JP7261104B2 - Condensation prediction system and program - Google Patents

Description

The present invention relates to a dew condensation prediction system and program for predicting dew condensation on windows.

Japanese Patent Application Laid-Open No. 2015-31415 (Patent Document 1) discloses that an air conditioner includes an indoor unit installed in a room facing a window, a wind direction adjustment unit that adjusts the wind direction of air blown out from the indoor unit into the room, It is attached to a window and includes a temperature sensor that detects the surface temperature of the interior side of the window, a humidity sensor that detects the humidity in the room, and a recognition unit that recognizes the position of the window. It discloses controlling the indoor unit and the wind direction adjusting section so that the indoor unit blows air toward the window based on the position of the window recognized by the recognition section when it is determined that it will occur.

JP 2015-31415 A

In Patent Document 1, the presence or absence of dew condensation on the window is determined based on the current window surface temperature and indoor humidity. Therefore, even if dew condensation can be suppressed, it is difficult to prevent dew condensation. Also, attaching a temperature sensor to the surface of the window as in Patent Document 1 is not realistic considering the difficulty of securing a power supply and the fact that windows are susceptible to cold drafts and solar radiation.

SUMMARY OF THE INVENTION The present invention has been made to solve the problems described above, and an object thereof is to provide a dew condensation prediction system and program suitable for preventing dew condensation on windows.

A dew condensation prediction system according to one aspect of the present invention is a dew condensation prediction system for predicting condensation on a window, and includes a model storage unit, detection means, prediction means, and risk determination means. The model storage unit is a prediction model generated by machine-learning the indoor temperature, indoor absolute humidity, and outdoor temperature for each hour, and the indoor temperature, indoor absolute humidity, and outdoor temperature for a predetermined period of time. memorize The detection means detects indoor temperature and indoor absolute humidity. The prediction means inputs the indoor temperature and indoor absolute humidity detected by the detection means, the outdoor temperature at that time, and the predicted outdoor temperature after a predetermined period into the prediction model, and calculates the indoor temperature and indoor temperature after the predetermined period. predict the absolute humidity of The risk determination means estimates the window surface temperature based on the indoor temperature predicted by the prediction means, the predicted outdoor temperature, and the virtual heat transmission coefficient of the window, and uses the estimated window surface temperature to prevent dew condensation on the window. Determine the presence or absence of risk.

Preferably, the risk determination means estimates the surface temperature of the window using the temperature near the window obtained by correcting the room temperature predicted by the prediction means with the installation height information of the window.

Preferably, the installation height information is information indicating the height from the floor or ceiling surface to the lower edge of the window, and the risk determination means estimates the surface temperature of the lower edge of the window.

The risk determination means calculates the dew point temperature near the window based on the indoor relative humidity obtained from the indoor temperature and absolute humidity predicted by the prediction means and the temperature near the window, and estimates the calculated dew point temperature near the window. It is desirable to determine whether there is a risk of dew condensation by comparing the measured surface temperature of the window.

More preferably, the virtual heat transmission coefficient is determined for each material of the window frame, and the risk determination means estimates the surface temperature of the window frame.

The virtual heat transmission coefficient may be determined according to the presence or absence of a curtain covering the surface of the window.

Preferably, the predetermined period is 24 hours, and the prediction model is learned and generated on a monthly basis in a year.

A dew condensation prediction program according to an aspect of the present invention is a dew condensation prediction program for predicting condensation on a window, and includes an hourly indoor temperature, an indoor absolute humidity, and an outdoor air temperature, and an indoor temperature before a predetermined period of each hour. a step of reading a prediction model generated by machine learning of temperature, indoor absolute humidity, and outdoor temperature from a storage unit; indoor temperature and indoor absolute humidity detected by the detection means, and outdoor temperature at that time; and the predicted outside air temperature after a predetermined period of time into a prediction model to predict the indoor temperature and indoor absolute humidity after the predetermined period of time; estimating the surface temperature of the window based on the through-flow rate, and using the estimated surface temperature of the window to determine whether or not there is a risk of dew condensation on the window.

ADVANTAGE OF THE INVENTION According to this invention, the dew condensation of a window can be prevented.

