Energy-saving control system and method for heating and ventilation equipment

Abstract

The invention discloses an energy-saving control system and method for heating and ventilation equipment, and belongs to the technical field of building energy conservation, and the system comprises a data obtaining module which is used for obtaining environment observation data of a target space; the flow field prediction module constructs a first fusion convolutional neural network and is used for predicting target space flow field information at a future moment and obtaining steady-state heat and humidity characteristics based on the target space flow field information; the operation backstepping module constructs a second fusion convolutional neural network and is used for predicting heating and ventilation equipment operation state data corresponding to the candidate environment data; and the operation optimization module constructs an optimization objective function, optimizes the environment data to be adjusted from the candidate environment data, and determines a heating and ventilation equipment control strategy. Through the method of fusing the convolutional neural network and data driving, the dependence on complex modeling and simulation tools is reduced, so that the implementation cost of the system is reduced, and the real-time responsiveness and adaptability of the model are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of supplier management, and more specifically, to a heating and ventilation equipment energy-saving control system and method. Background Art

[0002] Methods for optimizing control of heating, ventilation and air conditioning (HVAC) systems are mainly divided into two categories, namely model-based control methods and model-free control methods. The field of model-free control methods mainly includes rule-based control methods and reinforcement learning-based control methods. At present, rule-based control methods often combine the powerful prediction function of machine learning to achieve optimized control, but their intelligence is relatively low, relying on the experience of engineers and equipment managers, and the scalability of combined optimized control of complex HVAC control systems is insufficient. For example, the user activities of office buildings have obvious regularity. Air conditioning managers can formulate the operation schedule of the air conditioning system according to this regularity and control the HVAC system according to the schedule, but this control method lacks flexibility. In many cases, the set schedule does not match the actual flow of personnel. This difference may cause the air conditioning system to start meaninglessly when no one is indoors, and excessive ventilation when there are fewer people indoors, resulting in energy waste. One solution to this problem is to use the actual user occupancy rate measured by sensors to control the air conditioning system, but due to the thermal inertia of indoor air, the reactive control system may not be able to ensure thermal comfort when the user just enters the room.

[0003] In response to the above problems, the existing patent technology CN115540252A discloses a HVAC energy-saving control system and method based on machine learning. The method includes determining the target thermal zone according to the building model and building HVAC model of the target building; obtaining historical energy consumption observation data and energy consumption simulation data of the target thermal zone, optimizing the building HVAC model according to the energy consumption simulation data and the historical energy consumption observation data, and determining the target building HVAC model; obtaining energy consumption prediction data according to the target building HVAC model; obtaining human comfort data, and establishing a HVAC control model according to the energy consumption prediction data and the human comfort data; determining the HVAC control strategy of the target thermal zone according to the HVAC control strategy; and adjusting the HVAC supply air temperature of the target thermal zone according to the HVAC control strategy. Since this method requires the use of professional building simulation software such as EnergyPlus and TRNSYS to model the target room, or to reversely establish a data-driven room model through detailed sampling of indoor and outdoor parameters, and combine the air-conditioning model to optimize the temperature and humidity set points, the entire process requires a large amount of manpower and material resources, which is unacceptable to some users and is difficult to use as a universal solution, making its prospects for practical application unclear. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] The purpose of this application is to provide an energy-saving control system and method for HVAC equipment, which solves the technical problem that the existing energy-saving control system for HVAC equipment proposed in the above background technology relies on a large amount of detailed sampling, reverse modeling and professional simulation, resulting in high manpower and material costs and difficult to promote as a universal solution. By integrating convolutional neural networks and data-driven methods, the dependence on complex modeling and simulation tools is reduced, thereby reducing the implementation cost of the system and improving the real-time responsiveness and adaptability of the model.

[0006] 2. Technical solution

[0007] Based on the above purpose, the present application provides a HVAC equipment energy-saving control system, including:

[0008] A data acquisition module is used to obtain environmental observation data of the target space;

[0009] The flow field prediction module constructs the first fusion convolutional neural network to predict the target space flow field information at the future moment and obtain the steady-state thermal and humid characteristics based on the target space flow field information;

[0010] Run the reverse inference module to construct the second fusion convolutional neural network to predict the HVAC equipment operation status data corresponding to the candidate environmental data;

[0011] Run the optimization module, construct the optimization objective function, select the environmental data to be adjusted from the candidate environmental data, and determine the HVAC equipment control strategy.

