KR102338086B1 - Machine learning-based air conditioner and control method thereof - Google Patents

Abstract

The present invention provides a machine learning-based air conditioner and a method for controlling the same. According to the present invention, the method may comprise the steps of: collecting first data on air quality of air introduced into the air conditioner; controlling a purification unit based on the first data; collecting second data on air quality of the air that has passed through the purification unit; and controlling the purification unit according to a machine learning model generated based on a set value related to the control of the purification unit, the first data, and the second data.

Description

MACHINE LEARNING-BASED AIR CONDITIONER AND CONTROL METHOD THEREOF

The present invention relates to a machine learning-based air conditioner and a method for controlling the same. Specifically, the present invention measures the air quality before and after passing through a purifying unit that purifies air, and machine learning using a set value related to the control of the purifying unit to obtain the optimum air purifying efficiency according to the space where the air conditioner is installed. The present invention relates to an air conditioner and a method for controlling the same.

In general, air conditioners in large office buildings purify air by filtering and circulating pollutants floating in the air using a HEPA filter. However, when a conventional air conditioner using a HEPA filter is used for a long time, pores between the filters are clogged by fine dust, resulting in differential pressure and vibration, and the ability to filter fine dust is also reduced.

To improve this, an electrostatic precipitator for removing contaminants without differential pressure was considered. An electric dust collector is a device that is installed in an air conditioner such as an air purifier, air conditioner, or heater to collect dust by charging foreign substances such as dust contained in the air. The electrostatic precipitator may include a charging unit including a discharge device for forming an electric field, and a dust collecting unit in which pollutant particles charged by the charging unit are collected, and while the air passes through the charging unit and then passes through the dust collecting unit, air Contaminants in the dust may be collected in the dust collector. To this end, if a high voltage of several thousand volts is applied to the charged part, electrons are generated from the electrode itself or from the gas around the electrode to form plasma around the electrode. When electrons are separated from atoms or molecules in gaseous state by plasma and attached to particles in the air, the particles become (-) charged, and dust particles with (-) charges gain (+) charge by electrostatic attraction. It can be removed by moving to a dust collecting plate that is attached to it.

However, this electrostatic precipitation method may generate oxides such as ozone or nitrogen oxides during the air purification process. Since these oxides are highly reactive, they exhibit a sterilizing effect that promotes the decomposition of harmful substances in the air, but can increase the ozone concentration indoors, which can be dangerous to the human body. For example, when a person is exposed to ozone, it can cause eye irritation or damage to the bronchi. In addition, since this electrostatic precipitation method requires a high voltage, it requires high facility and operating costs, and there is a risk of explosion in an environment with high dust density.

In addition, the conventional electrostatic precipitation method used for air conditioning generates ions having a (-) charge and charges contaminants in the air that bind thereto with (-) charges, so contaminants in the dust collecting plate having a (+) charge Although it is possible to collect particles of (-), it is difficult to collect particles from a dust collecting plate with a negative charge. This lowers the dust collection efficiency, and if the negatively charged material is recovered in the air without being collected by the dust collecting plate, the (-) ions in the air become excessive, breaking the ion balance and generating more oxides. have.

In addition, in the conventional electrostatic precipitation method, unnecessary power is wasted because the air conditioner operates on the same basis regardless of the characteristics of the space in which the air conditioner is installed. However, even if an air conditioner of the same product is installed in the same building, air characteristics such as temperature, humidity, amount of fine dust, carbon dioxide, volatile organic compounds (VOCs) and formaldehyde may differ depending on the area within the building. There is a need to optimize the conditions for the operation of the air conditioner. Alternatively, for example, even if an air conditioner of the same product is installed in an underground station, the quality of the air in the platform may be different for each time period, so it is necessary to operate the device to purify the air with minimum power consumption.

An embodiment of the present specification has been proposed to solve the above-described problems, and an object of the present specification is to provide an air conditioner that does not generate ozone, which is a harmful substance, as a by-product that ionizes pollutants in the air, and a method for controlling the same.

In addition, the problem to be solved by the present embodiment is to derive an operating condition of a purification unit optimized according to a space in which the air conditioner is installed, and accordingly a machine learning-based air conditioner for controlling the purification unit of the air conditioner, and a control method thereof is to provide

The technical problems to be achieved by the present embodiment are not limited to the technical problems described above, and other technical problems may be inferred from the following embodiments.

In order to achieve the above object, according to an embodiment of the present specification, there is provided a method for controlling an air conditioner based on machine learning, comprising: collecting first data on air quality of air introduced into the air conditioner; controlling a purifier based on the first data; collecting second data on air quality of the air that has passed through the purification unit; and controlling the purification unit according to a set value related to the control of the purification unit, and a machine learning model generated based on the first data and the second data.

