KR102664464B1 - A deep-learning based energy-saving air-conditioner and control method thereof - Google Patents

Abstract

The present invention relates to a deep learning-based energy-saving air conditioner including an outdoor unit and an air conditioner (AHU) that cools or heats an indoor space by exchanging heat between the heat medium supplied from the outdoor unit and indoor air, and is provided on one side of the air conditioner. A rotary heat exchanger unit installed to exchange heat between indoor air exhausted to the outside and outdoor air flowing into the air conditioner, a temperature and humidity sensing unit measuring temperature and humidity information of the indoor and outdoor air to be exchanged heat, and the rotary heat exchanger. It includes a control module that controls the operation of the base, and the control module inputs temperature and humidity information about the outdoor air (OA) before the heat exchange measured by the temperature and humidity sensing unit to determine the rotation speed of the rotary heat exchanger unit. It is characterized by including a deep learning unit that has been machine learned to output.

Description

Deep learning based energy-saving air-conditioner and control method thereof}

The present invention relates to a deep learning-based energy-saving air conditioner and a control method thereof. More specifically, in an air conditioner applying a rotary total heat exchanger, the deep learning technique is used to control the air conditioner according to the operating conditions of the air conditioner. This study relates to a deep learning-based energy-saving air conditioner and its control method that can contribute to energy savings by improving the performance and total heat exchange efficiency of the air conditioner by controlling the rotation speed of the typical total heat exchanger at the optimal speed.

In general, an air conditioning device refers to a device that optimally adjusts the temperature, humidity, and airflow of air supplied indoors. It is usually used in small air conditioning devices for home use and in large spaces such as buildings and factories. It is classified into a circulating air conditioning device.

Among these, the circulating air conditioning device connects the indoor space with an air conditioner in which heat exchange between refrigerant and air is performed through an air channel formed on the wall or ceiling, and the air conditioner continuously supplies air with controlled temperature and humidity to the indoor space. This air conditioner is composed of a plurality of panels assembled to form a hexahedral tubular duct, and is mainly installed fixedly on the outside of a building or on the floor of a basement (lower floor).

In the case of a circulating air conditioning device configured as above, the air conditioner sucks in indoor air, exhausts some of it, mixes the rest with outdoor air, exchanges heat with a heat exchanger through which refrigerant flows, and returns to the indoor space through the air flow path. It is configured to supply.

In this case, in order to save energy by recovering the waste heat of the exhausted indoor air, most circulation air conditioners use a total heat exchanger that recovers the waste heat of the indoor air by exchanging heat between the exhausted indoor air and the incoming outdoor air. The detailed configuration of a circulation air conditioning device using a total heat exchanger for exhaust heat recovery is disclosed in detail in [Document 1] and [Document 2] below.

The total heat exchanger for exhaust heat recovery can be divided into a plate-type total heat exchanger (in the case of Document 1) and a rotary total heat exchanger (in the case of Document 2) depending on the structure of the heat exchanger and the total heat exchange method. Because of its relatively excellent efficiency, the spread of products using rotary total heat exchangers has been gradually increasing recently.

However, in the case of such a rotary total heat exchanger, it is known that the total heat exchange efficiency increases as the rotation speed increases, but this is due to various loss factors (increased leakage rate, increased input of the total heat exchanger drive motor, mechanical Because it does not reflect increased wear and failure rate), depending on the operating conditions of the air conditioner, rotating the total heat exchanger at high speed improves the performance and/or total heat exchange of the air conditioner compared to rotating the total heat exchanger at low or medium speed. There was a problem that made it more disadvantageous in terms of efficiency.

[Document 1] Korean Patent Publication No. 2001-0094498 (published on November 1, 2001)

[Document 2] Korean Patent No. 10-1036193 (announced on May 16, 2011)

The present invention is to solve the problems of the prior art as described above. The purpose of the present invention is to use a deep learning technique in an air conditioner applying a rotary total heat exchanger to determine the rotary total heat exchanger according to the operating conditions of the air conditioner. The purpose is to provide a deep learning-based energy-saving air conditioner and its control method that can contribute to energy savings by improving air conditioner performance and heat exchange efficiency by controlling the rotation speed of the air conditioner at an optimal speed.

In order to achieve the above object, the present invention provides a deep learning-based energy-saving air conditioning system that includes an outdoor unit and an air conditioner (AHU) that cools or heats the indoor space by exchanging heat between the heat medium supplied from the outdoor unit and indoor air. In the apparatus, a rotary heat exchanger unit installed on one side of the air conditioner to exchange heat between indoor air exhausted to the outside and outdoor air flowing into the air conditioner, and measuring temperature and humidity information of the indoor and outdoor air for which the heat is exchanged. It includes a temperature and humidity sensing unit and a control module that controls the operation of the rotary heat exchanger unit, wherein the control module receives temperature and humidity information about the outdoor air (OA) before the heat exchange measured by the temperature and humidity sensing unit as input. It is characterized by including a deep learning unit that is machine-learned to output the rotation speed of the rotary heat exchanger unit.

