CN117433078A

CN117433078A - Heating, ventilation and air conditioning system and method for controlling the same

Info

Publication number: CN117433078A
Application number: CN202210864373.8A
Authority: CN
Inventors: F·科拉; B·坦耶尔德兹; J·普拉迪普塔
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2022-07-21
Filing date: 2022-07-21
Publication date: 2024-01-23

Abstract

Various aspects relate to a method of controlling a heating, ventilation and air conditioning, HVAC, system, the method comprising: controlling indoor environmental conditions using the HVAC system, detecting a load of the HVAC system, inputting the detected indoor environmental conditions and the detected outdoor environmental conditions into a load prediction model to generate a predicted load, and training the load prediction model to reduce a difference between the predicted load and the detected load of the HVAC system; determining a required indoor environmental condition associated with a future time period; determining predicted outdoor environmental conditions for a future time period using weather forecast; inputting the desired indoor environmental conditions and the predicted outdoor environmental conditions into a trained load prediction model to determine a predicted load for a future time period; the HVAC system is controlled using the determined predicted load to reduce the load of the HVAC system over a future period of time.

Description

Heating, ventilation and air conditioning system and method for controlling the same

Technical Field

Aspects of the present disclosure relate to a heating, ventilation, and air conditioning (HVAC) system and a method of controlling the HVAC system.

Background

Heating, ventilation, and air conditioning (HVAC) systems can be used to control an indoor environment (e.g., in a building) to provide thermal comfort and/or a desired indoor air quality. However, to achieve a desired indoor environmental condition (e.g., temperature, e.g., humidity, etc.), the HVAC system may consume a particular load (e.g., a particular power consumption) that may depend on various parameters. The load (e.g., future power consumption) of the HVAC system over a future period of time is predicted to adjust the control of the HVAC system such that the load actually used over the period of time is reduced as compared to the predicted load. Various aspects relate to an HVAC system and a method of controlling the HVAC system that are capable of predicting a load for a future period of time and controlling the HVAC system to reduce an actual load over the future period of time as compared to the predicted load. For example, the heating/cooling rate of the HVAC system may be adapted to the actual required load that is reduced over a future period of time. This may reduce energy consumption of the HVAC system, which may also reduce costs and increase environmental sustainability. Illustratively, an energy efficient HVAC system and a method of energy efficient control of an HVAC system are provided.

Disclosure of Invention

Various embodiments relate to a method of controlling a heating, ventilation and air conditioning, HVAC, system, the method comprising: training a load prediction model, the training comprising: controlling indoor environmental conditions using the HVAC system, detecting indoor environmental conditions, load of the HVAC system, outdoor environmental conditions, inputting the detected indoor environmental conditions and the detected outdoor environmental conditions into a load prediction model to generate a predicted load, determining a loss value by comparing the predicted load with the detected load of the HVAC system, and training the load prediction model to reduce the loss value; determining a required indoor environmental condition associated with a future time period; determining predicted outdoor environmental conditions for a future time period using weather forecast; inputting the desired indoor environmental conditions and the predicted outdoor environmental conditions into a trained load prediction model to determine a predicted load for a future time period; and controlling the HVAC system using the determined predicted load to reduce the load of the HVAC system over a future period of time.

According to various embodiments, the HVAC system may include or may be a variable refrigerant flow system.

According to various embodiments, the predicted load may represent an amount of energy required by the HVAC system to achieve a desired indoor environmental condition over a future period of time. According to various embodiments, the indoor environmental conditions may include indoor temperature. According to various embodiments, the indoor environmental conditions may include indoor humidity.

According to various embodiments, the outdoor environmental condition may include an outdoor temperature. According to various embodiments, the outdoor environmental conditions may include solar surface radiation. According to various embodiments, the outdoor environmental condition may include outdoor humidity.

According to various embodiments, the method may further comprise: detecting occupancy of an indoor area with indoor environmental conditions controlled by an HVAC system; determining a predicted occupancy within a future time period using calendar information and/or occupancy statistics representing occupancy of the indoor area; inputting the detected indoor environmental condition and the detected outdoor environmental condition into a load prediction model to generate a predicted load may include inputting the detected indoor environmental condition, the detected outdoor environmental condition, and the detected occupancy into the load prediction model to generate a predicted load; inputting the desired indoor environmental condition and the predicted outdoor environmental condition into the trained load prediction model to determine the predicted load for the future time period may include inputting the desired indoor environmental condition, the predicted outdoor environmental condition, and the predicted occupancy into the trained load prediction model to determine the predicted load for the future time period.

