the translucent building, it indirectly controls the operating load and energy consumption of the air conditioning system. Ultimately, this achieves synergistic energy saving between the spray and air conditioning systems, optimizing energy consumption in both systems and in relation to light and heat comfort. Actual measurements show a 28.6% reduction in air conditioning power consumption, a 51.2% reduction in water consumption, and a 96% compliance rate with predicted average voting values. This system is suitable for large-area translucent roof buildings in various climate zones.

[0086] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0087] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0088] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0091] It should also be noted that the exemplary embodiments mentioned in this invention describe methods or systems based on a series of steps or apparatus. However, this invention is not limited to the order of the steps described above; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0092] The above are merely specific embodiments of the present invention. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. A multi-objective optimization control method for evaporative cooling and shading of translucent roofs using mist spraying, characterized in that, The method, applied to translucent buildings equipped with air conditioning and misting systems, includes: Based on the outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the translucent building, calculate the predicted average vote and glare probability of the translucent building; obtain the first real-time energy consumption of the air conditioning system, and obtain the real-time water consumption and second real-time energy consumption of the spray system; A multi-objective optimization model is established based on the predicted average vote, the glare probability, the first real-time energy consumption, the real-time water consumption, and the second real-time energy consumption. The predicted average vote and the glare probability are taken as the first optimization objective, and the minimum value of the first real-time energy consumption, the real-time water consumption and the second real-time energy consumption is taken as the second optimization objective. The multi-objective optimization model is solved by rolling using model predictive control or deep reinforcement learning algorithm to calculate the predicted value of spray pressure and the predicted value of spray flow. The spray system is controlled based on the predicted spray pressure and the predicted spray flow rate.

2. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 1, characterized in that, The multi-objective optimization model is established based on the predicted average vote, the glare probability, the first real-time energy consumption, the real-time water consumption, and the second real-time energy consumption, including: A first dynamic weight is set for the first real-time energy consumption of the air conditioning system, a second dynamic weight is set for the real-time water consumption of the spray system, a third dynamic weight is set for the second real-time energy consumption of the spray system, and the predicted average vote and the glare probability are combined into a comprehensive comfort deviation index. A fourth dynamic weight is set for the comprehensive comfort deviation index to construct an initial optimization objective function; wherein, a glare penalty coefficient is preset for the glare probability. The current energy consumption cost and the preset baseline energy consumption cost are obtained in real time. The comfort priority index is calculated by combining the predicted average vote and the glare probability calculated at the current time. The first dynamic weight, the second dynamic weight, the third dynamic weight, and the fourth dynamic weight are adjusted in real time according to the comfort priority index to establish the multi-objective optimization model in order to meet the normalization condition.

3. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 2, characterized in that, The process involves acquiring the current energy consumption cost and the preset baseline energy consumption cost in real time, combining them with the predicted average vote calculated at the current moment and the glare probability, to calculate a comfort priority index, including: The system acquires the real-time electricity price signal and / or real-time water price signal at the current moment, and calculates the real-time energy consumption cost, which represents the comprehensive operating cost of the current system, by weighting the energy type consumed by the first real-time energy consumption of the air conditioning system and the resource type consumed by the real-time water consumption and the second real-time energy consumption of the spray system. Wherein, if both the air conditioning system and the spray system consume electrical energy, the real-time energy consumption cost is taken as the time-of-use electricity price at the current moment; if multiple energy media with different billing standards are involved, the cost is obtained by weighted summation based on the consumption ratio of each energy media and its corresponding unit price. Obtain the historical average energy price of the location of the translucent building during the cooling season or typical operating cycle, or set the value according to the user's preset economic operating target, as the benchmark energy cost; Based on the predicted average vote and the glare probability, a comprehensive comfort deviation is constructed. The comprehensive comfort deviation is then correlated with the ratio of real-time energy consumption cost to baseline energy consumption cost to calculate the comfort priority index.

4. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 2, characterized in that, The step of adjusting the first dynamic weight, the second dynamic weight, the third dynamic weight, and the fourth dynamic weight in real time according to the comfort priority index includes: Compare the relationship between the currently calculated comfort priority index and the preset index threshold range; If the comfort priority index rises and exceeds the first preset threshold, it is determined that the current indoor light and heat comfort is deteriorating or the real-time energy consumption cost is low. Then, the value of the fourth dynamic weight is increased, and the values ​​of the first dynamic weight, the second dynamic weight, and the third dynamic weight are decreased accordingly. If the comfort priority index decreases and falls below the second preset threshold, it is determined that the current indoor light and heat comfort is good or the real-time energy consumption cost is high. Then, the values ​​of the first dynamic weight, the second dynamic weight, and the third dynamic weight are increased, and the value of the fourth dynamic weight is decreased accordingly. After the real-time adjustment is completed, the first dynamic weight, the second dynamic weight, the third dynamic weight, and the fourth dynamic weight are constrained to satisfy the normalization condition.

5. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 1, characterized in that, If model predictive control is used to solve the multi-objective optimization model in a rolling manner, after the prediction time domain and control time domain of the model predictive control are set, the optimal solution set is solved by a multi-objective evolutionary algorithm, and the knee point is selected from the optimal solution set by the fuzzy comprehensive evaluation method to calculate the predicted spray pressure value and the predicted spray flow rate value.

6. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 5, characterized in that, The step of using model predictive control to perform rolling solutions on the multi-objective optimization model specifically includes: The prediction time domain and control time domain of the model predictive control are set. At each rolling optimization time, based on the currently obtained outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, indoor thermal environment parameters, first real-time energy consumption, real-time water consumption and second real-time energy consumption, the multi-objective optimization model is used as the objective function, and the spray pressure and spray flow rate are used as decision variables to construct a constrained optimization problem with the prediction time domain as the span. The constrained optimization problem is solved by a multi-objective evolutionary algorithm to obtain a Pareto optimal solution set consisting of multiple non-dominated solutions. Each solution in the Pareto optimal solution set corresponds to a set of candidate control sequences for the spray pressure and the spray flow rate. The fuzzy comprehensive evaluation method is used to evaluate the overall satisfaction of each candidate solution in the Pareto optimal solution set, including: defining fuzzy membership functions for the first optimization objective and the second optimization objective respectively, calculating the weighted comprehensive satisfaction of each candidate solution, and selecting the solution with the highest weighted comprehensive satisfaction and the largest trade-off cost between the first optimization objective and the second optimization objective from the Pareto optimal solution set as the knee point; Extract the control increments of the spray pressure and spray flow rate corresponding to the knee point at the current control moment to generate the predicted values ​​of the spray pressure and spray flow rate.

7. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 1, characterized in that, If a deep reinforcement learning algorithm is used to solve the multi-objective optimization model in a rolling manner, the state space of the deep reinforcement learning algorithm includes the outdoor temperature of the light-transmitting building, the outdoor relative humidity, the outdoor solar radiation, the predicted average vote, the glare probability, and the real-time energy consumption cost; the action space of the deep reinforcement learning algorithm includes the spray pressure increment and the spray flow rate increment. A first reward weight is set for the first real-time energy consumption, a second reward weight is set for the real-time water consumption, a third reward weight is set for the second real-time energy consumption, a fourth reward weight is set for the predicted average vote, and a fifth reward weight is set for the glare probability, so as to establish the reward function of the deep reinforcement learning algorithm. Among them, the first reward weight, the second reward weight, the third reward weight, the fourth reward weight, and the fifth reward weight satisfy the normalization condition.

8. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 1, characterized in that, Also includes: When the intensity of transmitted solar radiation is detected to be lower than a preset radiation intensity threshold, the spray system is controlled based on the intermittent spray duty cycle, wherein the intermittent spray duty cycle is obtained by using the ratio of the spray on duration to the spray off duration.

9. The multi-objective optimization control method for evaporative cooling and shading of translucent roofs according to claim 1, characterized in that, The first real-time energy consumption is predicted based on a preset proxy model, the input of which includes the outdoor meteorological parameters and water mist parameters.

10. A multi-objective optimization control system for light-transmitting roof spray shading evaporative cooling, characterized in that, A translucent building equipped with an air conditioning system and a misting system, the system comprising: The data acquisition device is used to acquire outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the light-transmitting building; to acquire the first real-time energy consumption of the air conditioning system; and to acquire the real-time water consumption and second real-time energy consumption of the spray system. The data processing device is used to calculate the predicted average vote and glare probability of the translucent building based on the outdoor meteorological parameters, transmitted solar radiation intensity, indoor illuminance distribution, and indoor thermal environment parameters of the translucent building, and to establish a multi-objective optimization model based on the predicted average vote, the glare probability, the first real-time energy consumption, the real-time water consumption, and the second real-time energy consumption. The data processing device is further configured to take the predicted average vote and the glare probability as the first optimization objective, take the minimum value of the first real-time energy consumption, the real-time water consumption and the second real-time energy consumption as the second optimization objective, and use model predictive control or deep reinforcement learning algorithm to perform rolling solution on the multi-objective optimization model to calculate the predicted value of spray pressure and the predicted value of spray flow. An execution control device is used to control the spray system based on the predicted spray pressure and the predicted spray flow rate.

Images