In addition, the control module controls the outdoor air (SA) after total heat exchange and the indoor air (RA) before total heat exchange, which are measured for each rotational speed of the rotary total heat exchanger unit, with respect to the temperature and humidity conditions of outdoor air (OA). A memory unit storing indoor/outdoor air temperature/humidity information, which is temperature/humidity information for indoor/outdoor air, operating data calculating heat exchange efficiency at each rotational speed of the rotary heat exchanger unit for the temperature/humidity conditions of the outdoor air (OA) using the indoor/outdoor air temperature/humidity information. It further includes a heat efficiency calculating unit that generates a learning data generation unit that generates learning data for deep learning by correcting the heat exchange efficiency of the operation data by applying a predetermined weight for each rotation speed, wherein the deep learning unit generates the learning data. It is characterized by machine learning to output the rotation speed with the highest heat exchange efficiency for each temperature and humidity condition of outdoor air (OA).

In addition, the weight for each rotational speed includes at least one of the following factors: an increase in air leakage rate due to an increase in the rotational speed of the rotary heat exchanger, an increase in driving force of the rotary heat exchanger, an increase in wear of the mechanism, and an increase in failure rate of the mechanism. It is characterized by being determined.

In addition, the weight for each rotational speed is determined to decrease as the rotational speed of the rotary heat exchanger unit increases, and is determined differently depending on the operation mode of the air conditioner, which is either a cooling operation or a heating operation. .

In addition, the control unit collects the indoor and outdoor air temperature and humidity information while changing the rotation speed of the rotary heat exchanger unit with respect to the current temperature and humidity conditions of the outdoor air (OA) according to a predetermined method during operation of the air conditioner, and the memory unit collects the temperature and humidity information of the indoor and outdoor air. Further perform a temperature and humidity information collection mode to store the temperature and humidity information, and control the operations of the thermal efficiency calculation unit, the learning data generation unit and the deep learning unit so that the deep learning unit re-learns using the newly stored indoor and outdoor air temperature and humidity information according to a predetermined cycle. It is characterized by:

In addition, the present invention relates to a control method of a deep learning-based energy-saving air conditioner including an outdoor unit and an air conditioner (AHU) that cools or heats an indoor space by exchanging heat between the heat medium supplied from the outdoor unit and indoor air, (a) measuring temperature and humidity information of outdoor air (OA) before heat exchange in a rotary heat exchanger that exchanges heat between indoor air exhausted to the outside of the air conditioner and outdoor air flowing into the air conditioner, (b) determining the rotation speed of the rotary total heat exchanger using a machine-learned deep learning module to input the measured temperature and humidity information of the outdoor air (OA) and output the rotation speed of the rotary total heat exchanger. , and (c) rotating the rotary total heat exchanger at the determined rotation speed.

In addition, (d) further includes the step of machine learning the deep learning module, wherein the step (d) is performed after the total heat exchange according to the rotation speed of the rotary total heat exchanger with respect to the temperature and humidity conditions of the outdoor air (OA). Measuring indoor and outdoor air temperature and humidity information, which is temperature and humidity information about indoor air (RA) before heat exchange with outdoor air (SA), and using the indoor and outdoor air temperature and humidity information to determine the temperature and humidity conditions of the outdoor air (OA) Generating operation data by calculating the heat exchange efficiency for each rotational speed of a typical heat exchanger, generating learning data for deep learning by correcting the heat exchange efficiency of the operation data by applying a predetermined weight for each rotation speed, And a machine learning step using the learning data to output a rotational speed with the highest heat exchange efficiency for each temperature and humidity condition of outdoor air (OA).

In addition, (e) collecting the indoor and outdoor air temperature and humidity information while changing the rotational speed of the rotary total heat exchanger with respect to the current temperature and humidity conditions of the outdoor air (OA) according to a predetermined method while operating the air conditioner, ( f) further comprising the step of retraining the deep learning module using the indoor and outdoor air temperature and humidity information collected in step (e) according to a predetermined cycle.

The deep learning-based energy-saving air conditioner and its control method according to the present invention consider the influence of loss factors due to an increase in the rotation speed of the rotary heat exchanger in an air conditioner applying a rotary heat exchanger. Since it is configured to control the rotational speed of the total heat exchanger at an optimal speed, it has the advantage of contributing to energy saving by improving the performance of the air conditioner and total heat exchange efficiency.