1A is a block diagram showing a functional configuration of a dew condensation prediction system according to an embodiment of the present invention, and FIG. 1B is a block diagram showing a functional configuration of a learning device; FIG. (A) is a diagram schematically showing a room targeted for humidity adjustment by a dew condensation prediction system, and (B) is a diagram schematically showing a room targeted for learning by a learning device. It is a figure for demonstrating the prediction accuracy of the prediction model in embodiment of this invention. (A) and (B) are diagrams for explaining a method of determining the risk of condensation in the embodiment of the present invention. (A) and (B) are diagrams showing an overview of a test method for measuring surface temperatures at multiple locations on a window and test results, respectively. 4 is a flow chart showing the flow of processing executed by the controller of the dew condensation prediction system according to the embodiment of the present invention; 7 is a graph schematically showing an example of changes in estimated values of the window surface temperature and the dew point temperature near the window for one day. FIG. 4 is a diagram schematically showing an example of a screen displayed on the user terminal in the embodiment of the present invention; FIG.

An embodiment of the present invention will be described in detail with reference to the drawings. The same or corresponding parts in the drawings are denoted by the same reference numerals, and the description thereof will not be repeated.

<Configuration example of dew condensation prediction system>
FIG. 1A is a block diagram showing the functional configuration of a dew condensation prediction system 1 according to this embodiment. FIG. 2A is a diagram schematically showing a room 9 to which the dew condensation prediction system 1 is applied. The room 9 is typically a living room (living room, bedroom, etc.) of a house, and a window 90 is provided on the outer wall of the room 9 . Also, an air conditioner 91 is installed in the room 9 . The air conditioner 91 is an air conditioner that adjusts the temperature of the air inside the room 9 . In addition, the building in which the room 9 is provided may be other than a residence.

As shown in FIG. 2A, the dew condensation prediction system 1 includes a control device 10, a temperature sensor 71 and a humidity sensor 72 provided inside the room 9, and an outside air temperature sensor 73 provided outdoors. ing.

A control device 10 is a computer including a memory and a processor, and is connected to a network 11 such as the Internet. A temperature sensor 71 detects the temperature (room temperature) in the room 9 . Humidity sensor 72 detects the absolute humidity in room 9 . The outside air temperature sensor 73 detects outside air temperature.

Referring to FIG. 1A, a dew condensation prediction system 1 is a system for predicting dew condensation on a window 90, and functions as a model storage unit 2, a prediction unit 3, a risk determination unit 4, and an output unit 5 . These units are mounted on the control device 10 . Note that the model storage unit 2 may be provided on a server that can be accessed by the control device 10 .

The model storage unit 2 preliminarily stores a prediction model 20 for predicting the indoor temperature and indoor absolute humidity of the next day (that is, after 24 hours). This prediction model 20 is generated in advance by machine learning of the air environment data of room 109 (corresponding to room 9) shown in FIG. 2B by the learning device 100 shown in FIG. be. A learning method by the learning device 100 will be described later.

The prediction unit 3 inputs the current indoor temperature, absolute humidity, and outdoor temperature obtained from the sensors 71 to 73, respectively, and the predicted outdoor temperature (weather forecast) for the next day obtained via the network 11 to the prediction model 20. to predict the indoor temperature and absolute humidity for the next day. In other words, the indoor temperature and absolute humidity are predicted using the air environment history of one day before the prediction target date.

The actual measurement data of the indoor temperature, absolute humidity, and outdoor temperature of the day are temporarily stored in the data storage unit 30 in chronological order (for each time). When predicting unit 3 predicts the indoor temperature and absolute humidity for the next day, actual measurement data for one day is read from data storage unit 30 . The prediction unit 3 inputs the hourly measured value read from the data storage unit 30 and the hourly forecast outside air temperature (prediction value) for the next day into the prediction model 20 to obtain the hourly indoor temperature for the next day. and predict (calculate) absolute humidity. From this, the hourly transitions of the room temperature and absolute humidity for the next day are predicted. Indoor temperature and absolute humidity are predicted on an hourly basis, for example.

In the present embodiment, the outside air temperature on the day is actually measured, but the outside air temperature on the day may also be obtained from the network 11 . In this case, the outside air temperature sensor 73 does not have to be included in the dew condensation prediction system 1 .

