WO2025052607A1 - Ventilation system, learning device, and inference device - Google Patents

Abstract

This ventilation system is provided with: a ventilation device installed in a building; and a ventilation control device (2) for controlling the ventilation device on the basis of weather prediction data. The ventilation control device (2) has: a rainwater entry possibility determination unit (21) for determining whether or not there is a possibility that rainwater may enter the ventilation device on the basis of the weather prediction data; and an air supply amount management unit (22) for managing an air supply amount of the ventilation device by stopping an air supply with the ventilation device or reducing the air supply amount with the ventilation device when there is a possibility that rainwater may enter the ventilation device.

Description

Ventilation system, learning device and inference device

This disclosure relates to a ventilation system that includes a ventilation device installed in a building and a ventilation control device that controls the ventilation device, as well as a learning device and an inference device used therein.

In a ventilation system that includes a ventilation device installed in a building and a ventilation control device that controls the ventilation device, it is necessary to prevent rainwater and other substances from entering the room.

Patent Document 1 discloses a ventilation device that stops ventilation when the measured outdoor humidity is higher than the indoor humidity. The ventilation device disclosed in Patent Document 1 stops ventilation when rainfall causes the outdoor humidity to become higher than the indoor humidity, thereby preventing rainwater from entering during rainfall.

Japanese Patent Application Publication No. 5-280774

The ventilation device disclosed in Patent Document 1 has a problem in that when rain starts to fall suddenly due to a sudden downpour, the humidity sensor does not keep up with the sudden increase in outdoor humidity, resulting in a delay in stopping the air supply fan and allowing rainwater to seep in.

The present disclosure has been made in consideration of the above, and aims to provide a ventilation system that prevents rainwater from entering the ventilation device even if it suddenly starts raining.

In order to solve the above-mentioned problems and achieve the objectives, the ventilation system according to the present disclosure includes a ventilation device installed in a building, and a ventilation control device that controls the ventilation device based on weather forecast data. The ventilation control device has a rainwater intrusion possibility determination unit that determines whether or not there is a possibility of rainwater intrusion into the ventilation device based on weather forecast data, and an air supply volume management unit that manages the volume of air supply by the ventilation device by stopping air supply by the ventilation device or reducing the volume of air supply by the ventilation device when there is a possibility of rainwater intrusion into the ventilation device.

The ventilation system disclosed herein has the effect of preventing rainwater from entering the ventilation device even if it suddenly starts raining.

FIG. 1 is a diagram showing a configuration of a ventilation system according to a first embodiment. FIG. 1 is a diagram showing a configuration of a ventilation device of a ventilation system according to a first embodiment. FIG. 1 is a diagram showing a configuration of a ventilation control device of a ventilation system according to a first embodiment. A flowchart showing the flow of operations of the ventilation system according to the first embodiment. A flowchart showing the flow of operations of a ventilation system according to a second embodiment. FIG. 13 is a diagram showing an example of a rainwater intrusion determination table held by the ventilation control device according to the second embodiment; FIG. 13 is a diagram showing an example of installation of a ventilation device of a ventilation system according to a third embodiment. FIG. 13 is a diagram showing a configuration of a ventilation system according to a fifth embodiment. FIG. 13 is a diagram showing a configuration of a ventilation system according to a sixth embodiment. FIG. 13 is a diagram showing the configuration of a ventilation system according to a 9th embodiment. FIG. 13 is a diagram showing the configuration of a learning device for a ventilation system according to a 9th embodiment. A flowchart of a learning process of a learning device for a ventilation system according to a ninth embodiment FIG. 13 is a diagram showing a configuration of an inference device for a ventilation system according to a ninth embodiment. A flowchart of an inference process of an inference device of a ventilation system according to a ninth embodiment FIG. 13 is a diagram showing a configuration of a ventilation system according to a tenth embodiment. FIG. 13 is a diagram showing a hardware configuration of a learning device according to a ninth embodiment and a tenth embodiment.

The ventilation system, learning device, and inference device according to the embodiment are described in detail below with reference to the drawings.

Embodiment 1.
Fig. 1 is a diagram showing the configuration of a ventilation system according to embodiment 1. The ventilation system 100 includes a ventilation device 1 installed in a building, and a ventilation control device 2 that controls the ventilation device 1 based on weather forecast data.