In addition, the deep learning-based energy-saving air conditioner and its control method according to the present invention use deep learning techniques to optimize the rotary heat exchanger according to the operating conditions (operating mode and outdoor air temperature and humidity conditions) of the air conditioner. Since it is configured to determine the speed, it has the advantage of achieving a highly reliable energy saving effect even under arbitrary operating conditions of the air conditioner.

1 is a diagram showing the overall configuration of a deep learning-based energy-saving air conditioner according to an embodiment of the present invention;
Figure 2 is a diagram showing the configuration of an air conditioner applied to the deep learning-based energy-saving air conditioner shown in Figure 1;
Figure 3 is a diagram for explaining the total heat exchange process in the rotary total heat exchanger installed in the air conditioner shown in Figure 2;
Figures 4a and 4b are block diagrams for explaining the operation configuration of a deep learning-based energy-saving air conditioner, respectively;
Figure 5 is a diagram showing an example of operation data and learning data generated by the heat efficiency calculation unit and learning data generation unit shown in Figure 4b;
Figure 6 is a diagram for explaining the structure of the learning algorithm of the deep learning unit shown in Figure 4b;
Figures 7 and 8 are flow charts to explain the control method of the deep learning-based energy-saving air conditioner shown in Figure 1, respectively.

Hereinafter, preferred embodiments of the present invention will be described in detail using the accompanying drawings.

Figure 1 is a diagram showing the overall configuration of a deep learning-based energy-saving air conditioner according to an embodiment of the present invention, and Figure 2 is an air conditioner applied to the deep learning-based energy-saving air conditioner shown in Figure 1. This is a diagram showing the configuration, and Figure 3 is a diagram for explaining the total heat exchange process in the rotary total heat exchanger installed in the air conditioner shown in Figure 2.

The deep learning-based energy-saving air conditioner according to the present invention includes an outdoor unit (20) that supplies a heat medium, and air that cools or heats the indoor space (30) by heat exchanging the heat medium supplied from the outdoor unit (20) and indoor air. It consists of a regulator (10) (AHU, Air Handling Unit).

At this time, the outdoor unit 20 functions as a cold heat source or a hot heat source and may be composed of a normal air conditioner outdoor unit for cooling, a heat pump outdoor unit such as EHP/GHP, a chiller, or a boiler. For convenience of explanation, this embodiment In the example, the case where the outdoor unit 20 is an EHP outdoor unit will be described as an example.

In the case of this embodiment, the outdoor unit 20 has a compressor (not shown), a condenser (not shown), and an expansion valve (not shown) inside, together with a coil heat exchanger 142 inside the air conditioner 10 to be described later. Constituting a refrigeration cycle, the coil heat exchanger (142) of the air regulator (10) is operated through a discharge pipe (21) that supplies low-temperature, low-pressure refrigerant during cooling and a suction pipe (22) through which refrigerant that has exchanged heat with indoor air flows in. is connected to

In addition, the air conditioner 10 is connected to at least one indoor space 30 requiring cooling and heating through an intake pipe 40 and an air supply pipe 50, where the intake pipe 40 and the air supply pipe 50 are It is installed to communicate with the interior of the indoor space 30 through the wall or ceiling of the building.

In this case, the intake pipe 40 and the air supply pipe 50 are preferably configured in the form of pipes or ducts.

As described above, the air conditioner (10) connected to the indoor space (30) sucks indoor air through the intake pipe (40), then discharges some of it to the outside through the exhaust pipe (60) and the outdoor air inlet pipe (70). ) is mixed with the outdoor air introduced through the remaining indoor air to exchange heat with the coil heat exchanger 142 through which the refrigerant flows.

The air that has undergone heat exchange in this way is supplied to the indoor space 30 through the air supply pipe 50. On one side of the wall or ceiling of the indoor space, there is an intake hole 31 communicating with the intake pipe 40 and an air supply pipe ( A supply hole 32 communicating with 50 is formed to allow intake of indoor air and supply of air from the air conditioner 10.

On the other hand, the air conditioner 10 includes a housing 100 formed in the shape of a rectangular duct by combining a plurality of panels, an intake unit 110 formed inside the housing 100, an air mixing unit 120, and a It is comprised of a first heat exchange unit 130, a second heat exchange unit 140, and an air supply unit 150.

The intake unit 110 sucks indoor air from the indoor space 30 and flows it into the housing 100. For this purpose, the intake unit 110 has an intake port 112 that communicates with the intake pipe 40. It is configured to include a first blowing fan 111 that sucks indoor air through.