According to various embodiments, inputting the detected indoor environmental condition and the detected outdoor environmental condition into the load prediction model to generate the predicted load may include inputting the detected indoor environmental condition, the detected outdoor environmental condition, and a time of day at which the indoor environmental condition, the HVAC system load, the outdoor environmental condition are detected into the load prediction model to generate the predicted load; inputting the desired indoor environmental condition and the predicted outdoor environmental condition into the trained load prediction model to determine the predicted load for the future time period may include inputting the desired indoor environmental condition, the predicted outdoor environmental condition, and the time of day associated with the future time period into the trained load prediction model to determine the predicted load for the future time period.

According to various embodiments, the method may further comprise: detecting a load of the HVAC system over a future period of time; determining an additional loss value by comparing the predicted load for the future time period to the load of the HVAC system detected for the future time period; and further trains the load prediction model to reduce additional loss values.

Various embodiments relate to a heating, ventilation, and air conditioning HVAC system including one or more computers configured to: implementing a load prediction model trained according to the above embodiments; receiving a required indoor environmental condition describing a predicted indoor environmental condition for a future time period; receiving a weather forecast for a future time period, the weather forecast describing predicted outdoor environmental conditions for the future time period; determining a predicted load for a future time period using the trained load prediction model; and controlling the HVAC system using the determined predicted load to reduce the load of the HVAC system over a future period of time.

Drawings

The invention will be better understood with reference to the detailed description when considered in conjunction with the non-limiting examples and the accompanying drawings, in which:

FIGS. 1 and 5 each illustrate a processing system for controlling an HVAC system according to various embodiments;

FIGS. 2 and 4 each illustrate a processing system training a load prediction model for controlling an HVAC system according to various embodiments;

FIG. 3A illustrates a method of controlling an HVAC system according to various embodiments;

FIG. 3B illustrates a method of training a load prediction model for controlling an HVAC system according to various embodiments; and

FIG. 6 illustrates an exemplary HVAC system according to various embodiments.

Detailed Description

The following detailed description refers to the accompanying drawings that show, by way of illustration, specific details and embodiments in which the disclosure may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the disclosure. Other embodiments and structures may be utilized and logical changes may be made without departing from the scope of the present disclosure. The various embodiments are not necessarily mutually exclusive, as some embodiments may be combined with one or more other embodiments to form new embodiments.

The embodiments described in the context of one of the methods are similarly valid for the other method. Similarly, embodiments described in the context of HVAC systems are similarly effective for methods, and vice versa.

Features described in the context of embodiments may be correspondingly applicable to the same or similar features in other embodiments. Features described in the context of embodiments may be correspondingly applicable to other embodiments even if not explicitly described in these other embodiments. Furthermore, additions and/or combinations and/or substitutions described in the context of an embodiment for features may be applied accordingly to the same or similar features in other embodiments.

In the context of various embodiments, the articles "a," "an," and "the" are used with respect to a feature or element to include references to one or more features or elements.

As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.

In one embodiment, a "computer" may be understood as any kind of logic implementing entity, which may be hardware, software, firmware, or any combination thereof. Thus, in one embodiment, a "computer" may be hardwired logic or programmable logic, such as a programmable processor, such as a microprocessor (e.g., a Complex Instruction Set Computer (CISC) processor or a Reduced Instruction Set Computer (RISC) processor). A "computer" may also be software implemented or executed by a processor, e.g. any type of computer program, e.g. a computer program using virtual machine code such as Java. A "computer" may be or include one or more processors. Any other type of implementation of the various functions described in more detail below may also be understood as a "computer" according to alternative embodiments.

The "memory" may be used in processes performed by the computer and/or may store data used by the computer. "memory" as used in embodiments may be volatile memory, such as DRAM (dynamic random Access memory) or nonvolatile memory, such as PROM (programmable read Only memory), EPROM (erasable PROM), EEPROM (electrically erasable PROM), or flash memory, such as floating gate memory, charge trapping memory, MRAM (magnetoresistive random Access memory), or PCRAM (phase Change random Access memory).