[0012] Furthermore, the operation control module adjusts the operation status of the HVAC equipment in the target space according to the HVAC equipment control strategy determined by the operation optimization module, and controls the environmental data of the target space to the environmental data to be adjusted.

[0013] Preferably, a weighted fuzzy logic control algorithm is used in the operation control module to simultaneously control the temperature and humidity in the environmental data of the target space, so as to stabilize the temperature and humidity of the target space at the preferred environmental data to be adjusted.

[0014] Preferably, the data acquisition module includes sensors provided by the HVAC equipment itself, including but not limited to temperature sensors, humidity sensors, pressure sensors and air quality sensors, which are used to obtain the air temperature, humidity, pressure and CO2 concentration in the target space.

[0015] Preferably, the flow field prediction module constructs a first fused convolutional neural network based on the data set, including:

[0016] Combine environmental data with HVAC equipment operation data to form a fused data set;

[0017] The time dimension is divided into multiple time steps, and the convolutional neural network is trained based on the fused data set. For each time step, one network is selected as the representative network, and the representative networks of each time step are fused to form the first fused convolutional neural network.

[0018] Further preferably, the fused data set is input into the convolutional neural network in the order of time steps, the flow field of the corresponding time step is output, the output is substituted into the finite volume control equation of the corresponding time step for solution, the network that minimizes the residual of the finite volume control equation in the time step is used as the representative network, the flow field in the representative network is used as the basis for solving the next time step, and the loss function is used to associate and fuse the representative networks of all time steps to form a first fused convolutional neural network.

[0019] Further preferably, constructing a second fused convolutional neural network in the running back-inference module includes:

[0020] Combine environmental data with indoor load data to form a new fused data set;

[0021] Under the state of heat and moisture balance, the HVAC equipment operation data input in the input layer of the first fusion convolutional neural network and the flow field information data output in the output layer are swapped to establish the second fusion convolutional neural network.

[0022] Preferably, an optimization objective function is constructed in the operation optimization module to balance the operating power of the HVAC equipment and the environmental thermal comfort of the target space. The optimization objective function is:

[0023]

[0024] Among them, F i is the target subfunction to be balanced, γ i is the corresponding balance weight, μ i is the correction coefficient of the target sub-function to be balanced, and n represents the number of the target sub-functions to be balanced; the environmental data to be adjusted that minimizes the optimization target function is selected from the candidate environmental data;

[0025] Without considering the impact of the target space occupancy rate on the HVAC equipment operating power, the total power target sub-function F1 of the HVAC equipment is:

[0026]

[0027] Among them, P total is the total power of HVAC equipment, P min and P max It is the minimum and maximum value of the total power of HVAC equipment;

[0028] The value range of the environmental thermal comfort target sub-function F2 of the target space is [0,0.25].

[0029] Further preferably, a balance weight optimization model is constructed based on deep learning, and the balance weight is obtained according to the target space state using the balance weight optimization model to calculate the optimization objective function.

[0030] The present application also provides a method for controlling energy saving of a heating and ventilation equipment, comprising the following steps:

[0031] Step 1, obtaining environmental observation data of the target space;

[0032] Step 2: Combine the environmental data with the HVAC equipment operation data to form a fused data set, and build a first fused convolutional neural network based on the fused data set to predict the target space flow field information at the future moment, and obtain the steady-state thermal and humid characteristics based on the target space flow field information;

[0033] Step 3, constructing a second fused convolutional neural network based on the fused data set to predict the HVAC equipment operating status data corresponding to the candidate environmental data;

[0034] Step 4: construct an optimization objective function, select the environmental data to be adjusted from the candidate environmental data, and determine the control strategy of the HVAC equipment;

[0035] Step 5, using a weighted fuzzy logic control algorithm to simultaneously control the temperature and humidity in the environmental data of the target space, and stabilize the temperature and humidity of the target space at the preferred environmental data to be adjusted.