FIG. 2 is a diagram showing the configuration of the ventilation device of the ventilation system according to the first embodiment. The ventilation device 1 includes a ventilation blower 11, a speed regulator 12 that adjusts the air volume of the ventilation blower 11, and a damper 13 that opens and closes the ventilation port. The ventilation device 1 may also include a ventilation hood and a filter that is attached inside the ventilation duct. The speed regulator 12 stops the operation of the ventilation blower 11 and adjusts its speed upon receiving an instruction from the ventilation control device 2. The damper 13 adjusts the opening degree of the ventilation port upon receiving an instruction from the ventilation control device 2. The ventilation device 1 is, for example, a propeller fan type pressure ventilation fan installed on the wall of a building. There is a possibility that rainwater may enter the pressure ventilation fan through the wind tunnel of the propeller fan that connects the inside and outside of the building.

FIG. 3 is a diagram showing the configuration of the ventilation control device of the ventilation system according to the first embodiment. The ventilation control device 2 includes a rainwater intrusion possibility determination unit 21 and an air supply volume management unit 22. The rainwater intrusion possibility determination unit 21 determines whether or not there is a possibility of rainwater intrusion into the ventilation device 1 based on weather forecast data and ventilation control conditions set in advance. When there is a possibility of rainwater intrusion into the ventilation device 1, the air supply volume management unit 22 outputs a control instruction to the ventilation device 1 to stop the ventilation device 1 or to reduce the air supply volume. When there is no longer a possibility of rainwater intrusion into the ventilation device 1, the air supply volume management unit 22 outputs a control instruction to the ventilation device 1 to resume operation of the ventilation device 1 or to return the air supply volume to the original air supply volume. In this way, the air supply volume management unit 22 manages the air supply volume by instructing the ventilation device 1 to operate, stop, and increase or decrease the air supply volume.

The weather forecast data may be forecast result data based on current and past weather measurement data obtained by various weather sensors such as a temperature sensor, a humidity sensor, and anemometer installed at the installation site of the ventilation device 1, or it may be weather forecast data obtained from outside based on regional information for the installation site of the ventilation device 1 released by a meteorological agency or a weather information provider.

The weather forecast data includes at least the amount of precipitation. In addition to the amount of precipitation, the weather forecast data may also include one or more of the following: temperature, humidity, air pressure, solar radiation, wind direction, wind speed, etc. The forecast value of the weather forecast data is associated with the forecast time.

The ventilation control device 2 may be configured as an integral part of the ventilation device 1, or may be installed in a location separate from the ventilation device 1.

Next, the operation of the ventilation system 100 will be described. FIG. 4 is a flowchart showing the flow of the operation of the ventilation system according to the first embodiment. In step S1, the rainwater intrusion possibility determination unit 21 acquires weather forecast data. In step S2, the rainwater intrusion possibility determination unit 21 determines whether the expected precipitation amount after X minutes is greater than a preset value. Here, the value of X can be set arbitrarily in advance, but if the value of X is too large, rain may start falling before the ventilation device 1 is controlled. For this reason, it is preferable to set X to about 5 minutes. If the expected precipitation amount after X minutes is equal to or less than the preset value, step S2 becomes No, and the process returns to step S1. If the expected precipitation amount after X minutes is greater than the preset value, step S2 becomes Yes, and in step S3, the supply air volume management unit 22 controls the ventilation device 1 so that the supply air volume after X minutes is equal to or less than the preset air volume.

In step S4, the rainwater intrusion possibility determination unit 21 acquires weather forecast data. In step S5, the rainwater intrusion possibility determination unit 21 determines whether the expected precipitation Y minutes later is smaller than a preset value. Here, the value of Y can be set in advance as desired, but if the value of Y is too large, the time for which the ventilation device 1 is stopped will be long even though the possibility of rainwater intrusion is low. For this reason, it is preferable to set Y to about 5 minutes. If the expected precipitation Y minutes later is equal to or greater than the preset value, step S5 becomes No and the process returns to step S4. If the expected precipitation Y minutes later is smaller than the preset value, step S5 becomes Yes and in step S6, the supply air volume management unit 22 controls the ventilation device 1 so that the supply air volume Y minutes later returns to the original air volume. Note that the "original air volume" is the air volume before the change in step S3.