In the case of this embodiment, as an example, the intake port 112 is formed on the ceiling surface of one end of the housing 100, and the first blowing fan 111 blows the sucked indoor air into the housing 100 parallel to the ground. It is configured to discharge toward the internal flow path.

In addition, the air mixing unit 120 exhausts some of the indoor air sucked into the housing 100 by the first blowing fan 111 and introduces new outdoor air.

For this purpose, the air mixing unit 120 is divided into a first space 123 adjacent to the above-described intake unit 110 by the central partition 125 and a second space adjacent to the second heat exchange unit 140 described later ( 124), wherein an exhaust port 121 is formed in the upper part of the first space 123 through which some of the sucked indoor air is discharged to the outside, and outdoor air flows in in the upper part of the second space 124. An outdoor air inlet 122 is formed.

Therefore, some of the indoor air introduced into the first space 123 by the first blowing fan 111 is discharged to the outdoors through the exhaust port 121, and the remainder is discharged to the outside through the bypass damper 126 formed on the partition wall 125. ) flows into the second space 124 and mixes with the outdoor air flowing in through the outdoor air inlet 122.

In addition, the first heat exchange unit 130 is installed between the exhaust port 121 and the outdoor air inlet 122 and the air mixing unit 120 inside the housing 100 to discharge the air discharged from the air mixing unit 120. It consists of a rotary total heat exchanger that exchanges total heat between indoor air and outdoor air flowing into the air mixing unit 120.

Specifically, the first heat exchange unit 130 includes a disk-shaped casing 131 installed between the exhaust port 121 and the outdoor air inlet 122 and the air mixing unit 120 inside the housing 100, and the casing A rotating shaft 132 formed in the central portion of the casing 131, a disk-shaped total heat exchanger rotor 133 rotatably coupled to the rotating shaft 132 inside the casing 131, and the total heat exchanger rotor 133 are rotated. It is composed of a total heat exchanger driving unit 230.

At this time, as shown in FIG. 3, the total heat exchanger rotor 133 rotates and transfers the sensible heat and latent heat transferred from the indoor air exhausted from the first space 123 to the outdoor air flowing into the second space 124. Heat is exchanged through heat transfer.

In the detailed description and patent claims of the present invention, for convenience of explanation, the outdoor air before the total heat exchange is referred to as 'outdoor air (OA)', and the second space of the air conditioner 10 after the total heat exchange is performed ( 124) The outdoor air introduced inside will be referred to as ‘outdoor air (SA)’, and the indoor air before heat exchange will be referred to as ‘indoor air (RA)’.

In addition, the second heat exchange unit 140 is disposed behind the air mixing unit 120 and mixes the indoor air and outdoor air supplied from the air mixing unit 120 with the mixed air supplied from the outdoor unit 20. Heat exchange between heating mediums is performed. For this purpose, the second heat exchange unit 140 includes an air filter 141 to remove contaminants from the air flowing in from the air mixing unit 120, and an air filter 141 from which the contaminants are removed. It is configured to include a coil heat exchanger 142 that exchanges heat with air and the heat medium.

At this time, the air filter 141 and the coil heat exchanger 142 are installed vertically in the internal flow path of the housing 100, so that the mixed air of indoor air and outdoor air flowing through the housing 100 flows almost vertically. It is configured to pass through the front part of the air filter 141 and the coil heat exchanger 142.

In addition, as described above, the air that has passed through the second heat exchange unit 140 is supplied to each room through the air supply port 152 communicated with the air supply pipe 50 by the second blowing fan 151 of the air supply unit 150. By being supplied to the space 30, the indoor space 30 is cooled and heated.

Next, the operation configuration of the deep learning-based energy-saving air conditioner according to the present invention configured as described above will be described using FIGS. 4A to 7.

Figures 4a and 4b are block diagrams for explaining the operation configuration of a deep learning-based energy-saving air conditioner, and Figure 5 shows the operation data and learning generated by the thermal efficiency calculation unit and the learning data generation unit shown in Figure 4b. This is a diagram showing an example of data, and Figure 6 is a diagram for explaining the structure of the learning algorithm of the deep learning unit shown in Figure 4b.

The deep learning-based energy-saving air conditioner according to the present invention includes an input and output unit 210 that is installed on one side of the indoor space 30 or the air conditioner 10 and inputs or outputs information related to the operation of the air conditioner. A temperature and humidity sensing unit 220 for measuring information on temperature and humidity required for operation of the air conditioner, and operation of the air conditioner according to information input from the input/output unit 210 and the temperature and humidity sensing unit 220. It is configured to include a control module 200 that controls.