The risk determination unit 4 uses the indoor temperature and indoor absolute humidity predicted by the prediction unit 3 to determine whether there is a risk of condensation on the window 90 (hereinafter referred to as “risk of condensation”). The risk of condensation on window 90 can be determined by comparing the surface temperature of window 90 and the dew point temperature in the room (preferably the dew point temperature near window 90). Therefore, the risk determination unit 4 determines the risk of dew condensation on the window 90 by estimating the surface temperature of the window 90 and the indoor dew point temperature.

A method of determining the risk of condensation on the window 90 will be described with reference to FIG. FIG. 4A schematically shows a longitudinal section of the room 9 in which the window 90 is provided. FIG. 4B conceptually shows information used to determine the risk of condensation.

The risk determination unit 4 estimates the surface temperature TA of the window 90 based on the indoor temperature Ti, the outside air temperature To, and the virtual heat transmission coefficient (U value) of the window 90 predicted by the prediction unit 3 . The virtual heat transmission coefficient (U value) of the window 90 is a value predetermined by experiments. The heat transmission coefficient of the window is represented by "U = α × (Ti-TA) / (Ti-To)" when the surface heat transfer coefficient on the outdoor side is constant α (= 8.6). By keeping the outside air temperature (To) and the room temperature (Ti) constant in the experiment and measuring the window surface temperature (TA) at that time, the virtual heat transmission coefficient (U) can be calculated. By storing the virtual heat transmission coefficient (U) of the window 90 obtained by experiment in advance in memory, the surface temperature TA of the window 90 can be calculated if the outside air temperature To and the room temperature Ti are known.

Here, the lower the surface temperature of the window 90 is, the more likely it is that dew condensation will occur. It has been found from an actual measurement test that reproduces various outside air/indoor conditions that the surface temperature of the window 90 is lower at the lower end 90a than at the upper end 90b and the central portion 90c. FIGS. 5A and 5B are diagrams showing an overview of a test method for measuring surface temperatures at multiple locations on a window and test results, respectively. As shown in FIG. 5(A), windows (sweep-out windows), which are specimens, are placed on the boundary walls of constant temperature and humidity chambers <1> and <2>.

In the constant temperature and humidity chamber <1>, an outside air condition with a temperature of −12.6° C. was reproduced, and in the constant temperature and humidity chamber <2>, an indoor condition with a temperature of 20° C. and a relative humidity of 40 to 50% was reproduced. In addition, the inside of the room was adjusted to a negative pressure of 0 to 5 Pa by a ventilator. FIG. 5B shows the temperatures measured at a plurality of locations under such conditions, and it can be seen that the surface temperature of the window is lower at the lower end than at the upper end. That is, the closer the lower end portion 90a of the window 90 is to the floor FL, the lower the surface temperature of the window 90 becomes.

Therefore, the risk determination unit 4 in the present embodiment estimates the surface temperature TA of the window 90 using the window vicinity temperature Tw obtained by correcting the predicted indoor temperature Ti with the installation height information of the window 90 . The installation height information is information indicating the height from the floor FL of the room 9 to the lower end portion 90 a of the window 90 . Therefore, the window vicinity temperature Tw represents the temperature near the height of the lower end of the window 90 . Since the housing design information includes information such as the length and width dimensions of windows in each room, the height of window lintels, and the type of window frame (sash) (resin, aluminum, etc.), installation of windows 90 is difficult. Height information can be obtained from housing design information.

Specifically, by subtracting the lintel height L12 of the window 90 and the height dimension L13 of the window 90 from the ceiling height L11 of the room 9, the height L1 of the lower end portion 90a of the window 90 can be calculated. The risk determination unit 4 corrects the room temperature Ti such that the closer the height L1 of the lower end portion 90a of the window 90 to the floor FL, the lower the window vicinity temperature Tw. For example, assuming that the ceiling height L11 of the room 9 is 2400 mm, if the height L1 of the lower end portion 90a of the window 90 is 1200 mm from the floor FL (center height of the room 9), the correction value is set to "0°C ” (no correction), the correction value can be set to “−1.5° C.” if the distance is 0 mm from the floor FL, and the correction value can be set to “+1.5° C.” if the distance is 2300 mm from the floor FL. Note that this correction value may be determined according to the installation height of the temperature sensor 71 .