The rainwater infiltration possibility determination unit 21 may acquire precipitation prediction data in step S1, and determine whether the predicted precipitation amount after X minutes is greater than a preset value in step S2 based on the precipitation prediction data. The rainwater infiltration possibility determination unit 21 may also acquire one or more weather forecast data of temperature, humidity, air pressure, solar radiation, wind direction, wind speed, etc. in step S1, and determine whether the predicted precipitation amount after X minutes is greater than a preset value in step S2 based on the acquired weather forecast data.

The rainwater infiltration possibility determination unit 21 may also acquire precipitation forecast data in step S4, and determine whether the expected precipitation after Y minutes is smaller than a preset value in step S5 based on the precipitation forecast data. The rainwater infiltration possibility determination unit 21 may also acquire one or more weather forecast data of temperature, humidity, air pressure, solar radiation, wind direction, wind speed, etc. in step S4, and determine whether the expected precipitation after Y minutes is smaller than a preset value in step S5 based on the acquired weather forecast data.

The control of the ventilation device 1 in steps S3 and S6 above is performed by adjusting the speed of the ventilation blower 11 and adjusting the opening of the damper 13. If multiple pieces of weather forecast data of the same type are acquired in step S1, the determination in step S3 of whether the expected precipitation amount in X minutes is greater than a preset value is made based on the most recent weather forecast data. If multiple pieces of weather forecast data of the same type are acquired in step S4, the determination in step S5 of whether the expected precipitation amount in Y minutes is less than a preset value is made based on the most recent weather forecast data.

The ventilation system 100 according to the first embodiment controls the ventilation device 1 so that the air supply volume after X minutes is equal to or less than the preset volume when the expected precipitation volume after X minutes is predicted based on weather forecast data to be greater than a preset value. This reduces the amount of rainwater entering the ventilation device or air supply duct, compared to a ventilation system that reduces the air supply volume after actually detecting an increase in precipitation.

In addition, when it is determined based on weather forecast data that the expected amount of precipitation Y minutes from now is less than a preset value, the ventilation system 100 according to embodiment 1 controls the ventilation device 1 so that the amount of air supplied Y minutes from now is equal to or greater than the preset air volume. This allows the amount of air supplied to be restored to its original level more quickly compared to a ventilation system that increases the amount of air supplied only after actually detecting a decrease in precipitation.

Embodiment 2.
The configuration of the ventilation system 100 according to the second embodiment is similar to that of the ventilation system 100 according to the first embodiment. In the ventilation system 100 according to the second embodiment, the rainwater infiltration possibility determination unit 21 is different from the rainwater infiltration possibility determination unit 21 of the ventilation system 100 according to the first embodiment in that the rainwater infiltration possibility determination unit 21 determines the possibility of rainwater infiltration based not only on the amount of precipitation but also on a combination of the amount of precipitation and wind speed.

FIG. 5 is a flowchart showing the flow of operation of the ventilation system according to the second embodiment. In step S11, the rainwater intrusion possibility determination unit 21 acquires weather forecast data. In step S12, the rainwater intrusion possibility determination unit 21 determines whether the predicted precipitation and predicted wind speed after X minutes fall within an area where there is a possibility of intrusion in the rainwater intrusion determination table. If the predicted precipitation and wind speed after X minutes do not fall within an area where there is a possibility of intrusion in the rainwater intrusion determination table, step S12 becomes No and the process returns to step S11. If the predicted precipitation and predicted wind speed after X minutes fall within an area where there is a possibility of intrusion in the rainwater intrusion determination table, step S12 becomes Yes, and in step S13, the supply air volume management unit 22 controls the ventilation device 1 so that the supply air volume after X minutes is equal to or less than a preset air volume.

In step S14, the rainwater intrusion possibility determination unit 21 acquires weather forecast data. In step S15, the rainwater intrusion possibility determination unit 21 determines whether the predicted precipitation and predicted wind speed after Y minutes will be outside the area where intrusion is possible in the rainwater intrusion determination table. If the precipitation and wind speed after Y minutes will not be outside the area where intrusion is possible in the rainwater intrusion determination table, step S15 becomes No and the process returns to step S14. If the precipitation and wind speed after Y minutes will be outside the area where intrusion is possible in the rainwater intrusion determination table, step S15 becomes Yes, and in step S16, the air supply volume management unit 22 controls the ventilation device so that the air supply volume after Y minutes returns to the original air volume. The "original air volume" is the air volume before the change in step S13.