A machine learning-based air conditioner according to an embodiment of the present specification includes: a purification unit; a first data collection unit including at least one sensor and acquiring information related to air flowing into the purification unit; a second data collection unit including at least one sensor and acquiring information related to air flowing out from the purification unit; and a control unit, wherein the control unit collects first data on air quality obtained through the first data collection unit, controls the purification unit based on the first data, and through the second data collection unit Collects second data on the air quality of the air that has passed through the purification unit, and controls the purification unit according to a machine learning model generated based on a set value related to the control of the purification unit, the first data, and the second data can do.

According to the embodiment of the present specification, since pollutants in the air are ionized by using electromagnetic waves of low energy wavelength, there is an effect that not only a differential pressure is applied before and after passing through the air conditioner, but also almost no ozone is generated.

In addition, according to an embodiment of the present specification, it is possible to charge pollutants in the air with a (-) charge as well as a (+) charge, so that the charging unit in the air conditioner of the present invention is charged with a (+) charge. The dust collection efficiency is high as it can collect contaminant particles from both the dust collecting plate that has been charged with negative (-). In addition, even if some of the charged contaminants pass through the dust collector without being collected, it is easy to maintain the balance of ions in the air because the (-) charged material and the (+) charged material are balanced.

In addition, according to the embodiment of the present specification, since the voltage applied to the purification unit is set by machine learning according to the space in which the device is installed, a space-customized design is possible, and the power used by the device can be minimized by machine learning. Contaminant removal efficiency per usage can be increased.

Effects of the invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is a diagram illustrating an air purification operation of a purification unit of an air conditioner according to an exemplary embodiment.
2 is a diagram illustrating an air conditioner according to an exemplary embodiment.
3 is a diagram illustrating the purification unit 220 according to an embodiment.
4 is a flowchart illustrating a method of controlling an air conditioner according to an exemplary embodiment.
5 is a block diagram schematically illustrating an air conditioner 500 according to an exemplary embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

In describing the embodiments, descriptions of technical contents that are well known in the technical field to which the present invention pertains and are not directly related to the present invention will be omitted. This is to more clearly convey the gist of the present invention without obscuring the gist of the present invention by omitting unnecessary description.

For the same reason, some components are exaggerated, omitted, or schematically illustrated in the accompanying drawings. In addition, the size of each component does not fully reflect the actual size. In each figure, the same or corresponding elements are assigned the same reference numerals.

Advantages and features of the present invention, and a method for achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the technical field to which the present invention belongs It is provided to fully inform the possessor of the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

At this time, it will be understood that each block of the flowchart diagrams and combinations of the flowchart diagrams may be performed by computer program instructions. These computer program instructions may be embodied in a processor of a general purpose computer, special purpose computer, or other programmable data processing equipment, such that the instructions performed by the processor of the computer or other programmable data processing equipment are not described in the flowchart block(s). It creates a means to perform functions. These computer program instructions may also be stored in a computer-usable or computer-readable memory which may direct a computer or other programmable data processing equipment to implement a function in a particular manner, and thus the computer-usable or computer-readable memory. It is also possible that the instructions stored in the flow chart block(s) produce an article of manufacture containing instruction means for performing the function described in the flowchart block(s). The computer program instructions may also be mounted on a computer or other programmable data processing equipment, such that a series of operational steps are performed on the computer or other programmable data processing equipment to create a computer-executed process to create a computer or other programmable data processing equipment. It is also possible that instructions for performing the processing equipment provide steps for performing the functions described in the flowchart block(s).

Additionally, each block may represent a module, segment, or portion of code that includes one or more executable instructions for executing specified logical function(s). It should also be noted that in some alternative implementations it is also possible for the functions recited in blocks to occur out of order. For example, two blocks shown one after another may be performed substantially simultaneously, or the blocks may sometimes be performed in the reverse order according to a corresponding function.

At this time, the term '~ unit' used in this embodiment means software or hardware components such as FPGA or ASIC, and '~ unit' performs certain roles. However, '-part' is not limited to software or hardware. '~' may be configured to reside on an addressable storage medium or may be configured to refresh one or more processors. Accordingly, as an example, '~' indicates components such as software components, object-oriented software components, class components, and task components, and processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. The functions provided in the components and '~ units' may be combined into a smaller number of components and '~ units' or further separated into additional components and '~ units'. In addition, components and '~ units' may be implemented to play one or more CPUs in a device or secure multimedia card.

1 is a diagram illustrating an air purification operation of a purification unit of an air conditioner according to an exemplary embodiment.