The test results in FIG. 5B further indicate that the surface temperature of the window is lower in the window frame in which the window body is fitted than in the window body. Therefore, the surface temperature TA of the lower end portion 90a of the window 90 calculated by the risk determination section 4 is preferably the surface temperature of the lower end portion of the window frame 90d. In this case, the virtual heat transmission coefficient described above is the virtual heat transmission coefficient of the window frame 90d, and is determined for each material of the window frame 90d.

The risk determination unit 4 also calculates the indoor dew point temperature TB to be compared with the surface temperature TA of the window 90 . The indoor dew point temperature TB is also desirably calculated using the window vicinity temperature Tw obtained by correcting the indoor temperature Ti with the window height information. That is, the dew point temperature TB to be compared with the surface temperature TA of the window 90 is desirably the dew point temperature near the window 90 .

Therefore, the risk determination unit 4 in the present embodiment applies the window vicinity temperature Tw and the relative humidity Hi of the room 9 to a general formula to estimate (calculate) the dew point temperature TB in the vicinity of the window 90 . As a result, when the surface temperature TA of the window 90 exceeds the dew point temperature TB near the window 90, the risk determination unit 4 determines that there is a risk of condensation occurring at the lower end of the window frame 90d where condensation is most likely to occur. It can be determined (predicted).

It is known that the surface temperature of the window 90 is relatively low when the surface of the window 90 is covered with a curtain 92 (FIG. 4A). Therefore, the virtual heat transmission coefficient used to calculate the surface temperature TA of the window 90 may be determined to be higher with the curtain 92 than without the curtain 92 .

The output unit 5 outputs the result of determination by the risk determination unit 4 . The output unit 5 is implemented by, for example, a communication I/F (interface) that transmits the determination result to the user terminal 8 (FIG. 2(A)). As the user terminal 8, for example, a mobile phone such as a smart phone or a tablet terminal is assumed. Alternatively, the output section 5 may be realized by a display section such as an LCD (Liquid Crystal Display).

The functions of prediction unit 3 and risk determination unit 4 included in control device 10 are typically realized by software. Note that at least one of these may be implemented by hardware.

<Configuration example of learning device>
FIG. 1B is a block diagram showing the functional configuration of learning device 100 according to the present embodiment. FIG. 2B is a diagram schematically showing a room 109 to be learned by the learning device 100. As shown in FIG. A room 109 is a room imitating the room 9 of the dew condensation prediction system 1 . Note that room 109 may be the same as room 9 .

Referring to FIG. 1B, learning device 100 includes a model generation unit 101, a data storage unit 102, and a model storage unit 103 as its functions.

The data storage unit 102 stores and accumulates the indoor temperature, absolute humidity, and outdoor air temperature obtained from the sensors 71 to 73 shown in FIG. 2B in time series (for each time). The data storage unit 102 also stores (accumulates) the usage status of the air conditioner 91 installed in the room 109 in chronological order. The usage status of the air conditioner 91 includes ON/OFF information and set temperature information of the air conditioner 91 . Note that the outside air temperature may be obtained via the network 11 in the learning device 100 as well.

Model generator 101 generates predictive model 130 based on the air environment data accumulated in data storage 102 . Specifically, the prediction model 130 is generated by performing machine learning by associating the hourly indoor temperature, absolute humidity, outdoor temperature, and usage status of the air conditioner 91 with the previous day's indoor temperature, absolute humidity, and outdoor temperature. do. As a result, it is possible to predict the indoor temperature and absolute humidity after 24 hours by using the measured indoor temperature, absolute humidity, and outdoor temperature at a certain time (time) and the predicted outdoor temperature after 24 hours as keys. , a prediction model 130 is generated. The usage status of the air conditioner 91 is a parameter used to improve the accuracy of the prediction model 130, and indirectly contributes to the prediction of the indoor temperature and indoor absolute humidity.

The model generator 101 updates the prediction model 130 by performing machine learning on these air environment data over multiple days (preferably one year or more). Machine learning is performed, for example, by multiple regression analysis. The prediction model 130 generated in this manner is stored in the model storage unit 2 of the dew condensation prediction system 1 as the prediction model 20 . The prediction model 130 is desirably generated by learning on a monthly basis in a year.

The prediction accuracy by the prediction unit 3 using the prediction model 130 (20) will be described with reference to FIG. FIG. 3 shows indoor temperature prediction conditions by the prediction model 130 and a comparative example thereof. "Case 1" and "Case 3" shown in FIG. 3A are comparative examples, and "Case 2" corresponds to the embodiment of the present application.