In addition, the temperature and humidity sensing unit 220 detects outdoor air (OA) before total heat exchange, outdoor air (SA) after total heat exchange, and air before total heat exchange near the first heat exchange unit 130, which is a rotary total heat exchanger. It is configured to measure temperature and humidity information of indoor air (RA), and, if necessary, measures information on temperature and humidity for each part of the air conditioner 10, such as inside the indoor space 30 or near the coil heat exchanger 142. You may.

At this time, the temperature and humidity information may be expressed as temperature and humidity using a thermometer and a hygrometer, or as dry bulb temperature and wet bulb temperature using a wet and dry bulb thermometer.

In this embodiment, as an example, the temperature and humidity information is configured to be expressed as dry bulb temperature and wet bulb temperature. In this case, the humidity can be obtained by searching the difference between dry bulb temperature and wet bulb temperature in a humidity table or psychrometric diagram.

In addition, the deep learning-based energy-saving air conditioner according to the present invention includes a total heat exchanger driving part 230 that rotates the total heat exchanger rotor 133, and a blowing fan driving part 240 that drives the first and second blowing fans 111 and 151. ) and an outdoor unit driving unit 250 that drives the compressor of the outdoor unit 20, and the control module 200 uses information input from the input/output unit 210 and the temperature and humidity sensing unit 220. Thus, the operation of the air conditioner is controlled by controlling the operations of the total heat exchanger driver 230, the blower fan driver 240, and the outdoor unit driver 250.

In addition, the control module 200 of the deep learning-based energy-saving air conditioner according to the present invention includes a control unit 201 that controls the operation of the air conditioner in the above-described manner, and temperature and humidity information measured by the temperature and humidity sensing unit. A stored memory unit 202, a heat transfer efficiency calculation unit 203 that calculates the heat exchange efficiency of the first heat transfer unit 130 (i.e., a rotary heat exchanger or a heat exchange rotor) using the temperature and humidity information, and a deep learning technique. A deep learning unit 204 determines the rotation speed of the heat exchanger rotor 133 according to the temperature and humidity information, and generates learning data for the deep learning unit 204 using the heat exchange efficiency. It is composed of a unit 205.

At this time, the memory unit 202 contains the outdoor air (SA) and indoor air measured at the rotation speed of the rotary heat exchanger unit (i.e., the total heat exchanger rotor 133) according to the temperature and humidity conditions of the outdoor air (OA). Temperature and humidity information of the air (RA) may be stored, and the temperature and humidity information may be measured and stored in advance through test runs, etc.

In addition, the thermal efficiency calculation unit 203 uses the temperature and humidity information stored in the memory unit 202 to determine the rotary heat exchanger unit (i.e., the total heat exchanger rotor 133) according to the temperature and humidity conditions of the outdoor air (OA). The heat exchange efficiency is calculated for each rotation speed, and the operation data generated using this is stored in the memory unit 202.

In addition, the total heat exchange efficiency (η _TH ) of the rotary total heat exchanger unit (i.e., the total heat exchanger rotor 133) can be obtained by [Equation 1] below.

[Formula 1]

At this time, h in the above formula refers to enthalpy, which can be obtained using the temperature and humidity information of the relevant air and the hygroscopic diagram.

Figure 5(a) shows an example of operation data generated by the thermal efficiency calculation unit 203. The operation data includes heat transfer at the rotation speed of the heat exchanger rotor 133 according to the temperature and humidity conditions of the outdoor air (OA). Exchange efficiency is saved.

In this embodiment, as an example, a case where the rotation speed of the total heat exchanger rotor 133 is comprised of three predetermined levels of low speed, medium speed, and high speed is described, but the case is not limited to this, and if necessary, the rotation speed is performed in RPM units. It may be possible.

Additionally, the operation data may be generated for cooling operation and heating operation of the air conditioner, respectively.

Meanwhile, the total heat exchange efficiency according to the rotation speed of the total heat exchanger rotor 133 shown in (a) of Figure 5 shows that the total heat exchange efficiency is highest when the rotation speed is high under the overall temperature and humidity conditions, which is simply the temperature and humidity of the indoor and outdoor air. This is a result calculated according to the above [Equation 1] using only information, and is a value that does not take into account the loss factor due to an increase in the rotation speed of the total heat exchanger rotor 133.

Therefore, if the total heat exchanger rotor 133 is unconditionally rotated at high speed in all operating conditions of the air conditioner, the performance and total heat exchange efficiency of the air conditioner may be reduced for certain operating conditions.

The present invention is intended to solve this problem, by applying a weight considering the loss factor due to an increase in the rotation speed of the total heat exchanger rotor 133 to the calculated heat exchange efficiency at each rotation speed, thereby determining the air conditioner operating conditions (i.e. It is characterized in that it is configured to rotate the total heat exchanger rotor 133 by determining the optimal rotation speed for each condition (cooling or heating operation, temperature and humidity conditions of indoor and outdoor air).