The air conditioner of the present invention includes a purifying unit 100, and the purifying unit 100 includes a charging unit 110 that ionizes at least some of the substances contained in the air and a dust collecting unit 120 that collects the ionized substances. may include

Similar to the electrostatic precipitator using the conventional photoionization method, the purification unit 100 of the present invention ionizes particles and molecules of pollutants in the air in the charging unit 110 and then collects the dust in the dust collecting unit 120 . Specifically, a voltage may be applied to the charging unit 110 to ionize the contaminant 111 , and the ionized contaminant 112 may be collected by the dust collecting plates 121 and 122 of the dust collecting unit 120 . According to an embodiment, since the charging unit 110 of the present invention uses electromagnetic waves of a low energy wavelength, ozone is hardly generated. In addition, the charging unit 110 of the present invention is capable of charging pollutants in the air with a (-) charge as well as a (+) charge using electromagnetic waves of a low energy wavelength, so It is possible to collect contaminant particles from both the dust collecting plate and the negatively charged dust collecting plate. Accordingly, the contaminant 112 having a (-) charge may move to the dust collecting plate 121 charged with (+) by electrostatic attraction and be collected, and the contaminant 112 having a (+) charge may be collected. Silver may be collected by moving to the dust collecting plate 122 charged with (-). On the other hand, since the clean gas 113 such as oxygen, which is not a pollutant in the air, is not ionized by the charging unit 110 and does not have a specific electric charge, it passes without being collected by the dust collecting unit 120 . After all, when the purification unit 100 of the present invention is used, it is possible to ionize and collect only contaminants. Examples of contaminants may include particulate matter, volatile organic compounds (VOCs) and formaldehyde.

On the other hand, since a high voltage of 8 to 15 kV is applied to the charged part using the conventional electrostatic precipitation method, plasma is formed around the electrode, which ionizes oxygen rather than contaminants, and ozone may be generated as a by-product. On the other hand, since the charged part 110 of the present invention uses electromagnetic waves having a wavelength in a relatively low energy range, it is possible to apply a voltage of 5 kV or less, and thus plasma is not generated, and ozone, which is a byproduct of this, is not generated. This does not occur, or even if it does occur, it occurs only in very small amounts. The charged unit 110 according to an embodiment of the present invention may ionize a material using electromagnetic waves having a wavelength in the range of 0.1 nm or more and 100 nm or less, and specifically, extreme ultraviolet (EUV) radiation having a wavelength of 13.5 nm. Alternatively, soft X-rays having a wavelength of 0.1 nm or more and 10 nm or less may be used. Since differential pressure and ozone are not generated when electromagnetic waves having a wavelength in a low energy range, such as extreme ultraviolet or soft X-rays, are not generated, the purification unit 100 of the present invention can be installed in an existing air conditioner, and a high bypass (by-by-) pass) performance. For example, extreme ultraviolet rays have 50% bypass performance, and soft X-rays have 75% bypass performance.

2 is a diagram illustrating an air conditioner according to an exemplary embodiment.

The air conditioner 200 according to the present invention includes a purifying unit 220 for purifying air, a first data collecting unit 210 for acquiring information related to air flowing into the purifying unit 220, and a purifying unit ( It may include a second data collection unit 230 for acquiring information related to air leaking from 220 . The first data collection unit 210 and the second data collection unit 230 may each include at least one sensor to collect air quality-related data.

The purification unit 220 according to an embodiment may include at least one charging unit 240 for ionizing at least a portion of the material contained in the air, and a dust collecting unit for collecting the material ionized in the charging unit 240 . 250 may be included. The charging unit 240 may include at least one tube 241 for emitting electromagnetic waves. According to an embodiment, the tube 241 is an X-ray tube that generates X-rays through collision of electrons with a metal target, and may include an anode and a cathode, and additionally a gate electrode if necessary. can do. Here, the cathode emitting an electric field may include a field emission device that is composed of a carbon nanotube (CNT) and emits soft X-rays having a wavelength of 0.1 nm or more and 10 nm or less. In this case, the field emission device of the cathode, that is, the emitter may include a carbon nanotube structure including a plurality of unit yarns having a structure in which a plurality of carbon nanotubes are aggregated and extended in the first direction. . According to an embodiment, the charging unit 240 may ionize the material using electromagnetic waves having a wavelength in a range of 0.1 nm or more and 100 nm or less. For example, the charging unit 240 may perform photoionization on the contaminants using extreme ultraviolet or soft X-rays. The dust collecting unit 250 may include at least one dust collecting plate 251 for collecting the ionized material.