"Case 1" shows prediction conditions for predicting the indoor temperature without using past history data of indoor temperature and outdoor temperature. In this case, the parameters of the prediction model for predicting the indoor temperature are only the month/hour, the outside air temperature at that time, and the use of the air conditioner.

"Case 2" shows the prediction conditions for predicting the indoor temperature using the history data of the indoor temperature and the outdoor temperature from the previous day. In this case, the parameters of the predictive model for predicting the indoor temperature are the month/hour, the outdoor temperature at that time and the usage of the air conditioner, and the indoor temperature and outdoor temperature at the same time the day before.

"Case 3" shows the prediction conditions for predicting the room temperature using history data from 10 minutes ago. In this case, the parameters of the predictive model for predicting the indoor temperature are the month/hour, the outdoor temperature at that time and the usage of the air conditioner, and the indoor and outdoor temperatures 10 minutes ago.

As shown in FIG. 3B, Case 1 has a coefficient of determination of 0.1 (temperature error: about ±3° C.), and prediction accuracy is extremely low when past history data is not used. On the other hand, the coefficient of determination of Case 2 is 0.48 (temperature error: about ±1.5° C.), so it can be seen that the prediction accuracy is improved compared to Case 1.

The coefficient of determination of Case 3 is 0.98 (temperature error: about ±0.3°C), and the prediction accuracy is very high. Since the determination is made, dew condensation countermeasures cannot be taken efficiently.

Regarding the prediction accuracy of the indoor absolute humidity, the same coefficient of determination (0.4 to 0.6) as in Case 2 was shown when prediction was made using the history data of the absolute humidity and outside air temperature from the day before. Therefore, by following the prediction model 130 (20), it is possible to roughly predict the indoor temperature and absolute humidity after 24 hours. Thus, by roughly predicting the indoor temperature and absolute humidity after 24 hours, the risk of condensation after 24 hours can be determined (predicted). Countermeasures can be taken.

Note that parameters for generating the prediction model 130 (20) are parameters other than the above ( air environment data). Alternatively, the parameters for generating the prediction model 20 (130) may not include the usage status of the air conditioner 91.

Also, in the present embodiment, the common prediction model 20 is used to predict the room temperature and the absolute humidity, but this is not restrictive. That is, the prediction models stored in the model storage unit 2 may be composed of a prediction model for room temperature and a prediction model for absolute humidity. The prediction model for the room temperature is generated by machine learning the room temperature, the outside air temperature, the usage status of the air conditioner 91, and the room temperature and the outside air temperature 24 hours before each hour. The prediction model for absolute humidity is generated by machine-learning the absolute humidity, the outside air temperature, the usage status of the air conditioner 91 for each hour, and the absolute humidity and the outside air temperature 24 hours before each hour.

<Operation of the condensation prediction system>
The operation of the dew condensation prediction system 1 according to this embodiment will be described. FIG. 6 is a flow chart showing the flow of processing executed by the control device 10 of the dew condensation prediction system 1. As shown in FIG. The processing shown in FIG. 6 is started by the processor of the control device 10 reading out and executing a program pre-stored in the memory at the same time every day, such as midnight. In parallel with the processing shown in FIG. 4, the control device 10 stores the measured values detected by the sensors 71 to 73, that is, the indoor temperature, the absolute humidity, and the outdoor temperature in the data storage unit 30 in chronological order. It is assumed that the process to

Referring to FIG. 6, first, prediction unit 3 of control device 10 reads prediction model 20 stored in model storage unit 2 (step S1). Also, the measured values for the previous day stored in the data storage unit 30 are read out (step S2). Specifically, the indoor temperature, absolute humidity, and outdoor temperature for each hour of the previous day are read.

Furthermore, the prediction unit 3 acquires the predicted outside air temperature for today through the network 11 (step S3). That is, the hourly outside air temperature included in the weather forecast information is acquired. The order of steps S1 to S3 is not particularly limited.

Subsequently, the prediction unit 3 predicts the indoor temperature and absolute humidity for today's day based on yesterday's actual measurement values read in step S2 and today's forecast outside air temperature obtained in step S3 (step S5 ). That is, by inputting yesterday's actual measured value, today's forecast outside temperature, and today's month (for example, March) into the prediction model 20 read out in step S1, today's indoor temperature prediction value and today's An estimate of absolute humidity is obtained. These predicted values are obtained, for example, every hour.