FIG. 6 is a diagram showing an example of a rainwater intrusion determination table held by a ventilation control device according to embodiment 2. In the rainwater intrusion determination table, a first judgment reference precipitation and a second judgment reference precipitation are set for the amount of precipitation. The first judgment reference precipitation is greater than the second judgment reference precipitation. The rainwater intrusion determination table can be divided into six regions, from the first region to the sixth region. The first region is a region where the amount of precipitation is equal to or less than the second judgment reference precipitation and the wind speed is equal to or less than the judgment reference wind speed. The second region is a region where the amount of precipitation is equal to or less than the second judgment reference precipitation and the wind speed exceeds the judgment reference wind speed. The third region is a region where the amount of precipitation exceeds the second judgment reference precipitation and is equal to or less than the first judgment reference precipitation and the wind speed is equal to or less than the judgment reference wind speed. The fourth region is a region where the amount of precipitation exceeds the second judgment reference precipitation and is equal to or less than the first judgment reference precipitation and the wind speed exceeds the judgment reference wind speed. The fifth region is a region where the precipitation exceeds the first criteria precipitation and the wind speed is equal to or less than the criteria wind speed. The sixth region is a region where the precipitation exceeds the first criteria precipitation and the wind speed exceeds the criteria wind speed.

If the combination of the predicted precipitation and the predicted wind speed falls in the first, second, or third region, it is determined that there is no possibility of rainwater infiltration. On the other hand, if the combination of the predicted precipitation and the predicted wind speed falls in the fourth, fifth, or sixth region, it is determined that there is a possibility of rainwater infiltration. In other words, if the predicted precipitation exceeds the first judgment standard precipitation, it is determined that there is a possibility of rainwater infiltration regardless of the predicted wind speed. If the predicted precipitation is equal to or less than the first judgment standard precipitation but exceeds the second judgment standard precipitation, it is determined that there is a possibility of rainwater infiltration if the predicted wind speed at the same time exceeds the judgment standard wind speed. In other words, if the predicted precipitation exceeds the second judgment standard precipitation, it is determined that there is a possibility of rainwater infiltration if the predicted wind speed is strong enough to be greater than a certain level.

In step S11, the ventilation control device 2 may acquire weather forecast data on one or more of temperature, humidity, air pressure, solar radiation, and wind direction, as well as wind speed, and may determine in step S12 based on the acquired weather forecast data whether the predicted precipitation and predicted wind speed after X minutes fall within an area where there is a possibility of infiltration in the rainwater infiltration determination table.

In step S14, the ventilation control device 2 may acquire weather forecast data for one or more of temperature, humidity, air pressure, solar radiation, and wind direction, as well as wind speed, and may determine in step S15 based on the acquired weather forecast data whether the amount of precipitation and wind speed Y minutes from now will be outside the area where infiltration is possible according to the rainwater infiltration determination table.

The ventilation system 100 according to the second embodiment reduces the amount of air supplied when the expected wind speed is high, even if the expected precipitation is not extremely high, and therefore is able to better prevent rainwater from entering compared to the ventilation system 100 according to the first embodiment.

Embodiment 3.
The ventilation system 100 according to the third embodiment has a ventilation device 1 and a ventilation control device 2, similar to the ventilation system 100 according to the first embodiment. Fig. 7 is a diagram showing an example of installation of the ventilation devices in the ventilation system according to the third embodiment. The ventilation system 100 according to the third embodiment includes a plurality of ventilation devices 1. The ventilation devices 1 include a west-side ventilation device 3 installed on a west wall 31 of the building 30, and a north-side ventilation device 4 installed on a north wall 32 of the building 30.