[0036] 3. Beneficial effects

[0037] The HVAC equipment energy-saving control system and method of the present application have at least the following technical effects or advantages:

[0038] Through the first and second fusion convolutional neural networks based on deep learning, this solution does not need to rely on cumbersome building and HVAC model construction, and can quickly and accurately predict the flow field information of the target space and the operating status of HVAC equipment, thereby greatly reducing the modeling cost. In addition, the introduction of the optimization objective function and weighted fuzzy logic control algorithm effectively balances energy consumption and environmental thermal comfort control, further improving the energy-saving effect and user experience of the system. Compared with the existing software methods based on detailed simulation, this solution has higher versatility and real-time responsiveness, is applicable to more scenarios, and has achieved significant improvements in energy saving and comfort control.

[0039] By integrating convolutional neural networks to predict future flow field information and HVAC equipment operating status, and based on the optimization objective function, the equipment operating power and environmental thermal comfort are balanced. The weighted fuzzy logic control algorithm is used to accurately control temperature and humidity, achieving efficient energy saving while ensuring environmental comfort. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a structural diagram of the energy-saving control system of the HVAC equipment in this application;

[0041] Figure 2 This is the structure diagram of the data acquisition module in the energy-saving control system of the HVAC equipment in this application;

[0042] Figure 3 It is a flow chart of the energy-saving control method of HVAC equipment in this application. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0044] Theoretically, energy-saving control of HVAC equipment can be divided into two technical routes. The first is to optimize hardware equipment, such as improving the performance of compressors or heat exchangers, and gradually replace fixed-frequency compressors with variable-frequency compressors. In theory, variable-frequency compressors can better match the room load under partial load conditions and avoid energy loss caused by frequent switching on and off. The second is to improve the control algorithm to give full play to the capabilities of variable-frequency compressors. For example, how to better match the air-conditioning cooling capacity with the room load under complex and changeable real working conditions, thereby reducing unnecessary losses caused by fluctuations in indoor thermal and humid environment. It can also further optimize the air-conditioning control targets to meet users' thermal comfort needs while reducing air-conditioning energy consumption.

[0045] Energy-saving control algorithms for HVAC equipment are generally divided into reactive control and predictive control. From a time perspective, reactive control responds to current or historical parameters, including commonly used feedback control methods, while predictive control focuses on making predictions about the future and taking actions based on the predictions. It should be noted that some predictive algorithms also include reactive control, such as determining the temperature set point based on predictions of air conditioning energy consumption at different temperature set points. After determining the set point, feedback methods are also relied upon to achieve temperature control. They are generally classified based on their technical core. The core of this example is to determine the temperature set point based on predictions, which can be considered predictive control.

[0046] Among them, reactive control can be further divided into two situations: objective state-oriented and user-oriented. Objective state includes measurement parameters such as indoor temperature, humidity, CO2 concentration, volatile organic compound (VOC) concentration, and calculation parameters such as predicted average vote and predicted dissatisfaction percentage calculated based on these measurement parameters. Reactive control oriented to objective state is based on the premise that there is a close and universal connection between user thermal comfort and objective parameters, and finds energy-saving space for air conditioning through more scientific parameter sets and / or more advanced control algorithms. User-oriented reactive control focuses on the randomness and differences of users. When there are no users or a small number of users in a certain air-conditioned space, the average energy consumption of the air-conditioning system can be reduced by shutting down the air-conditioning system, raising the air-conditioning set point, and reducing the air volume of the fresh air system, that is, reactive control based on user occupancy rate. In addition, the air-conditioning system can be adjusted based on the differences in thermal comfort perception of different individuals or groups, that is, reactive control based on individual thermal comfort.

[0047] The research and design of the energy-saving control algorithm for HVAC equipment in the present invention aims to organically combine reactive control and predictive control and reduce its dependence on complex modeling and simulation tools.

[0048] Example 1

[0049] The embodiment of the present invention provides a heating and ventilation equipment energy-saving control system, see Figure 1 to Figure 2 ,include:

[0050] The data acquisition module 100 is used to acquire the environmental observation data of the target space;

[0051] The flow field prediction module 200 constructs a first fusion convolutional neural network to predict the target space flow field information at a future moment, and obtains the steady-state thermal and humid characteristics based on the target space flow field information;

[0052] Running the inference module 300 to construct a second fused convolutional neural network for predicting HVAC equipment operating status data corresponding to the candidate environmental data;

[0053] The optimization module 400 is run to construct an optimization objective function, select the environmental data to be adjusted from the candidate environmental data, and determine the control strategy of the HVAC equipment.