The air conditioner 200 according to the present invention may collect first data regarding the air quality of the air flowing into the apparatus 200 from the first data collection unit 210 . The first data may include at least one or more of information about the size of the space in which the device 200 is installed, humidity, temperature, wind speed, carbon dioxide and fine dust, volatile organic compounds (VOCs), and air pollutants such as formaldehyde. and may include all information related to air quality. The information on fine dust may include a concentration of fine dust and a concentration of ultrafine dust. The air passes through the purification unit 220 , and the purification unit 220 may be controlled based on the first data collected at this time. For example, when the device 200 is set to operate when the fine dust concentration in the air is 75 μg/m ³ or more and the ultrafine dust concentration is 30 μg/m ^{3 or more among the parameters for air quality, the first data collection} If the concentration of fine dust in the air according to the first data collected by the unit 210 is 80 μg/m ^{3 , the} device 200 will be operated.

According to an embodiment, after the air passes through the purification unit 220 , the second data collection unit 230 may collect second data regarding the air quality of the air. The second data may include at least one or more of information about the size of the space in which the device 200 is installed, humidity, temperature, wind speed, carbon dioxide and fine dust, volatile organic compounds (VOCs), and air pollutants such as formaldehyde. and may include all information related to air quality. As long as the air has passed through the purification unit 220 , the numerical value shown in the second data may be an improved value than the numerical value shown in the first data.

In the air conditioner 200 of the present invention, the set value related to the control of the purification unit 220 , the first data measured by the first data collection unit 210 , and the second data measured by the second data collection unit 230 . A model may be generated by machine learning based on the data, and the purification unit 220 may be controlled according to the generated model. In this case, the control of the purification unit 220 may mean controlling the voltage applied to the charging unit 240 and the voltage applied to the dust collecting unit 250 . The voltage applied to the charging part 240 may correspond to the voltage applied to the tube 241 . According to an embodiment, the tube 241 may include carbon nanotubes, and controlling the voltage applied to the electric charge unit 240 is an individual to the voltage applied to the anode, cathode, or gate of the tube 241 . control may be included. Before machine learning, the device 200 may be driven according to an initial value, and the initial value may be preset as a minimum air quality standard that satisfies the indoor air quality management method.

The air conditioner 200 of the present invention may control the purifier 220 to allow a customized operation for the space in which the apparatus 200 is installed. To this end, a threshold value may be derived based on the first data and the second data regarding the air quality of the air that has passed through the purifier 220 . According to an embodiment, the second data is data related to the air quality of the air that has passed through the purification unit 220 set according to the set value or initial value related to the control of the purification unit 220 derived through the machine learning model. can In this case, the threshold value may be a threshold value for a parameter related to air quality, and the parameters are the size of the space in which the air conditioner 200 is installed, humidity, temperature, wind speed, carbon dioxide and fine dust, volatile organic compounds (VOCs), and form. It may include at least one or more of information regarding air pollutants such as aldehydes. In addition, the threshold value may be different for each space in which the air conditioner 200 is installed. For example, since the air quality data for two places will be different when the space in which the device 200 is installed is inside a vehicle that is periodically ventilated and when it is a closed underground station, the The control conditions may also be different. For example, in the case of an underground station, since it is an enclosed space, it is necessary to set a lower threshold to perform air purification frequently. In addition, the threshold value may be distinguished according to the size of fine dust included in the second data. For example, if the size of fine dust particles is 10 μm or less (ie, PM 10), the standard applicable to fine dust may be applied, and if the size of fine dust particles is 2.5 μm or less (ie PM 2.5) case) standards applicable to ultrafine dust may be applied.

According to an embodiment, the air conditioner 200 may derive a control condition of the purifier 220 that causes the parameter to be equal to or less than a threshold value. The control condition of the purification unit 220 may include a voltage applied to the charging unit 240 and a voltage applied to the dust collecting unit 250 . The voltage applied to the charging part 240 may correspond to the voltage applied to the tube 241 . According to an embodiment, the tube 241 may include carbon nanotubes, and controlling the voltage applied to the electric charge unit 240 is an individual to the voltage applied to the anode, cathode, or gate of the tube 241 . control may be included.

On the other hand, depending on the product characteristics of the air conditioner 200, the number of tubes 241 used in the charging unit 240, the position of the tube 241, and the size of the dust collecting unit 250 (or the dust collecting plate 251), etc. may be different. Accordingly, the apparatus 200 determines the number of at least one tube 241 emitting electromagnetic waves from the electric charge unit 240, at least one tube 241 before machine learning for controlling the purification unit 220 is performed. The position of and the size of the dust collecting unit 250 may be input in advance.