When the indoor temperature and absolute humidity are predicted, the risk determination unit 4 uses these values to calculate the indoor relative humidity (step S7). Thereby, the relative humidity of the room 9 is calculated (estimated) for each hour.

In addition, the risk determination unit 4 determines the temperature near the window (Tw) obtained by correcting the indoor temperature predicted in step S5 with the installation height information, the forecast outside temperature (To) obtained from the weather forecast via the network 11, and A window surface temperature TA is calculated based on a virtual heat transmission coefficient (U) determined according to the material of the window frame 90d (step S9). That is, each value is applied to the following equation to calculate the window surface temperature TA, that is, the surface temperature of the lower end portion of the window frame 90d.
TA=Tw−U(Tw−To)/α

Furthermore, the risk determination unit 4 applies the window vicinity temperature (Tw) and the relative humidity (Hi) calculated in step S7 to a general formula, and determines the dew point temperature near the lower end of the window frame 90d, that is, the window vicinity dew point temperature. TB is calculated (step S11). Note that the window vicinity dew point temperature TB may be obtained based on the window vicinity temperature Tw and the absolute humidity. In this case, the process of step S7 may be omitted.

Thus, transitions of the window surface temperature TA and the window vicinity dew point temperature TB for today are estimated. The graph of FIG. 7 shows an example of transition of the window surface temperature TA and the window vicinity dew point temperature TB in one day.

The risk determination unit 4 compares the window surface temperature TA and the window vicinity dew point temperature TB in units of time to determine (predict) whether or not there is a risk of dew condensation (step S13). In the example of FIG. 7, the window vicinity dew point temperature TB exceeds the window surface temperature TA in the time period from 16:00 to 22:00. Therefore, in this example, it is determined that there is a risk of condensation during the time period from 16:00 to 22:00. That is, it is predicted that condensation will occur from 16:00 to 22:00.

When the determination (prediction) by the risk determination unit 4 is completed, the output unit 5 outputs the prediction result to the user terminal 8, for example (step S15). FIG. 8 schematically shows an example of screen display on the user terminal 8. As shown in FIG. In the example of FIG. 8, the display unit 80 of the user terminal 8 displays that there is a risk of condensation, and displays an advice button 81 and an action button 82 . When the advice button 81 is selected, a pop-up screen 83 displays a message such as "Condensation will occur between 16:00 and 22:00. Let's start the heating operation." Note that a graph such as that shown in FIG. 7 may be displayed as the prediction result.

As described above, according to the present embodiment, it is possible to estimate transitions of the window surface temperature TA and the window vicinity temperature TB on the next day based on the measured values of the air environment data for one day. Therefore, even if a temperature sensor is not attached to the surface of the window 90, it is possible to determine (predict) whether or not there is a risk of dew condensation on the window 90. FIG. Moreover, in the present embodiment, the surface temperature of the portion of the window 90 that is most likely to get cold (the lower end portion of the window frame 90d) is estimated, so the risk of dew condensation can be predicted with high accuracy.

In addition, since the risk determination unit 4 determines (predicts) the presence or absence of condensation risk not on a daily basis, but on an hourly basis (typically, on an hourly basis), a period of time during the day when countermeasures are required is determined. User can be notified. In addition, by outputting a graph such as that shown in FIG. 7, it is possible to visualize the time zone in which countermeasures are required. Therefore, the user can systematically take dew condensation countermeasures, such as starting the heating operation before the time when there is a risk of dew condensation. As a result, dew condensation can be efficiently prevented.

(Modification)
In the present embodiment, the prediction model 20 (130) is generated using history data from one day ago, but this is not restrictive. That is, the period (predetermined period) that goes back from the time when the air environment data was measured is not limited to 24 hours. For example, the prediction model 20 (130) may be generated using history data from half a day ago so that the above-mentioned coefficient of determination is 0.5 or more (preferably 0.6 or more). In this case, the prediction unit 3 predicts the indoor temperature and absolute humidity half a day later.