As used herein, a "load prediction model" may be any type of model that is capable of predicting a load in response to input of one or more parameters and/or information as described herein. Illustratively, a "load prediction model" may map input parameters and/or information according to the description herein to a predicted load. For example, the "model" may be based on machine learning (e.g., a machine learning algorithm may be employed). Illustratively, machine learning may be used to adjust (e.g., train) the "model. The "model" may be a decision tree model, a random forest model, a gradient lifting model, a support vector machine, a k-nearest neighbor model, a neural network, and the like. The "neural network" may be any type of neural network, such as an automatic encoder network, a convolutional neural network, a variational automatic encoder network, a sparse automatic encoder network, a cyclic neural network, a deconvolution neural network, a generation countermeasure network, a prospective neural network, a product neural network, and the like. The "neural network" may include any number of layers. The neural network may be trained by any training principle, such as back propagation.

As used herein, the "load" of an HVAC system may represent the amount of energy required to achieve a relevant indoor environmental condition (e.g., indoor temperature, e.g., indoor humidity). Illustratively, the "load" of the HVAC system may be the power consumption of the HVAC system. The "load" of the HVAC system may be the amount of energy required to maintain the condition of the relevant area within the desired/required conditions. The "load" of the HVAC system may be a cold load and/or a hot load.

While the present disclosure has been particularly shown and described with reference to particular embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. The scope of the invention is therefore indicated by the appended claims, and all changes that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

Various aspects relate to methods of predicting load of an HVAC system over a future time period and controlling the HVAC system (e.g., adjusting a currently set parameter) such that the HVAC system consumes less energy during the time period. For example, on a particular day, the load of the HVAC system on the next day may be predicted, and the control of the HVAC system adjusted such that the load actually consumed on the next day (i.e., the day) is reduced as compared to the predicted load. Illustratively, future loads are predicted and the settings are changed such that the load for future consumption is reduced.

FIG. 1 illustrates a processing system 100 for controlling an HVAC system according to various embodiments. The HVAC system may be or include a Variable Refrigerant Flow (VRF) system. The HVAC system may be or include a VRF system and/or a chiller system. The HVAC system may include one or more HVAC devices. The one or more HVAC devices may be configured to control (e.g., remain stable, e.g., change) one or more environmental parameters surrounding the one or more HVAC devices in accordance with a set operating parameter (e.g., a set HVAC temperature). As used herein, an operating parameter of an HVAC device may be any kind of parameter related to changing environmental conditions, such as fan speed, valve opening (e.g., a valve for controlling flow of cooling liquid, such as throughput of cooling water). One of the one or more environmental parameters may be, for example, temperature, humidity, or dew point. HVAC systems may be associated with indoor environments (e.g., environments within a building) and outdoor environments (e.g., environments outside of a building, such as environments of an exterior wall of a building). The indoor environment may include a plurality of areas (e.g., zones). For example, one of the plurality of zones associated with the HVAC system may be a room in an indoor environment. Each of the plurality of zones may be associated with at least one HVAC device of the one or more devices. At least one HVAC device associated with the zone may be configured to control (e.g., remain stable, e.g., change) one or more environmental parameters within the zone.