According to an embodiment, the air conditioner 200 of the present invention may continuously update the machine learning model even after the purifier 220 is operated by deriving a control condition of the purifier 220 through machine learning. . Machine learning may be continued using the data sets of the first data and the second data obtained while the purification unit 220 normally operates according to the control condition derived through machine learning. When the air quality is improved even under the same control condition and the air can be purified with less power consumption, it is possible to derive the optimum threshold value at which the air purification is maintained while reducing the voltage supplied to the purification unit 220 . . If the air quality deteriorates under the same control condition, the device 200 will be continuously driven to adjust the air quality parameter to less than or equal to the threshold value. Machine learning may be continued using the data set of the data and the second data. The air conditioner 200 according to an embodiment may collect the first data and the second data in units of a predetermined time for a preset period for machine learning. For example, the device 200 may build a data set by collecting the first data and the second data in units of 5 seconds for a week.

According to an embodiment, the air conditioner 200 of the present invention generates a machine learning model by reinforcement learning that compensates based on parameters related to air quality and power used by the purifier 220 . can do. Specifically, the voltages of the charging unit 240 and the dust collecting unit 250 may be randomly selected every predetermined time within a range derived through machine learning, and first data and second data may be collected accordingly. In addition, reinforcement learning may be performed in a manner in which a score is given in a direction in which the power used by the device 200 decreases while a parameter related to air quality satisfies a threshold value. For example, when the voltage range applied to the charging unit 240 is set to 4.0 to 4.8 kV and the voltage range applied to the dust collector 250 is set to 2.8 to 4.4 kV as an initial condition, each partial voltage is a preset voltage It is possible to collect air quality data before and after the purification unit 220 by randomly selecting every 20 seconds within the range. As a result of reinforcement learning, when the voltage applied to the charging unit 240 is 4.3 kV and the voltage applied to the dust collecting unit 250 is 3.5 kV, the fine dust removal efficiency compared to the used power can come out to the maximum value, and accordingly, the threshold value is It can be set to 4.3 kV and 3.5 kV. According to an embodiment, the charging unit 240 may include a plurality of tubes 241 , and accordingly, the voltage applied to each tube 241 may be slightly different, but in this case, each tube 241 ) can be viewed as the voltage of the charging unit 240 by deriving an average value. Alternatively, there may be a plurality of charged units 240 in the purification unit 220 , but machine learning is performed by applying the same voltage value to each charged unit 240 rather than machine learning for each charged unit 240 . can To this end, the number of the charged parts 240 or the number of tubes 241 is input before the first machine learning, and will be considered during machine learning.

Meanwhile, a machine learning model may be generated by reinforcement learning, but a model applied to the air conditioner 200 of the present invention may be generated using other machine learning algorithms. For example, various machine learning algorithms such as unsupervised learning and rule-based machine learning algorithms may be applied to the present invention.

3 is a diagram illustrating the purification unit 220 according to an embodiment. Referring to FIG. 3 , the purification unit 220 may include a plurality of charging units 240 - 1 and 240 - 2 and a dust collecting unit 250 . Each of the charged units 240-1 and 240-2 is a module for ionizing pollutants in the air, and the first and second charged units 240-1 and 240- 2), or a voltage value sufficient to ignore the difference may be applied to the first charged unit 240 - 1 and the second charged unit 240 - 2 .

On the other hand, since the air quality data changes according to the positions of the sensors of the first data collection unit 210 and the second data collection unit 230 , the purification derived by the model learned based on the first data and the second data The control condition of the unit 220 may also be changed. Accordingly, the positions of the sensors of the first data collection unit 210 and the second data collection unit 230 may be changed according to the structure of the space in which the device 200 is installed, air quality characteristics, and the like.

According to an embodiment, a plurality of air conditioners 200 may be installed in the same space. In this case, the respective purification units 220 may be controlled according to different machine learning models for each of the devices 200 . That is, the threshold value applied to the air quality parameter may be different for each device 200 .

Compared to the conventional electric dust collector using the corona discharge method, the air conditioner 200 of the present invention has a higher dust collection efficiency compared to the power used, and the ozone generation amount converges to zero. For example, if the conventional electrostatic precipitator consumes 110 W of power and the dust collection efficiency against ionization is 10%, but the ozone generation amount is approximately 65 ppm, the air conditioner of the present invention consumes 40 W of power and is The amount of ozone generated may be approximately 0 ppm while the dust collection efficiency is 50%.