At this time, the optimal rotation speed for each operating condition of the total heat exchanger rotor 133 is obtained based on deep learning, as will be described later.

To this end, the learning data generation unit 205 first applies predetermined weights for each rotational speed to the heat exchange efficiency of the operation data to generate learning data for deep learning (or machine learning) of the deep learning unit 204, which will be described later. is generated, and Figure 5(b) shows an example of the learning data.

At this time, the weights for each rotational speed include an increase in air leakage rate, an increase in the driving force of the total heat exchanger, an increase in wear of the mechanical part, an increase in the failure rate of the mechanical part, etc., as the rotational speed of the rotary heat exchanger (i.e., the total heat exchanger rotor 133) increases. It is desirable to determine it by including at least one loss factor among the loss factors including.

Therefore, the weight for each rotation speed is determined to decrease as the rotation speed of the total heat exchanger rotor 133 increases. This weight is determined differently depending on the operation mode of the air conditioner, which is either cooling operation or heating operation. It is desirable.

In this embodiment, as an example, the weights for each rotational speed were applied as shown in [Table 1] in consideration of the air leakage rate increase factor, the total heat exchanger driving force increase factor, the mechanical part wear increase factor, and the mechanical part failure rate increase factor. These weights were applied based on the air Of course, it may be set differently depending on the type of conditioner and installation environment.

sleaze medium speed high speed Weight during cooling operation +8.5% +2% +0% Weight during heating operation +3% +1% +0%

In this way, when a weight for each rotation speed is applied to the heat exchange efficiency for each rotation speed of the operation data, the heat exchanger rotor ( 133), it can be seen that the rotation speed appears different.

In addition, the deep learning unit 204 performs machine learning to output the rotational speed of the rotary heat exchanger (i.e., the total heat exchanger rotor 133) by inputting the measured temperature and humidity information of the outdoor air (OA). Specifically, using the learning data, machine learning is performed to output the rotational speed with the highest heat exchange efficiency (i.e., heat exchange efficiency corrected by applying weights) for each temperature and humidity condition of outdoor air (OA).

The deep learning unit 204 may be configured to perform the machine learning by any one of known deep learning algorithms such as RNN, CNN, DNN, and MLP. In this embodiment, the structure shown in FIG. 6 is an example. Machine learning was performed using a multi-layer perceptron (MLP) neural network model.

At this time, the deep learning unit 204 uses temperature and humidity information (dry bulb temperature and wet bulb temperature in this embodiment) among the learning data as input and performs machine learning to output the rotation speed with the highest heat exchange efficiency.

When the machine learning of the deep learning unit 204 is completed using the temperature and humidity information stored in the memory unit 202 in the same manner as described above, the control unit 201 determines the outdoor air measured by the temperature and humidity sensing unit 220. The temperature and humidity information of (OA) is input to the deep learning unit 204, and the total heat exchanger driving unit 230 is operated so that the total heat exchanger rotor 133 rotates at the output rotation speed.

Additionally, the electric heat exchanger driving unit 230 may be preferably implemented by any one of known electric motors capable of variable rotation speed.

Meanwhile, the control unit 201 changes the rotation speed of the rotary heat exchanger (i.e., the total heat exchanger rotor 133) with respect to the current temperature and humidity conditions of the outdoor air (OA) according to a predetermined method during operation of the air conditioner. It may be configured to further perform a temperature and humidity information collection mode in which temperature and humidity information about the indoor and outdoor air is collected and stored in the memory unit 202 while doing so.

At this time, the temperature and humidity information collection mode is set to be performed at a predetermined period, performed during a specific operation mode such as dehumidification operation, or performed when the outdoor air (OA) temperature and humidity conditions change beyond the set value. It can be.

In this temperature and humidity information collection mode, the amount of temperature and humidity information initially stored in the memory unit 202 is limited, so that the accuracy of the learning results of the deep learning unit 204 using this mode is reduced. This is performed to additionally secure the learning data.

Therefore, the control unit 201 operates the thermal efficiency calculation unit 203, the learning data generation unit 205, and the deep learning unit 204 to re-learn the deep learning unit 204 using the newly stored temperature and humidity information according to a predetermined cycle. The operation of the unit 204 is controlled, and the re-learning cycle can be set on a daily basis, weekly basis, monthly basis, or yearly basis as needed.

Finally, the control method of the deep learning-based energy-saving air conditioner according to the present invention configured as described above will be explained using the flowcharts shown in FIGS. 7 and 8.