Processing system 100 may include a computer 110. The computer 110 may be configured to control an HVAC system. For example, the computer 110 may be configured to control one or more HVAC devices (e.g., by setting a respective HVAC temperature associated with each of the one or more HVAC devices). The computer 110 includes one or more processors. As described above, computer 110 may be any kind of logic implementing entity. The processing system 100 may include a memory 102. Memory 102 may be used in the processing performed by computer 110. The memory 102 may be part of an HVAC system. The memory 102 may be external to the HVAC system and may be, for example, cloud memory. The memory 102 may include a plurality of memory devices and one or more of the plurality of memory devices may be part of an HVAC system and other memory devices of the plurality of memory devices may be part of a cloud memory. The data stored in memory 102 may be stored in local memory and/or cloud memory. The memory 102 may store a desired indoor environmental condition 104 (e.g., data representing a desired indoor environmental condition) associated with a future time period. The future time period may be any time period that is future from the current time. The future time period may begin at the current time and may end at the future time (i.e., a future point in time at the current time). For example, the future time period may be a time period from the current time to a point in time that is one or more hours (e.g., 1 hour, e.g., 2 hours, etc.), one or more days (e.g., 1 day, e.g., 2 days), or the like, later than now. The future time period may begin at a future time and may end at a point in time that is later than the future time. For example, the future time period may be the next day. For example, the future time period may begin at a first point in time on the second day and may end at a second point in time on the second day, the second point in time being after the first point in time. As used herein, an "indoor environmental condition" may describe one or more environmental parameters, such as indoor temperature (e.g., room temperature) and/or indoor humidity. For example, the desired indoor environmental conditions may include a desired indoor temperature over a future time period. As used herein, "indoor environmental conditions" may refer to an indoor environment associated with an HVAC system. The predefined setpoint schedule may include predefined indoor environmental conditions (e.g., indoor temperature and/or indoor humidity) as a function of time of day and/or day of the week. The predefined setpoint schedule may be stored in the memory 102. The computer 110 may be configured to obtain a predefined setpoint schedule and determine a predefined indoor environmental condition over a future time period as the desired indoor environmental condition 104. The HVAC system may include a user interface. The user may be able to set indoor environmental conditions via the user interface. The set indoor environmental conditions may be stored in the memory 102. The computer 110 may be configured to obtain a set indoor environmental condition and determine the set indoor environmental condition over a future time period as the desired indoor environmental condition 104. The memory 102 may store predicted outdoor environmental conditions 106 over a future period of time. As used herein, an "outdoor environmental condition" may describe one or more environmental parameters, such as outdoor (e.g., air) temperature, outdoor humidity, and/or solar surface radiation. As used herein, "outdoor environmental conditions" may refer to the outdoor environment associated with an HVAC system. The solar surface radiation may describe a cloud level. The computer 110 may be configured to determine cloud deck height using solar surface radiation. Weather forecast may be used to determine predicted outdoor environmental conditions 106. For example, weather forecast may provide future outdoor environmental conditions. The computer 110 may be configured to obtain (e.g., download) the weather forecast for the future time period from a weather forecast service (e.g., from cloud storage of the weather forecast service). The predicted outdoor environmental condition 106 may affect the (cooling/heating) load required to achieve the desired indoor environmental condition 104. As an example, relatively high solar surface radiation with an outside temperature at least 10 ℃ higher than the indoor temperature and/or indicative of a cloudless condition may increase the cooling load of the HVAC system required to reduce the indoor temperature.

The processing system 100 may be configured to implement a trained load prediction model 112. The trained load prediction model 112 may be configured to generate (e.g., output) a predicted load in response to input indoor and outdoor environmental conditions. The trained load prediction model 112 may be obtained by training the load prediction model shown in fig. 2. The trained load prediction model 112 may determine the predicted load 114 for the future time period in response to inputting the desired indoor environmental condition 104 and the predicted outdoor environmental condition 106 associated with the future time period.

The computer 110 may be configured to provide control instructions 116 to control the HVAC system using the determined predicted load 114 to reduce the load of the HVAC system over a future period of time. For example, in the case of a relatively high predicted load 114, the control instructions 116 may include instructions to begin cooling or heating earlier than a previously set schedule in order to reduce the slope of the cooling ramp or heating ramp. The lower heating or cooling ramp may reduce the amount of energy required to achieve the desired indoor environmental conditions 104 over a future period of time.