The air conditioner 200 of the present invention can derive optimal operating conditions according to the location and environment in which the device is installed, thereby minimizing the amount of energy used. In addition, since it uses machine learning, it may not be possible to configure the system unless it is a digital generating device that can control energy in real time. For this reason, it is difficult to construct the air conditioner 200 of the present invention using conventional ultraviolet or X-ray hot cathode tubes, and carbon nanotubes (CNTs), etc. that generate electromagnetic waves with relatively little energy and are easy to control energy It will be easy to construct the air conditioner 200 of the present invention.

For example, a conventional thermionic emission tube such as an ultraviolet or X-ray hot cathode tube is composed of a cathode made of a metal filament and an anode that is a metal target, and uses the principle that electrons are emitted from a heated metal filament (cathode). That is, the metal filament of the thermionic emission tube is heated to 1000° C. or higher to emit electrons, and the electrons are accelerated by the applied electric field to attack the metal target to generate electromagnetic waves. Such a thermionic emission tube has a limitation in that the response time is long because it must be heated to a high temperature in order to emit electrons from the metal filament.

On the other hand, in the field emission (FE) tube, electrons are emitted from the metal cathode by an applied electric field, but the temperature of the emitter is much lower than the temperature of the filament of the thermionic emission tube. In other words, since the temperature of the cathode is lower than that of the thermionic emission tube, the field emission tube has a longer lifespan and a faster response. In addition, it is possible to fine-tune the voltage control between the electrodes, and accordingly, it is easy to control the emission of electromagnetic waves at a required level. As such a field emission tube, a CNT tube using a carbon nanotube cathode may be used.

A CNT tube basically includes an anode of an anode and a cathode of a cathode and a gate for inducing the emission of electrons. Accordingly, the magnitude of the voltage applied to the anode, the cathode or the gate can be individually set and learned. Due to the voltage applied to the anode and the cathode, a current flows in the tube, and a current is generated as much as the voltage difference between the anode and the cathode. In reality, a high voltage is applied to the tube to give a magnetic field strong enough to emit electrons from the cathode. Therefore, leakage current occurs in the gate. Accordingly, the anode, cathode, or gate voltages may be individually controlled to generate a desired amount of current in the tube while minimizing gate leakage current.

4 is a flowchart illustrating a method of controlling an air conditioner according to an exemplary embodiment.

In step S401, the method may collect first data on the air quality of the air introduced into the air conditioner. According to an embodiment, the first data is at least one of information about the size of the space in which the air conditioner is installed, humidity, temperature, wind speed, carbon dioxide and fine dust, volatile organic compounds (VOCs), and information about air pollutants such as formaldehyde. may include more than one.

In step S402, the method may control the purification unit based on the collected first data. According to an embodiment, the purification unit may include at least one charged unit that ionizes at least a portion of the material included in the air, and a dust collector that collects the material ionized in the charged unit. The charged part may ionize a material using electromagnetic waves having a wavelength in the range of 0.1 nm or more and 100 nm or less, and specifically photoionize pollutants in the air using extreme ultraviolet or soft X-rays.

In step S403, the method may collect second data regarding the air quality of the air that has passed through the purification unit. According to an embodiment, the second data is at least one of information about the size of a space in which the air conditioner is installed, humidity, temperature, wind speed, carbon dioxide and fine dust, volatile organic compounds (VOCs), and air pollutants such as formaldehyde. may include more than one.

In step S404, the method may control the purification unit according to a machine learning model generated based on the set value related to the control of the purification unit, the first data, and the second data. According to an embodiment, the set value related to the control of the purification unit may include a set value for a voltage applied to the charged unit and a voltage applied to the dust collecting unit. According to an embodiment, the charged unit may include at least one tube emitting electromagnetic waves, and the set value for the voltage applied to the charged unit is a set value for the voltage applied to each of the plurality of electrodes constituting the tube. may include Accordingly, machine learning may be individual learning for each of the plurality of electrodes constituting the tube. Meanwhile, the method may collect the first data and the second data in units of a predetermined time for a preset period in order to control the purification unit according to the machine learning model.

In order to control the purification unit according to the machine learning model, the present invention derives a threshold value for a parameter related to air quality based on first data and second data according to a set value related to the control of the purification unit, and the parameter is It is possible to derive a control condition for the purification unit that falls below the threshold value. The control condition of the purification unit may include a voltage applied to the charging unit and a voltage applied to the dust collecting unit. According to an embodiment, the control condition for the voltage applied to the charged part may include individual control conditions for the voltage applied to the plurality of electrodes constituting the tube in the charged part. The parameter may include at least one of information about a size of a space in which the air conditioner is installed, humidity, temperature, wind speed, carbon dioxide, and air pollutants. In this case, the air pollutants may include, for example, one or more of airborne substances harmful to the human body, such as fine dust, volatile organic compounds (VOCs), and formaldehyde. Also, the threshold value may be different for each space in which the air conditioner is installed, and may be distinguished according to the size of fine dust included in the second data. For example, different threshold values may be derived according to ultrafine particles having a particle size of PM 2.5 or fine particles having a particle size of PM 10.