When an operation signal of the air conditioner (or air conditioner 10) is received through the input/output unit 210 (S110), the control unit 201 controls the blower fan driver 240 and the outdoor unit driver 250. The outdoor unit 20 and the first and second blowing fans 111 and 151 are operated to operate the air conditioner (S120).

When step S120 is completed, the control unit 201 measures temperature and humidity information about the outdoor air (OA) using the temperature and humidity sensing unit 220 as described above (S130), and stores the measured temperature and humidity information. It is input into the deep learning unit 204.

In addition, when step S130 is completed, the control unit 201 determines the rotation speed of the total heat exchanger rotor 133 at the rotation speed output from the deep learning unit 204 (S140), and according to the determined rotation speed The operation of the total heat exchanger driving unit is controlled so that the total heat exchanger rotor 133 rotates (S150).

When step S150 is completed, the control unit 201 determines whether the temperature and humidity information collection mode operation conditions pre-stored in the driving program, etc. are satisfied (S160), and if the determination result is satisfied, performs step S200 as described later. do.

On the other hand, if the determination result in step S160 is that the temperature and humidity information collection mode operation conditions are not met, the control unit 201 determines whether the air conditioner operation stop signal has been received (S170). If not, the control unit 201 determines whether the air conditioner operation stop signal has been received (S170). Step S130 is performed repeatedly.

In addition, if an air conditioner operation stop signal is received as a result of the determination in step S170, the control unit 201 controls the outdoor unit 20, the first and second blowing fans 111 and 151, and the total heat exchanger rotor 133. By turning off the drive, the operation of the air conditioner is stopped and control is terminated.

Meanwhile, as a result of the determination in step S160, if the temperature and humidity information collection mode operation conditions are met, the control unit 201 adjusts the rotational speed of the total heat exchanger rotor 133 with respect to the current temperature and humidity conditions of the outdoor air (OA). While changing, the temperature and humidity information of the outdoor air (SA) and indoor air (RA) are measured at each rotation speed (S210).

When step S210 is completed, the control unit 201 calculates the heat exchange efficiency for each rotational speed of the heat exchanger rotor 133 as described above through the heat efficiency calculation unit 203 and generates new operation data (S220).

When step S220 is completed, the control unit 201 uses the newly generated driving data to generate new learning data by the learning data generating unit 205 in the same manner as described above (S230).

At this time, the control unit 201 may store the newly measured temperature and humidity information, newly generated operation data, and learning data in the memory unit 202.

When step S240 is completed, the control unit 201 controls the operation of the deep learning unit 204 to re-learn using the newly generated training data (S240), and then performs step S170.

The deep learning-based energy-saving air conditioner and its control method according to the present invention configured as described above are the effects of loss factors due to an increase in the rotation speed of the rotary heat exchanger in an air conditioner applying a rotary heat exchanger. Since it is configured to control the rotational speed of the rotary total heat exchanger at an optimal speed in consideration of, it has the advantage of contributing to energy saving by improving the performance of the air conditioner and total heat exchange efficiency.

In addition, the deep learning-based energy-saving air conditioner and its control method according to the present invention use deep learning techniques to optimize the rotary heat exchanger according to the operating conditions (operating mode and outdoor air temperature and humidity conditions) of the air conditioner. Since it is configured to determine the speed, it has the advantage of achieving a highly reliable energy saving effect even under arbitrary operating conditions of the air conditioner.

10: Air conditioner (AHU) 20: Outdoor unit
110: intake section 120: air mixing section
130: first heat exchanger 133: total heat exchanger rotor
140: second heat exchange unit 150: air supply unit
200: Control module 203: Heat efficiency calculation unit
204: Deep learning unit 205: Learning data generation unit
220: Temperature and humidity sensing unit 230: Electric heat exchanger driving unit

Claims (10)