[0054] Based on the above system, the HVAC equipment energy-saving control system also includes:

[0055] The operation control module 500 adjusts the operation state of the HVAC equipment in the target space according to the HVAC equipment control strategy determined by the operation optimization module 400, and controls the environmental data of the target space to the environmental data to be adjusted. Specifically, the operation control module adopts a weighted fuzzy logic control algorithm to simultaneously control the temperature and humidity in the environmental data of the target space, and stabilizes the temperature and humidity of the target space at the preferred environmental data to be adjusted.

[0056] In this embodiment, the data acquisition module 100 is first used to acquire the environmental observation data of the target space, such as using the sensors provided by the HVAC equipment itself, wherein the sensors include temperature sensors, humidity sensors, pressure sensors and air quality sensors, respectively acquiring the measurement parameters such as the air temperature, humidity, pressure, CO2 concentration and volatile organic compound (VOC) concentration in the target space. Considering that the CO2 concentration can be used as a reference indicator for the number of people in the user's room, the CO2 concentration is used as a direct indicator of the objective state of indoor air quality in this embodiment. At the same time, the operation data of the HVAC equipment is recorded and stored in the database.

[0057] The flow field prediction module 200 retrieves the environmental observation data collected and stored in the database by the data acquisition module 100, and retrieves the operation data of the HVAC equipment, and uses the environmental observation data and the HVAC equipment operation data to build and train a prediction model that can be used to quickly predict the numerical simulation results of the target space flow field, including:

[0058] The environmental data is combined with the HVAC equipment operation data to form a fused data set. Specifically, the following method can be used: based on different temperature and humidity combinations in the environmental data and different speed combinations of compressor speed and fan speed in the HVAC equipment, a plurality of different training data are randomly combined to establish the training data, and the training data is not marked. In this embodiment, the temperature and humidity combination includes dry-bulb temperature, wet-bulb temperature and relative humidity parameters. The compressor speed and fan speed combination in the HVAC equipment can be set with the same change trend, such as setting the ratio of the speed to the rated speed as a parameter, and the interval difference between two adjacent parameter points is kept consistent. The two parameter data are randomly combined to generate 200 training data [temperature, humidity, compressor speed, fan speed] under different temperature and humidity combinations and different compressor and fan speed combinations to form a fused data set;

[0059] The time dimension is divided into multiple time steps, and the convolutional neural network is trained based on the fused data set. For each time step, a network is selected as the representative network, and the representative networks of each time step are fused to form the first fused convolutional neural network. Specifically, the fused data set is input into the convolutional neural network in the order of the time steps, and the flow field of the corresponding time step is output. The output is substituted into the finite volume control equation of the corresponding time step for solution. The network that minimizes the residual of the finite volume control equation in the time step is used as the representative network, and each convolutional neural network as the representative network corresponds to one time step. The representative networks of each time step are fused to form the first fused convolutional neural network, and the flow field in the representative network is used as the basis for solving the next time step. The representative networks of all time steps are associated and fused using the loss function to form the first fused convolutional neural network.

[0060] The fusion data [temperature, humidity, compressor speed, fan speed] formed by the newly acquired environmental data and the HVAC equipment operation data is input into the trained first fusion convolutional neural network to obtain the flow field at any time step. Based on the flow field information data and the HVAC equipment operation data, the sensible heat and cooling capacity and latent heat and cooling capacity output under the corresponding temperature and humidity combination and the compressor and fan speed combination are calculated. Under the heat and humidity balance state, the heat load and humidity load of the target space are equal to the sensible heat and cooling capacity and latent heat and cooling capacity of the air conditioner, respectively. The heat load Q of the target space S and wet load Q L The calculation formula is:

[0061]

[0062] Among them, T in and W in Indoor temperature and humidity in space can be obtained through the temperature and humidity sensors of HVAC equipment. and is the constant term of heat and humidity load. By using any two sets of target space heat and humidity load data under different temperature and humidity combinations, the relationship between the room heat and humidity load and the indoor temperature and humidity under a stable state can be solved, that is, the steady-state heat and humidity characteristics of the target space.