In relation to machine learning in step S404, the machine learning model according to an embodiment may be generated by inputting the number of at least one tube emitting electromagnetic waves from the electric charge part, the position of the at least one tube, and the size of the dust collecting part as input values. have. To this end, the method may pre-enter the input value prior to machine learning.

According to an embodiment, the method of the present invention may update the machine learning model based on the first data collected in step S401 and the second data according to the control condition of the purification unit derived in step S404. For example, first data obtained by sensing the air quality of the air flowing into the purification unit and second data obtained by sensing the air quality of the air passing through the purification unit after setting the purification unit to a set value derived through machine learning Based on this, the machine learning model can be continuously modified.

According to an embodiment, the machine learning model may be generated by reinforcement learning that performs compensation based on a parameter related to air quality and power used by the purifier. For example, in an environment in which a parameter satisfies a threshold value, a higher score may be given as the power used by the purification unit is lower, and a model may be generated in a direction to minimize the power used and increase the purification efficiency.

According to an embodiment, since a plurality of air conditioners may be installed, in this case, the respective purification units may be controlled according to different machine learning models for each air conditioner.

5 is a block diagram schematically illustrating an air conditioner 500 according to an exemplary embodiment. Referring to FIG. 5 , the air conditioner 500 may include a purification unit 510 , a first data collection unit 520 , a second data collection unit 530 , and a control unit 540 .

The first data collection unit 520 includes at least one sensor and may acquire information related to air flowing into the purification unit 510 . For example, the first data collection unit 520 may be provided before the purification unit 510 to collect first data regarding the air quality of the air before passing through the purification unit 510 . According to an embodiment, the first data collection unit 520 may be located before the charging unit of the purification unit 510 or may be located just before the dust collecting unit past the charging unit to sense air. This is an element that can be changed according to the characteristics of the space in which the air conditioner 500 is installed and the needs of the installing user.

The second data collection unit 530 includes at least one sensor and may acquire information related to the air flowing out from the purification unit 510 . For example, the second data collection unit 530 may be provided after the purification unit 510 to collect second data regarding the air quality of the air that has passed through the purification unit 510 .

The purification unit 510 may include at least one charged unit that ionizes at least a portion of the material contained in the air and a dust collector that collects the ionized material. By controlling the voltage applied to each of the charging unit and the dust collecting unit, the air purification efficiency per power consumption of the device 500 may be increased.

The control unit 540 may collect first data on air quality obtained through the first data collection unit 520 and control the purification unit 510 based on the first data. In addition, the control unit 540 collects second data related to the air quality of the air that has passed through the purification unit 510 through the second data collection unit 530 , and sets values related to the control of the purification unit 510 , The purification unit 510 may be controlled according to a machine learning model generated based on the first data and the second data.

According to an embodiment, the controller 540 is based on a machine learning model, and based on first data and second data according to a set value related to the control of the purifier 510 , a threshold value for a parameter related to air quality. can be derived. In addition, the controller 540 may control the purification unit 510 by deriving a control condition for the purification unit 510 in which the parameter falls below a threshold value. In particular, the machine learning model may be generated by reinforcement learning that performs compensation based on parameters related to air quality and power used by the purifier 510 . For example, as the air quality-related parameter satisfies a threshold value and the power used by the purification unit 510 decreases, the model may be reinforced by giving a large score to the model.

For example, when the air conditioner to which the machine learning of the present invention is not applied operates at maximum performance, a voltage of 4.8 kV is applied to the charging part and a voltage of 4.4 kV is applied to the dust collector, so that the total power used is 20 W and fine dust The removal efficiency may be 15% on average. That is, when the maximum performance operating condition is applied, the fine dust removal efficiency per power consumption corresponds to 0.75%/W. On the other hand, when the air conditioner to which the machine learning of the present invention is applied operates with optimal performance, a voltage of 4.3 kV is applied to the charging part and a voltage of 3.5 kV is applied to the dust collecting part, so that the total power used is 8 W, and the fine dust removal efficiency is This may be 14% on average. That is, when the optimal operating conditions are applied, the efficiency of removing fine dust per power consumption is 1.75%/W, which is higher than the case where machine learning is not applied. Therefore, the fine dust removal performance does not decrease despite the decrease in the amount of electricity used, so that the fine dust removal efficiency per power consumption can be increased.