In the ventilation system 100 according to the third embodiment, the ventilation control device 2 controls the ventilation device 1 based on the predicted wind direction when precipitation that may cause rainwater to infiltrate is predicted. For example, if a westerly wind is predicted during the time period when precipitation that may cause rainwater to infiltrate is predicted, the west-side ventilation device 3 installed on the west wall 31 may have rain blown in by the westerly wind because the wind tunnel of the propeller fan faces west. On the other hand, the north-side ventilation device 4 installed on the north wall 32 has a small possibility of having rain blown in by the westerly wind because the wind tunnel of the propeller fan does not face west. Therefore, the ventilation control device 2 stops the operation of the west-side ventilation device 3 or reduces the amount of air supplied by the west-side ventilation device 3, and maintains the amount of air supplied by the north-side ventilation device 4 unchanged. In this way, even when precipitation that may cause rainwater infiltration is predicted, the ventilation system 100 according to embodiment 3 can increase the ventilation volume during precipitation by continuing to operate the ventilation devices 1 that are unlikely to blow rain into the building due to wind, compared to stopping the operation of all ventilation devices 1 or reducing the air supply volume regardless of wind direction.

Embodiment 4.
The configuration of the ventilation system 100 according to embodiment 4 is similar to that of the ventilation system 100 according to embodiment 3. In the ventilation system 100 according to embodiment 4, the ventilation control device 2 separately acquires weather forecast data for the vicinity of the west wall 31 of the building 30 and weather forecast data for the vicinity of the north wall 32 of the building 30, and separately determines whether or not there is a possibility of rainwater infiltrating into the west-side ventilation device 3 and whether or not there is a possibility of rainwater infiltrating into the north-side ventilation device 4. Then, the ventilation control device 2 stops operation or reduces the amount of air supply only for those ventilation devices 1 into which rainwater is likely to infiltrate.

For example, there is a method of predicting the local wind speed around the ventilation device 1, determining the possibility of rainwater infiltration, and controlling the ventilation device 1. By predicting the wind speed that changes locally due to the unique structure of the building 30 and its surroundings separately from the wind speed based on weather forecasts for the area around the building 30 in which the ventilation device 1 is installed, it is possible to determine the possibility of rainwater infiltration for each installation position of the ventilation device 1 and control the ventilation device 1.

Embodiment 5.
Fig. 8 is a diagram showing the configuration of a ventilation system according to embodiment 5. The ventilation system 100 according to embodiment 5 includes a ventilation control history storage device 5 that stores the ventilation control history by the ventilation control device 2, and a display device 6 that displays the ventilation control history, weather forecast data, and actual weather measurement data together with time data.

The ventilation control history storage device 5 stores the ventilation control history that associates the on/off command for the ventilation blower 11, the speed adjustment, or the opening/closing amount instruction for the damper 13 sent from the ventilation control device 2 to the ventilation device 1 with time data, and outputs the ventilation control history to the display 6 in response to the instruction from the ventilation control device 2.

The weather forecast data and actual weather measurement data are input to the display 6 in association with one or more of the weather data such as precipitation, temperature, humidity, air pressure, solar radiation, wind direction, and wind speed, and time data.

The weather forecast data and actual weather measurement data may be obtained by various weather sensors installed on-site, such as temperature sensors, humidity sensors, and anemometers, or may be obtained from external sources, such as meteorological agencies or businesses that provide weather information.

The display 6 shows the weather forecast data, actual weather data, and ventilation control history so that it is clear what the conditions were at the same time on the same date. For example, the display 6 shows a graph with time on the horizontal axis and actual precipitation, predicted precipitation, and ventilation fan speed adjustment on the vertical axis. In this way, the user of the ventilation system 100 can determine whether the weather forecast data was correct compared to the actual weather data, and whether the control results of the ventilation device 1 based on the weather forecast data were appropriate.

The ventilation control device 2, ventilation control history storage device 5, and display device 6 may be integrated together, or may be stored in separate housings. For example, the ventilation control device 2 and ventilation control history storage device 5 may be provided in the same computer system, and a monitor serving as the display device 6 may be connected to a separate computer system.

Embodiment 6.
9 is a diagram showing the configuration of a ventilation system according to embodiment 6. The ventilation system 100 according to embodiment 6 differs from the ventilation system 100 according to embodiment 5 in that it includes a ventilation control condition input device 7 for inputting and changing control conditions to the ventilation control device 2.

The ventilation control condition input device 7 inputs and changes the weather data thresholds set in advance in the ventilation control device 2 and the control content for the ventilation device 1 when the thresholds are exceeded. For example, if a control condition is set in advance to change the speed adjustment amount of the ventilation blower so as to reduce the amount of air supplied at the predicted time when a certain amount of precipitation or wind speed is predicted, the ventilation control condition input device 7 can change the threshold value of the amount of precipitation or wind speed and the speed adjustment amount of the ventilation blower 11.