[0063] A second fused convolutional neural network is constructed in the operation reverse inference module 300 to predict the HVAC equipment operation status data corresponding to the candidate environmental data. Since the cooling capacity is equal to the load under the heat and humidity balance state, the HVAC equipment operation data input in the input layer of the first fused convolutional neural network and the flow field information data output in the output layer are swapped to establish a new prediction model. At this time, the environmental data is combined with the indoor load data to form a new fused data set. The construction method of the fused data set used in the operation reverse inference module 300 is the same as that in the flow field prediction module 200, and the two fused data sets can be constructed based on the same original data without the need to add additional original data. Based on the new fused data set, the new prediction model is trained to obtain the second fused convolutional neural network. The second fused convolutional neural network input is the indoor dry-bulb temperature, wet-bulb temperature and the corresponding indoor heat and humidity load calculated, and the output is the compressor speed and fan speed that meet the corresponding refrigeration and dehumidification requirements under the temperature and humidity combination.

[0064] After using the first fused convolutional neural network to obtain the steady-state heat and humidity characteristics of the target space, the indoor heat and humidity load under the environmental data of any candidate temperature and humidity combination can be calculated. The candidate temperature and humidity combination and the corresponding indoor heat and humidity load are input into the second fused convolutional neural network to calculate the operating state required to stabilize the target space under the candidate temperature and humidity combination. The air conditioning power can be estimated through the operating state.

[0065] An optimization objective function is constructed in the operation optimization module 400 to select the environmental data to be adjusted from the candidate environmental data to balance the operating power of the HVAC equipment and the environmental thermal comfort of the target space. The optimization objective function is:

[0066]

[0067] Among them, F i is the target subfunction to be balanced, γ i is the corresponding balancing weight, n represents the number of the objective sub-functions to be balanced; the environmental data to be adjusted that minimizes the optimization objective function is selected from the candidate environmental data;

[0068] When the impact of the target space occupancy rate on the operating power of the HVAC equipment is not considered, the optimization objective function is the weighted sum of the two objective sub-functions, where the HVAC equipment total power objective sub-function F1 is:

[0069]

[0070] Among them, P total is the total power of HVAC equipment, the total power of HVAC equipment P total =P C +P F , P C and PF Represent the power of the compressor and fan respectively. The power of the compressor and fan at different speeds can be obtained by the HVAC equipment manufacturer based on the factory test data. min and P max It is the minimum and maximum total power of the HVAC equipment. The minimum total power is the sum of the minimum compressor power and the minimum fan power in the factory test data, and the maximum total power is the sum of the maximum compressor power and the maximum fan power in the factory test data. By adjusting the compressor speed and the fan speed, the total cooling capacity output by the HVAC equipment can be distributed between sensible heat and latent heat.

[0071] The value range of the target space environmental thermal comfort target subfunction F2 is [0, 0.25]. Taking PMV as an example, F2 = PMV 2 , the corresponding PMV value range can be determined according to different temperature and humidity combinations, and the corresponding PMV value range is [-0.5, 0.5], which meets the requirements of Class B thermal comfort environment specified in GB / T18049-2017 "Ergonomics of thermal environments-Analysis, measurement and interpretation of thermal comfort by calculating PMV and PPD indices and local thermal comfort criteria". It can also be adjusted to the requirements of Class A thermal comfort environment according to the characteristics of the target space, that is, the F2 value range is adjusted to [0, 0.04].

[0072] For example, according to GB / T 18049-2017 "Ergonomics of thermal environments - Analysis, determination and interpretation of thermal comfort by calculating PMV and PPD indices and local thermal comfort criteria", the PMV value is an index used to evaluate thermal comfort. It combines environmental parameters (such as temperature, humidity, wind speed, etc.) with human activities and clothing conditions to predict human thermal perception of the environment. Some of its input parameters are set as follows: the wind speed is set to 0.1m / s, the human metabolic rate is 1.0 (MetUnites), the average radiant temperature is equal to the indoor temperature, and taking office attire as an example, the thermal resistance of clothing is 1.0clo. Some PMV values ​​calculated in combination with different indoor temperature and humidity combinations are shown in the following table.