The air conditioner 500 of FIG. 5 is exemplary, and may further include other components in addition to the components shown in FIG. 5 . Also, the device 500 may implement the above-described embodiments through components.

On the other hand, in the present specification and drawings, preferred embodiments of the present invention have been disclosed, and although specific terms are used, these are only used in a general sense to easily explain the technical content of the present invention and to help the understanding of the present invention, It is not intended to limit the scope of the invention. In addition to the embodiments disclosed herein, it is apparent to those of ordinary skill in the art that other modifications based on the technical spirit of the present invention can be implemented.

Claims (19)

A method for controlling an air conditioner based on machine learning, the method comprising:
collecting first data on air quality of air introduced into the air conditioner;
controlling a purification unit including a charging unit having at least one tube including an anode and a cathode based on the first data;
collecting second data on air quality of the air that has passed through the purification unit; and
Controlling the purification unit according to the number of the at least one tube, a set value related to the control of the purification unit, and a machine learning model generated based on the first data and the second data,
The set value related to the control of the purifier includes at least one of a voltage value applied to the anode and a voltage value applied to the cathode. According to claim 1,
The method of controlling an air conditioner, wherein the first data and the second data include at least one of information about a size of a space in which the air conditioner is installed, humidity, temperature, wind speed, carbon dioxide, and air pollutants. According to claim 1,
The control method of the air conditioner, wherein the purification unit further includes a dust collecting unit for collecting the ionized material. 4. The method of claim 3,
The set value related to the control of the purifier further includes a set value for a voltage applied to the dust collector. delete 4. The method of claim 3,
The method of controlling an air conditioner, wherein the charged part ionizes the material using electromagnetic waves having a wavelength in a range of 0.1 nm or more and 100 nm or less. 4. The method of claim 3,
The machine learning model is
The method of controlling an air conditioner, wherein the position of the at least one tube and the size of the dust collecting unit are generated as input values. According to claim 1,
The step of controlling the purification unit according to the machine learning model,
deriving a threshold value for a parameter related to air quality based on the first data and the second data according to a set value related to control of the purifier; and
and deriving a control condition of the purifier in which the parameter falls below the threshold value. 9. The method of claim 8,
The method of controlling an air conditioner, wherein the parameter includes at least one of information about a size, humidity, temperature, wind speed, carbon dioxide, and air pollutants of a space in which the air conditioner is installed. 9. The method of claim 8,
The threshold value is different for each space in which the air conditioner is installed and is distinguished according to size information of fine dust included in the second data. 9. The method of claim 8,
The purification unit may include at least one charging unit; and
including a dust collector,
The control method of the air conditioner, wherein the control condition of the purification unit includes a voltage applied to the charging unit and a voltage applied to the dust collecting unit. 9. The method of claim 8,
and updating the machine learning model based on the first data and the derived second data according to the control condition of the purification unit. According to claim 1,
The step of controlling the purification unit according to the machine learning model,
The method of controlling an air conditioner further comprising the step of collecting the first data and the second data in units of a predetermined time period for a preset period. According to claim 1,
The machine learning model is generated by reinforcement learning for performing compensation based on a parameter related to air quality and power used by the purifier. According to claim 1,
A control method of an air conditioner, wherein each of the plurality of air conditioners is controlled according to different machine learning models. 10. The method of claim 2 or 9,
The method for controlling an air conditioner, wherein the air pollutants include at least one of fine dust, volatile organic compounds (VOCs), and formaldehyde. A machine learning-based air conditioner comprising:
a purifier comprising a charging section having at least one tube comprising an anode and a cathode;
a first data collection unit including at least one sensor and acquiring information related to air flowing into the purification unit;
a second data collection unit including at least one sensor and acquiring information related to air flowing out from the purification unit; and
comprising a control unit;
The control unit is
Collecting first data on air quality acquired through the first data collection unit,
controlling the purifier based on the first data,
Collecting second data on the air quality of the air that has passed through the purification unit through the second data collection unit,
controlling the purification unit according to a machine learning model generated based on the number of the at least one tube, a set value related to the control of the purification unit, and the first data and the second data,
The set value related to the control of the purifier includes at least one of a voltage value applied to the anode and a voltage value applied to the cathode. 18. The method of claim 17,
The control unit is
Based on the machine learning model, deriving a threshold value for a parameter related to air quality based on the first data and the second data according to a set value related to the control of the purifier,
and controlling the purification unit by deriving a control condition for the purification unit in which the parameter falls below the threshold value. 18. The method of claim 17,
The machine learning model is generated by reinforcement learning for performing compensation based on a parameter related to air quality and power used by the purifier.

Images