FIG. 2 illustrates a processing system 200 for training a load prediction model 222 for controlling an HVAC system 202. The HVAC system 202 can include an indoor environment 208 (e.g., a building, a room in a building, and/or an area in a building). The HVAC system 202 may include one or more HVAC devices 204 within an indoor environment 208. The one or more HVAC devices 204 may be configured to control indoor environmental conditions within the indoor environment 208. The HVAC system 202 may include one or more environmental sensors 206 within an indoor environment 208. The one or more indoor environment sensors 206 may be configured to detect indoor environmental conditions within the indoor environment 208. The HVAC system 202 may include one or more outdoor environment sensors 210 located outside of the indoor environment 208. The one or more outdoor environment sensors 210 may be configured to detect outdoor environmental conditions outside of the indoor environment 208. Processing system 200 may include a computer 212 (e.g., similar in configuration to computer 110). The computer 212 may be configured to control the HVAC system 202 (e.g., one or more HVAC devices 204) to control (e.g., maintain, e.g., change) indoor environmental conditions within the indoor environment 208. The computer 212 may be configured to obtain (e.g., data representative of) a load of the HVAC system 202 (e.g., a load of one or more HVAC devices 204). The computer 212 may be configured to acquire (e.g., data representative of) the indoor environmental conditions 216 detected by the one or more indoor environmental sensors 206. The computer 212 may be configured to obtain (e.g., data representative of) outdoor environmental conditions 220 detected by one or more outdoor environmental sensors 210. The computer 212 may be configured to implement a load prediction model 222. The load prediction model 222 may be configured to generate a predicted load in response to input indoor environmental conditions and outdoor environmental conditions. The load prediction model 222 may generate the predicted load 224 in response to inputting the acquired indoor environmental conditions 216 and the acquired outdoor environmental conditions 220 into the load prediction model 222. The computer 212 may be configured to determine the loss value 226 by comparing the predicted load 224 to the acquired load 214. The computer 212 may be configured to train the load prediction model 222 to reduce the loss value 226. Training of the load prediction model 222 as described herein may be one iteration of the training process, and training may be performed as multiple iterations.

FIG. 3A illustrates a flowchart of a method 300 of controlling an HVAC system according to various embodiments. The method 300 may include providing a trained load prediction model capable of generating a predicted load (in 302) in response to inputting the indoor environmental condition and the outdoor environmental condition into the load prediction model. The method 300 may include determining a required indoor environmental condition associated with a future time period (in 304). The method 300 may include determining (in 306) a predicted outdoor environmental condition over a future time period using weather forecast (e.g., by a weather forecast service). The method 300 may include inputting the desired indoor environmental condition and the predicted outdoor environmental condition into a trained load prediction model to determine (e.g., generate) a predicted load for a future time period (in 308). The method 300 may include controlling the HVAC system using the determined predicted load to reduce the load of the HVAC system over a future period of time. The trained load prediction model may be provided by training the load prediction model (in 302). FIG. 3B illustrates a flowchart of a method of training a load prediction model for controlling an HVAC system according to various embodiments. The training method may include controlling indoor environmental conditions using the HVAC system (in 302A). The training method may include detecting an indoor environmental condition, a load of the HVAC system, and an outdoor environmental condition (in 302B). The training method may include inputting the detected indoor environmental condition and the detected outdoor environmental condition into a load prediction model to generate a predicted load (in 302C). The training method may include determining a loss value (in 302D) by comparing the predicted load to the detected load of the HVAC system. The training method may include training a load prediction model to reduce loss values (in 302E). The training method may include one or more iterations (e.g., multiple iterations), and each iteration may include the training method described above.

FIG. 4 illustrates a processing system 400 for training the load prediction model 222 for controlling the HVAC system 202. In addition to the processing system 200, the computer 212 may be configured to obtain occupancy information 404 describing occupancy of personnel within the indoor environment 208. For example, the HVAC system 202 may also include one or more occupancy sensors 402 (e.g., wireless occupancy sensors, e.g., passive infrared sensors, e.g., motion sensors, etc.) configured to detect occupancy within the indoor environment 208 (e.g., infrared-based, e.g., ultrasonic-based, e.g., radar-based, e.g., microwave-based, etc.). The one or more occupancy sensors 402 may be configured to provide the detected occupancy as occupancy information 404 to the computer 212. According to various aspects, the memory may store personal calendar information, meeting information, and the like. The computer 212 may be configured to use personal calendar information, meeting information, etc. to determine occupancy of the indoor environment 208 in order to obtain occupancy information 404. The computer 212 may be configured to implement an occupancy prediction model (e.g., using occupancy statistics) that describes times when occupants of the indoor environment 208 are commuted, and/or daily habits of people within the indoor environment 208 to predict occupancy within the indoor environment 208. The computer 212 may be configured to determine a predicted occupancy within the indoor environment 208 as occupancy information 404 using the occupancy prediction model. The computer 212 may be configured to obtain the time 410 (e.g., using the computer's internal clock and/or an external server providing clock time) at which the load 214, the indoor environmental condition 216, the outdoor environmental condition 220 of the HVAC system 202 are detected during the day. Optionally, the computer 212 may be configured to obtain user comfort 408 for each user within the indoor environment 208. The user comfort level may represent thermal comfort level of the user. User comfort 408 may be user feedback provided via user device 406. The computer 212 may be configured to implement a comfort model capable of predicting a user comfort for each user using indoor environmental conditions. The user comfort 408 may be a predicted average vote representing an average of all user comfort. The load prediction model 222 may be configured to generate the predicted load 224 in response to inputting the acquired indoor environmental conditions 216, the acquired outdoor environmental conditions 220, the acquired occupancy information 216, the acquired time of day 410, and optionally the acquired user comfort 408 into the load prediction model 222. As described with reference to processing system 200, computer 212 may be configured to determine loss value 226 by comparing predicted load 224 to acquired load 214, and may be configured to train load prediction model 222 to reduce loss value 226 (e.g., using multiple iterations).