[0073] Table 1 PMV values ​​under different indoor temperature and humidity combinations (excerpt)

[0074]

[0075] The PMV value falls between [-0.5, 0.5] to ensure that thermal comfort is close to neutral (PMV value is close to 0). According to the PMV formula and thermal comfort theory, the human body usually feels warmer by increasing the temperature or reducing the humidity, and feels cooler by reducing the temperature or increasing the humidity.

[0076] Based on deep learning, a balance weight optimization model is constructed, and the balance weight is obtained according to the state of the target space using the balance weight optimization model to calculate the optimization objective function. Specifically, in this embodiment, based on different target sub-functions under different indoor temperature and humidity combinations, a weight allocation model is constructed using a deep neural network to determine the weights corresponding to the total power target sub-function F1 of the HVAC equipment and the environmental thermal comfort target sub-function F2 of the target space, and the optimization objective function is further calculated based on this, and the candidate environmental data that can minimize the optimization objective function is used as the environmental data to be adjusted. The deep neural network structure adopted is as follows: the input layer contains 2 neurons, namely the total power target subfunction F1 of the HVAC equipment and the environmental thermal comfort target subfunction F2 of the target space; the hidden layer has two layers, each containing 16 neurons; the hidden layer uses the ReLU function as the activation function; the output layer has 2 neurons, representing the weights corresponding to the total power target subfunction F1 of the HVAC equipment and the environmental thermal comfort target subfunction F2 of the target space; the activation function uses the linear function f(x)=wx+b; the loss function is the smooth L1 loss L1=(1 / n)∑|yy′|, where y is the true value, y′ is the predicted value, and n is the number of samples; the optimizer uses the stochastic gradient descent optimizer, the initial learning rate is 0.001, and it is decayed according to the change of the loss function, and the gradient is calculated using back propagation.

[0077] The balanced weight optimization model is used to determine the weights corresponding to the total power target sub-function F1 of the HVAC equipment and the environmental thermal comfort target sub-function F2 of the target space under different indoor temperature and humidity combinations. The weights are substituted into the optimization objective function for calculation to achieve a balance between thermal comfort and air-conditioning power. The calculation results are sorted, and the temperature and humidity data corresponding to the smallest calculation result are used as the environmental data to be adjusted. That is, the optimal temperature and humidity set points are selected from a series of candidate temperature and humidity combinations, and the HVAC equipment control strategy is determined accordingly.

[0078] Through the first and second fusion convolutional neural networks based on deep learning, this solution does not need to rely on cumbersome building and HVAC model construction, and can quickly and accurately predict the flow field information of the target space and the operating status of HVAC equipment, thereby greatly reducing the modeling cost. This solution has higher versatility and real-time responsiveness and is suitable for more scenarios. And by introducing the optimization objective function and weighted fuzzy logic control algorithm, an effective balance between energy consumption and environmental thermal comfort control is achieved.

[0079] Example 2

[0080] The difference from Example 1 is that, in Example 2, the influence of the target space occupancy rate on the operating power of the HVAC equipment is considered in the optimization objective function in the operation optimization module 400, and the optimization objective function is:

[0081]

[0082] Among them, F i is the target subfunction to be balanced, γ i is the corresponding balance weight, μ i is the correction coefficient of the objective sub-function to be balanced, and n represents the number of objective sub-functions to be balanced.

[0083] When considering the impact of the target space occupancy rate on the operating power of the HVAC equipment, the optimization objective function is the weighted sum of the two modified objective sub-functions. For example, the total power target sub-function F1 of the HVAC equipment is modified, while the environmental thermal comfort target sub-function F2 of the target space is not modified. At this time, the optimization objective function is: Fitness = γ1×μ1F1+γ2×F2, where the value of μ1 is closely related to the occupancy rate R(t), set to μ1 = δR(t), δ is a constant. Taking the office building as the research object, all the clock-in systems in the office building are connected to the background for statistics, and the number of people entering and leaving the office building every day P(t) is recorded, and time t is used as a variable to form a time series. The occupancy rate R(t) can be calculated by the following formula:

[0084] R(t)=P 当前 / P 最大

[0085] Among them, P 当前 Indicates the number of people in the current room, P 最大 Indicates the maximum number of people that can be accommodated in the room.