Therefore, by using the ventilation control condition input device 7 in conjunction with the ventilation control history storage device 5, it is possible to adjust the control conditions to more appropriate conditions while checking weather forecast data, actual weather data, and ventilation control history. Since the sensitivity of the sensor used and its ability to track weather changes, the ease with which rainwater can penetrate the ventilation device 1, and regional weather conditions vary depending on the individual property and time of year, by making it possible to adjust the precipitation or wind speed threshold and the speed adjustment amount of the ventilation blower 11, it is possible to set ventilation control conditions that appropriately suppress the infiltration of rainwater regardless of the location and time of year.

Embodiment 7.
The configuration of the ventilation system 100 according to the seventh embodiment is the same as that of the ventilation system 100 according to any one of the first to sixth embodiments. In the ventilation system 100 according to the seventh embodiment, the supply air amount management unit 22 controls, stores, and displays the supply air amount based on the predicted data of the dispersion of pollen and fine particulate matter. For example, when it is predicted that the dispersion amount of pollen or fine particulate matter will increase, the supply of air by the ventilation device 1 is stopped or the supply amount is reduced to suppress the intrusion of pollen or fine particulate matter into the building 30. In this case, when acquiring the weather forecast data, a particle measuring device for measuring the dispersion of pollen and fine particulate matter may be used, or external pollen dispersion forecast data and fine particulate matter distribution forecast data may be referred to.

The ventilation system 100 according to the seventh embodiment can prevent the air in the building 30 from becoming contaminated with pollen or fine particulate matter by stopping the air supply by the ventilation device 1 or reducing the amount of air supply when it is predicted that the amount of pollen or fine particulate matter dispersed in the air is increased.

Embodiment 8.
The configuration of the ventilation system 100 according to the eighth embodiment is the same as that of the ventilation system 100 according to any one of the first to sixth embodiments. In the ventilation system 100 according to the eighth embodiment, the supply air amount management unit 22 controls, stores, and displays the supply air amount based on the predicted data of the outdoor carbon monoxide concentration or carbon dioxide concentration. For example, when it is predicted that the outdoor carbon monoxide concentration or carbon dioxide concentration will be high, the supply air by the ventilation device 1 is stopped or the supply air amount is reduced, thereby suppressing the increase in the carbon monoxide concentration or carbon dioxide concentration in the air inside the building 30. In this case, in addition to using a carbon monoxide sensor or a carbon dioxide sensor, road traffic information such as an external congestion prediction may be used to obtain weather forecast data, and traffic volume may be obtained by a camera. In this way, when the concentration of automobile exhaust gas in the air becomes high, the air can be prevented from being taken into the room. The transition of the automobile exhaust gas concentration can be predicted by the carbon monoxide concentration or carbon dioxide concentration, and can also be predicted based on whether the traffic volume on nearby roads is high or low from road traffic information.

The ventilation system 100 according to the eighth embodiment can prevent the carbon monoxide or carbon dioxide concentration in the air inside the building 30 from increasing by stopping the supply of air by the ventilation device 1 or reducing the amount of air supplied when it is predicted that the carbon monoxide or carbon dioxide concentration outdoors will become high.

Embodiment 9.
10 is a diagram showing the configuration of a ventilation system according to a ninth embodiment. The ventilation system 100 according to the ninth embodiment includes a ventilation device 1, a ventilation control device 2, a rainwater infiltration amount measuring device 8, a learning device 23, a learned model storage unit 24, and an inference device 25. The learning device 23 learns the relationship between the ventilation control conditions, the weather forecast data, and the amount of rainwater infiltration. The learned model storage unit 24 stores a learned model 241. The inference device 25 infers the ventilation control conditions using the learned model 241.

FIG. 11 is a diagram showing the configuration of a learning device for a ventilation system according to embodiment 9. The learning device 23 includes a learning data acquisition unit 231 and a model generation unit 232. The learning data acquisition unit 231 acquires the ventilation control conditions, weather forecast data, and rainwater infiltration amount as learning data. The ventilation control conditions are composed of one or more of the speed of the ventilation blower and the opening/closing angle of the damper 13. The weather forecast data is composed of one or more of the amount of precipitation, temperature, humidity, air pressure, solar radiation, wind direction, wind speed, etc.