As described herein, computer-acquired data (e.g., corresponding conditions) may refer to data acquired directly from a sensor or device or indirectly from a memory storing the data. For example, one or more acquisition modules may be configured to acquire data and store the data in memory. The data may be stored in a database within the memory. Note that a load prediction model as described herein may include a plurality of individual models, and each individual model may be configured to provide a respective predicted load in response to input of one or more data/information as described herein. In this case, the predicted load described herein may be the sum of all the respective predicted loads determined by the plurality of individual models.

FIG. 5 illustrates a processing system 500 for controlling an HVAC system according to various embodiments. Processing system 500 may be similar to processing system 100 in that computer 110 is configured to implement a trained load prediction model 512. The trained load prediction model 512 may be configured to generate (e.g., output) a predicted load in response to input indoor environmental conditions, outdoor environmental conditions, time of day 502, and predicted occupancy 504 (and optionally user comfort). The computer 110 may be configured to determine the predicted occupancy 504 using personal calendar information, meeting information, etc., stored in memory, using occupancy statistics, and/or using an occupancy prediction model capable of predicting occupancy within the indoor environment. The trained load prediction model 512 may be obtained by training a load prediction model as shown with reference to fig. 4. The time of day 502 may be determined using the time of the future time period.

Depending on the processing system 100 and/or the processing system 500, the computer 110 may be configured to detect a load of the HVAC system over a future period of time. The computer 110 may be configured to determine additional loss values by comparing the predicted load for the future time period to the detected load of the HVAC system over the future time period. The computer 110 may be configured to adjust (e.g., train) a load prediction model (e.g., load prediction model 112, e.g., load prediction model 512) to reduce additional loss values. Training may be performed as described at processing system 200 and processing system 400. Illustratively, the trained load prediction model may be further trained. For example, the trained load prediction model may be further trained at constant time intervals (e.g., 1 day, e.g., 1 week, e.g., 1 month, etc.).

Fig. 6 illustrates an exemplary HVAC system 600 according to various embodiments. HVAC system 600 may be configured to perform method 300. The HVAC system 600 can include a control 604. The control 604 may include an internal memory and/or may be in communication with an external memory (e.g., a cloud). The memory may store a trained load prediction model (e.g., the trained load prediction model 112, e.g., the trained load prediction model 512). The memory may also store a desired indoor environmental condition for a future time period and a weather forecast for the future time period describing a predicted outdoor environmental condition for the future time period. Alternatively, the memory may store data representing a predicted occupancy or occupancy model to determine the predicted occupancy. Alternatively, the memory may store data representative of thermal comfort of at least one occupant of an indoor environment controlled by the HVAC system. The control 604 may include an internal computer or may be in communication (e.g., using cloud computing) with an external computer configured to implement a trained load prediction model to determine a predicted load (indoor environmental conditions and predicted outdoor environmental conditions in response to input requirements, and optionally time of day, data representing predicted occupancy, and/or data representing thermal comfort of at least one occupant). The control device 604 may be configured to control the HVAC system 600 (e.g., the plurality of HVAC devices 602 (1), 602 (2), 602 (3), 602 (4), 602 (5)) using the determined predicted load to reduce the load of the HVAC system 600 over a future period of time. According to various aspects, the indoor environment may include a plurality of areas (e.g., rooms in a building) 606 (1), 606 (2), 606 (3). The control device 604 may be configured to determine a respective predicted load for each of the plurality of zones 606 (1), 606 (2), 606 (3) using a trained load prediction model and may be configured to reduce the load over a future period of time using the separately determined predicted loads to control HVAC devices associated with the respective zones. As an example, the control device 604 may determine the predicted load of the region 606 (1) using a load prediction model, and may control the first HVAC device 602 (1) and the second HVAC device 606 (2) to reduce the load of the first HVAC device 602 (1) and the second HVAC device 606 (2) over a future period of time.