[0086] Feedforward control uses real-time data on people entering and exiting the room to predict changes in demand for HVAC equipment based on trends. For example, when the clock-in data detects a large number of people entering, the system immediately increases the processing power of the HVAC equipment to respond to changes in indoor temperature, humidity and air quality in advance, thereby avoiding the lag problem in traditional feedback control. In an air-conditioning system based on feedforward control, the adjustment of the total power of the HVAC equipment can be correlated and optimized with the occupancy rate. Combined with the real-time data on people entering and exiting the room collected by the clock-in system, the feedforward algorithm is used to estimate changes in the number of people in the room, and the power of the air conditioning and fresh air treatment system is quickly adjusted to achieve energy saving and comfort maintenance.

[0087] By adding feedforward control of user occupancy rate, it can respond quickly when the occupancy rate changes, avoiding the lag problem of feedback control. By adjusting the power of HVAC equipment according to the occupancy rate, energy consumption can be significantly reduced. By adjusting the power of the air conditioning system in advance, the temperature, humidity and air quality can be kept stable when people enter and exit. This feedforward control method is highly applicable in large buildings and can further optimize the energy consumption management of HVAC equipment.

[0088] Example 3

[0089] The following is a description of the HVAC equipment energy-saving control method provided by the present invention, which can be referred to in correspondence with the HVAC equipment energy-saving control system described above. Figure 3 As shown, the embodiment of the present invention discloses a method for energy-saving control of HVAC equipment, comprising the following steps:

[0090] Step 1, obtaining environmental observation data of the target space;

[0091] Step 2: Combine the environmental data with the HVAC equipment operation data to form a fused data set, and build a first fused convolutional neural network based on the fused data set to predict the target space flow field information at the future moment, and obtain the steady-state thermal and humid characteristics based on the target space flow field information;

[0092] Step 3, constructing a second fused convolutional neural network based on the fused data set to predict the HVAC equipment operating status data corresponding to the candidate environmental data;

[0093] Step 4: construct an optimization objective function, select the environmental data to be adjusted from the candidate environmental data, and determine the control strategy of the HVAC equipment;

[0094] Step 5, using a weighted fuzzy logic control algorithm to simultaneously control the temperature and humidity in the environmental data of the target space, and stabilize the temperature and humidity of the target space at the preferred environmental data to be adjusted.

[0095] In this embodiment, the above method is applied to a small office building. The system first obtains environmental data such as temperature, humidity, and pressure in various areas of the building through sensors, and combines the historical operating data of the HVAC equipment to construct a fusion data set. Based on this data set, the first fusion convolutional neural network predicts the flow field information of the building in the future period of time, and calculates the steady-state thermal and humidity characteristics of each area. Subsequently, the second fusion convolutional neural network predicts the corresponding operating status of the HVAC equipment based on the candidate environmental data. By optimizing the objective function, the system selects the most energy-saving and comfort-guaranteed environmental data, and determines the control strategy so that the operation of the HVAC equipment can take into account both energy saving and employee comfort. Finally, the weighted fuzzy logic control algorithm is used to accurately adjust the temperature and humidity of the building and stabilize it in the optimal state, ensuring energy saving while improving the comfort of employees.

[0096] By adopting the above method, the dependence on complex building simulation software can be reduced, and more efficient prediction and control can be achieved through the fusion of deep learning convolutional neural networks. There is no need for a large amount of sampling or professional simulation tools, which reduces the system deployment and operation costs. At the same time, it has higher real-time and adaptability, can be widely used in various types of buildings, and provide more efficient and energy-saving control effects.

[0097] It should be noted that the system provided in the above embodiment only uses the division of the above functional modules as an example to implement its functions. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0098] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

Claims

1. A heating and ventilation equipment energy-saving control system, characterized in that: include: A data acquisition module is used to obtain environmental observation data of the target space; The flow field prediction module constructs the first fusion convolutional neural network to predict the target space flow field information at the future moment and obtain the steady-state thermal and humid characteristics based on the target space flow field information; Run the reverse inference module to construct the second fusion convolutional neural network to predict the HVAC equipment operation status data corresponding to the candidate environmental data; Run the optimization module, construct the optimization objective function, select the environmental data to be adjusted from the candidate environmental data, and determine the HVAC equipment control strategy.