Further, the model storage unit 2 of the condensation prediction system 1 according to the present embodiment stores the prediction model 20 (130) generated in advance by the learning device 100. A device 100 may be provided. That is, the dew condensation prediction system 1 may update the prediction model 20 stored in the model storage unit 2 by repeating the actual measurement and prediction of the room temperature and the absolute humidity. In this case, it is possible to improve the prediction accuracy of the prediction model 20 as the system is used.

Note that the dew condensation prediction method described above can also be provided as a program. Such a program can be provided by being recorded on an optical medium such as a CD-ROM or a computer-readable non-transitory recording medium such as a memory card. In this case, the control device 10 includes a drive device (not shown) capable of reading/writing programs and data from a recording medium (not shown). Also, the program can be provided by downloading via the network 11 .

It should be considered that the embodiments disclosed this time are illustrative in all respects and not restrictive. The scope of the present invention is indicated by the scope of the claims rather than the above description, and is intended to include all modifications within the meaning and range of equivalents of the scope of the claims.

1 dew condensation prediction system, 2,103 model storage unit, 3 prediction unit, 4 risk determination unit, 5 output unit, 8 user terminal, 9,109 room, 10 control device, 11 network, 20,130 prediction model, 71 temperature sensor , 72 humidity sensor, 73 outside air temperature sensor, 90 window, 90d window frame, 100 learning device, 101 model generating unit.

Claims (8)

A condensation prediction system for predicting condensation on a window,
A model memory that stores a prediction model generated by machine learning of the indoor temperature, indoor absolute humidity, and outdoor temperature for each hour and the indoor temperature, indoor absolute humidity, and outdoor temperature for a predetermined period of time. Department and
detection means for detecting indoor temperature and indoor absolute humidity;
The indoor temperature and indoor absolute humidity detected by the detection means, the outdoor temperature at that time, and the predicted outdoor temperature after the predetermined period are input to the prediction model, and the indoor temperature and indoor temperature after the predetermined period are input to the prediction model. a prediction means for predicting the absolute humidity of
Estimate the surface temperature of the window based on the indoor temperature predicted by the prediction means, the predicted outdoor temperature, and the virtual heat transmission coefficient of the window, and use the estimated surface temperature of the window to determine the presence or absence of the risk of condensation on the window. A dew condensation prediction system, comprising a risk determination means for determining. 2. The dew condensation prediction system according to claim 1, wherein the risk determination means estimates the surface temperature of the window using the temperature near the window obtained by correcting the room temperature predicted by the prediction means with information on the installation height of the window. The installation height information is information indicating the height from the floor surface or ceiling surface to the lower end of the window,
3. The dew condensation prediction system according to claim 2, wherein said risk determination means estimates the surface temperature of the lower edge of the window. The risk determination means calculates the dew point temperature near the window based on the indoor relative humidity obtained from the indoor temperature and absolute humidity predicted by the prediction means and the window near temperature, and calculates the calculated dew point near the window. 4. The dew condensation prediction system according to claim 2, wherein the presence or absence of dew condensation risk is determined by comparing the temperature with the estimated surface temperature of the window. The virtual heat transmission coefficient is determined for each material of the window frame,
The dew condensation prediction system according to any one of claims 1 to 4, wherein said risk determination means estimates a surface temperature of a window frame. The dew condensation prediction system according to any one of claims 1 to 5, wherein said virtual heat transmission coefficient is determined according to the presence or absence of a curtain covering the window surface. the predetermined period of time is 24 hours;
7. The dew condensation prediction system according to any one of claims 1 to 6, wherein said prediction model is generated by learning on a monthly basis in a year. A condensation prediction program for predicting condensation on a window,
A prediction model generated by machine learning of the indoor temperature, indoor absolute humidity, and outdoor temperature for each hour and the indoor temperature, indoor absolute humidity, and outdoor temperature for a predetermined period of time is stored from the storage unit. a step of reading;
The indoor temperature and indoor absolute humidity detected by the detection means, the outdoor temperature at that time, and the predicted outdoor temperature after the predetermined period are input to the prediction model, and the indoor temperature after the predetermined period and the indoor temperature are calculated. predicting absolute humidity;
estimating the surface temperature of the window based on the predicted indoor temperature, the predicted outdoor temperature, and the virtual heat transmission coefficient of the window, and using the estimated window surface temperature to determine whether there is a risk of dew condensation on the window; A dew condensation prediction program that causes a computer to execute

Images