Claims

1. A method (300) of controlling a heating, ventilation, and air conditioning (HVAC) system, the method (300) comprising:

training a load prediction model (302), the training comprising:

controlling indoor environmental conditions using the HVAC system (302A),

detecting indoor environmental conditions, load of HVAC system and outdoor environmental conditions (302B),

inputting the detected indoor environmental condition and the detected outdoor environmental condition into the load prediction model to generate a predicted load (302C),

determining a loss value (302D) by comparing the predicted load to a detected load of the HVAC system, an

Training the load prediction model to reduce loss values (302E);

determining a required indoor environmental condition associated with a future time period (304);

determining (306) predicted outdoor environmental conditions over a future time period using weather forecast;

inputting the required indoor environmental conditions and the predicted outdoor environmental conditions into a trained load prediction model to determine a predicted load for a future time period (308); and

the HVAC system is controlled using the determined predicted load to reduce the load of the HVAC system over a future period of time (310).

2. The method (300) of claim 1, wherein,

the HVAC system includes a variable refrigerant flow system.

3. The method (300) according to claim 1 or 2, wherein,

the predicted load represents an amount of energy required by the HVAC system to achieve a desired indoor environmental condition over a future period of time.

4. The method (300) according to any one of claims 1 to 3, wherein,

indoor environmental conditions include indoor temperature; and/or

Indoor environmental conditions include indoor humidity.

5. The method (300) according to any one of claims 1 to 4, wherein,

outdoor environmental conditions include outdoor temperature.

6. The method (300) of claim 5, wherein,

outdoor environmental conditions also include solar surface radiation and/or outdoor humidity.

7. The method (300) according to any one of claims 1 to 6, wherein the method further comprises:

detecting occupancy of an indoor area with indoor environmental conditions controlled by an HVAC system; and

determining a predicted occupancy within a future time period using calendar information and/or occupancy statistics representing occupancy of the indoor area;

wherein inputting the detected indoor environmental condition and the detected outdoor environmental condition into the load prediction model to generate a predicted load (302C) comprises:

inputting the detected indoor environmental conditions, the detected outdoor environmental conditions, and the detected occupancy into the load prediction model to generate a predicted load;

inputting the desired indoor environmental condition and the predicted outdoor environmental condition into a trained load prediction model to determine a predicted load for a future time period (308) includes:

inputting the required indoor environmental conditions, the predicted outdoor environmental conditions, and the predicted occupancy into a trained load prediction model to determine a predicted load for a future time period.

8. The method (300) according to any one of claims 1 to 7, wherein,

inputting the detected indoor environmental condition and the detected outdoor environmental condition into the load prediction model to generate a predicted load (302C) includes:

inputting the detected indoor environmental condition, the detected outdoor environmental condition, and the time of day that the indoor environmental condition, the load of the HVAC system, and the outdoor environmental condition are detected into the load prediction model to generate a predicted load;

inputting the required indoor environmental conditions, the predicted outdoor environmental conditions, and the time of day associated with the future time period into a trained load prediction model to determine a predicted load for the future time period.

9. The method (300) according to any one of claims 1 to 8, wherein the method further comprises:

detecting a load of the HVAC system over a future period of time;

determining an additional loss value by comparing the predicted load for the future time period to the load of the HVAC system detected for the future time period; and

further training the load prediction model to reduce further loss values.

10. A heating, ventilation, and air conditioning (HVAC) system comprising one or more computers (110) configured to:

-implementing a load prediction model (112) trained according to any one of claims 1 to 9;

-receiving a required indoor environmental condition (104), the required indoor environmental condition (104) describing a predicted indoor environmental condition for a future time period;

-receiving a weather forecast (106) for a future time period, the weather forecast describing predicted outdoor environmental conditions over the future time period;

-determining a predicted load (114) for a future time period using a trained load prediction model (112); and

controlling the HVAC system using the determined predicted load (114) to reduce the load of the HVAC system over a future period of time.