2. The HVAC equipment energy-saving control system according to claim 1, characterized in that: Also includes: The operation control module adjusts the operation status of the HVAC equipment in the target space according to the HVAC equipment control strategy determined by the operation optimization module, and controls the environmental data of the target space to the environmental data to be adjusted.

3. The HVAC equipment energy-saving control system according to claim 2, characterized in that: A weighted fuzzy logic control algorithm is adopted in the operation control module to simultaneously control the temperature and humidity in the environmental data of the target space, so as to stabilize the temperature and humidity of the target space at the preferred environmental data to be adjusted.

4. The HVAC equipment energy-saving control system according to claim 1, characterized in that: The data acquisition module includes sensors provided by the HVAC equipment itself, including but not limited to temperature sensors, humidity sensors, pressure sensors and air quality sensors, which are used to obtain the air temperature, humidity, pressure and CO2 concentration in the target space.

5. The HVAC equipment energy-saving control system according to claim 1, characterized in that: The flow field prediction module constructs a first fused convolutional neural network based on the data set, including: Combine environmental data with HVAC equipment operation data to form a fused data set; The time dimension is divided into multiple time steps, and the convolutional neural network is trained based on the fused data set. For each time step, one network is selected as the representative network, and the representative networks of each time step are fused to form the first fused convolutional neural network.

6. The HVAC equipment energy-saving control system according to claim 5, characterized in that: The fused data set is input into the convolutional neural network in the order of time steps, and the flow field of the corresponding time step is output. The output is substituted into the finite volume control equation of the corresponding time step for solution. The network that minimizes the residual of the finite volume control equation in this time step is used as the representative network, and the flow field in the representative network is used as the basis for solving the next time step. The loss function is used to associate and fuse the representative networks of all time steps to form the first fused convolutional neural network.

7. The HVAC equipment energy-saving control system according to claim 5, characterized in that: The second fused convolutional neural network is constructed in the running back-stepping module, including: Combine environmental data with indoor load data to form a new fused data set; Under the state of heat and moisture balance, the HVAC equipment operation data input in the input layer of the first fusion convolutional neural network and the flow field information data output in the output layer are swapped to establish the second fusion convolutional neural network.

8. The HVAC equipment energy-saving control system according to claim 1, characterized in that: The optimization objective function is constructed in the operation optimization module to balance the operating power of the HVAC equipment and the environmental thermal comfort of the target space. The optimization objective function is: Among them, F i is the target subfunction to be balanced, γ i is the corresponding balance weight, μ i is the correction coefficient of the target sub-function to be balanced, and n represents the number of the target sub-function to be balanced; the environmental data to be adjusted that minimizes the optimization target function is selected from the candidate environmental data; Without considering the impact of the target space occupancy rate on the HVAC equipment operating power, the total power target sub-function F1 of the HVAC equipment is: Among them, P total is the total power of HVAC equipment, P min and P max It is the minimum and maximum value of total power of HVAC equipment; The value range of the environmental thermal comfort target sub-function F2 of the target space is [0,0.25].

9. The HVAC equipment energy-saving control system according to claim 8, characterized in that: A balance weight optimization model is constructed based on deep learning. The balance weight is obtained according to the target space state using the balance weight optimization model to calculate the optimization objective function.

10. A method for controlling energy saving of HVAC equipment, characterized in that: Applied to any one of the HVAC equipment energy-saving control systems as described in claims 1-9, the method comprises the following steps: Step 1, obtaining environmental observation data of the target space; Step 2: Combine the environmental data with the HVAC equipment operation data to form a fused data set, and build a first fused convolutional neural network based on the fused data set to predict the target space flow field information at the future moment, and obtain the steady-state thermal and humid characteristics based on the target space flow field information; Step 3, constructing a second fused convolutional neural network based on the fused data set to predict the HVAC equipment operating status data corresponding to the candidate environmental data; Step 4: construct an optimization objective function, select the environmental data to be adjusted from the candidate environmental data, and determine the control strategy of the HVAC equipment; Step 5, using a weighted fuzzy logic control algorithm to simultaneously control the temperature and humidity in the environmental data of the target space, and stabilize the temperature and humidity of the target space at the preferred environmental data to be adjusted.

Images