In a deep learning-based energy-saving air conditioner that includes an outdoor unit and an air conditioner (AHU) that cools or heats the indoor space by exchanging heat between the heat medium supplied from the outdoor unit and indoor air,
A rotary heat exchanger unit installed on one side of the air conditioner to exchange heat between indoor air exhausted to the outside and outdoor air flowing into the air conditioner;
A temperature and humidity sensing unit that measures temperature and humidity information of the indoor and outdoor air undergoing heat exchange;
It includes a control module that controls the operation of the rotary heat exchanger unit,
The control module is,
Indoor and outdoor air, which is temperature and humidity information for the outdoor air (SA) after total heat exchange and indoor air (RA) before total heat exchange, measured at each rotation speed of the rotary heat exchanger unit for the temperature and humidity conditions of the outdoor air (OA) A memory unit storing temperature and humidity information;
A thermal efficiency calculation unit that generates operating data that calculates heat exchange efficiency for each rotational speed of the rotary heat exchanger unit for the temperature and humidity conditions of the outdoor air (OA) using the indoor and outdoor air temperature and humidity information;
A learning data generator that generates learning data for deep learning by correcting the heat exchange efficiency of the operation data by applying a predetermined weight for each rotation speed; and
It includes a machine-learned deep learning unit to output the rotation speed of the rotary heat exchanger unit by inputting the temperature and humidity information about the outdoor air (OA) before the heat exchange measured by the temperature and humidity sensing unit,
The deep learning unit uses the learning data to perform machine learning to output the rotation speed with the highest heat exchange efficiency for each temperature and humidity condition of outdoor air (OA),
The weight for each rotational speed is determined to include at least one of the following factors: an increase in air leakage rate due to an increase in the rotational speed of the rotary heat exchanger, an increase in driving force of the rotary heat exchanger, an increase in wear of the mechanism, and an increase in failure rate of the mechanism. A deep learning-based energy-saving air conditioner characterized by: delete delete In paragraph 1:
The weight for each rotational speed is determined to become smaller as the rotational speed of the rotary heat exchanger unit increases, but is determined differently depending on the operation mode of the air conditioner, which is either a cooling operation or a heating operation. based energy-saving air conditioner. In paragraph 1 or 4:
During operation of the air conditioner, the control module collects the indoor and outdoor air temperature and humidity information by changing the rotational speed of the rotary heat exchanger unit with respect to the current temperature and humidity conditions of the outdoor air (OA) according to a predetermined method, and collects the indoor and outdoor air temperature and humidity information in the memory unit. Further performing a storage temperature and humidity information collection mode, and controlling the operations of the thermal efficiency calculation unit, the learning data generation unit and the deep learning unit so that the deep learning unit re-learns using the newly stored indoor and outdoor air temperature and humidity information according to a predetermined cycle. A deep learning-based energy-saving air conditioner characterized by: In the control method of a deep learning-based energy-saving air conditioner including an outdoor unit and an air conditioner (AHU) that cools or heats the indoor space by exchanging heat between the heat medium supplied from the outdoor unit and indoor air,
(a) measuring temperature and humidity information of outdoor air (OA) before heat exchange in a rotary heat exchanger that exchanges heat between indoor air exhausted to the outside of the air conditioner and outdoor air flowing into the air conditioner;
(b) determining the rotation speed of the rotary total heat exchanger using a machine-learned deep learning module to input the measured temperature and humidity information of the outdoor air (OA) and output the rotation speed of the rotary total heat exchanger. ;
(c) rotating the rotary total heat exchanger at the determined rotation speed; and
(d) including the step of machine learning the deep learning module,
In step (d),
(d-1) Temperature and humidity information for the outdoor air (SA) after the total heat exchange and the indoor air (RA) before the total heat exchange according to the rotation speed of the rotary total heat exchanger for the temperature and humidity conditions of the outdoor air (OA) Measuring indoor and outdoor air temperature and humidity information;
(d-2) generating operation data by calculating the heat exchange efficiency at each rotational speed of the rotary heat exchanger for the temperature and humidity conditions of the outdoor air (OA) using the indoor and outdoor air temperature and humidity information;
(d-3) generating learning data for deep learning by correcting the heat exchange efficiency of the operation data by applying a predetermined weight for each rotation speed; and
(d-4) using the learning data to perform machine learning to output the rotational speed with the highest heat exchange efficiency for each temperature and humidity condition of outdoor air (OA),
The weight for each rotational speed is determined to include at least one of the following factors: an increase in air leakage rate due to an increase in the rotational speed of the rotary heat exchanger, an increase in driving force of the rotary heat exchanger, an increase in wear of the mechanism, and an increase in failure rate of the mechanism. A control method for an energy-saving air conditioner based on deep learning, characterized by: delete In paragraph 6:
(e) collecting the indoor and outdoor air temperature and humidity information while changing the rotational speed of the rotary heat exchanger with respect to the current temperature and humidity conditions of outdoor air (OA) according to a predetermined method while operating the air conditioner; and
(f) a deep learning-based energy-saving air conditioner further comprising the step of retraining the deep learning module using the indoor and outdoor air temperature and humidity information collected in step (e) according to a predetermined cycle. control method. delete In paragraph 6 or 8:
The weight for each rotational speed is determined to be smaller as the rotational speed of the rotary heat exchanger unit increases, but is determined differently depending on the operation mode of the air conditioner, which is either a cooling operation or a heating operation. Control method of an energy-saving air conditioner based on