The amount of rainwater infiltration is measured by rainwater infiltration amount measuring device 8. Rainwater infiltration amount measuring device 8 is constructed by installing a commercially available rain sensor or rain gauge on the indoor side of ventilation device 1. If a rain sensor that cannot detect the amount of rain is used, multiple rain sensors are installed and the amount of rainwater infiltration can be determined according to the number of rain sensors that react. As another method, when rainwater infiltration is visually confirmed, an operator or manager can directly input the amount of rainwater infiltration into rainwater infiltration amount measuring device 8. In this case, a rain sensor is not necessary.

The model generation unit 232 learns the ventilation control conditions based on learning data including the ventilation control conditions, weather forecast data, and the amount of rainwater infiltration. In other words, it generates a learned model 241 that infers the ventilation control conditions from the weather forecast data and the amount of rainwater infiltration of the ventilation system 100.

The learning algorithm used by the model generation unit 232 may be a known algorithm such as supervised learning, unsupervised learning, or reinforcement learning. As an example, a case where reinforcement learning is applied will be described. In reinforcement learning, an agent, which is the subject of action in a certain environment, observes the parameters of the environment, which is the current state, and determines the action to be taken. The environment changes dynamically due to the agent's actions, and the agent is given a reward according to the change in the environment. The agent repeats this process, and learns the course of action that will obtain the most reward through a series of actions. Q-learning and TD-learning are known as representative methods of reinforcement learning. For example, in the case of Q-learning, the general update formula for the action value function Q(s, a) is expressed as the following formula (1).

In the above formula (1), s _t represents the state of the environment at time t, and a _t represents the action at time t. The state changes to s _t+1 due to the action a _t . _{r t+1} represents the reward obtained due to the change in state, γ represents the discount rate, and α represents the learning coefficient. Note that γ is in the range of 0<γ≦1, and α is in the range of 0<α≦1. The ventilation control condition becomes the action a _t , the amount of rainwater infiltration becomes the state s _t , and the best action a _t in the state s _t at time t is learned.

In addition, since the amount of precipitation and wind speed affect the amount of rainwater infiltration, there is a certain correlation between the weather forecast data and the amount of rainwater infiltration. Therefore, the best action a _t in state s _t at time t is associated with the weather forecast data at the same time. In other words, the best action a _t in the weather forecast data at time t is learned.

The update formula expressed by equation (1) increases the action value Q if the action value Q of the action a with the highest Q value at time t+1 is greater than the action value Q of the action a executed at time t, and decreases the action value Q in the opposite case. In other words, it updates the action value function Q(s, a) so that the action value Q of action a at time t approaches the best action value at time t+1. This allows the best action value in a certain environment to be propagated sequentially to the action value in the previous environment.

As described above, the model generation unit 232 that generates the trained model 241 by reinforcement learning includes a reward calculation unit 232a and a function update unit 232b.

The reward calculation unit 232a calculates the reward based on the ventilation control conditions and the amount of rainwater infiltration. The reward calculation unit 232a calculates the reward r based on the increase or decrease in the amount of rainwater infiltration. For example, if the amount of rainwater infiltration decreases, the reward r is increased, and if the amount of rainwater infiltration increases, the reward r is decreased. For example, if the amount of rainwater infiltration decreases, a reward of "1" is given, and if the amount of rainwater infiltration increases, a reward of "-1" is given.

The function update unit 232b updates the function for determining the ventilation control condition in accordance with the reward calculated by the reward calculation unit 232a, and outputs the function to the learned model storage unit 24. For example, in the case of Q-learning, the action value function Q(s _t , a _t ) expressed by the formula (1) is used as a function for calculating the ventilation control condition.

The learning process described above is repeated. The learned model storage unit 24 stores the action value function Q(s _t , a _t ) updated by the function update unit, that is, the learned model 241.

Next, the learning process performed by the learning device 23 will be described. Figure 12 is a flowchart showing the learning process performed by the learning device of the ventilation system according to embodiment 9.

In step S21, the learning data acquisition unit 231 acquires the ventilation control conditions, weather forecast data, and the amount of rainwater infiltration as learning data.

In step S22, the model generation unit 232 calculates the reward based on the ventilation control conditions and the amount of rainwater infiltration. Specifically, the reward calculation unit 232a acquires the ventilation control conditions and the amount of rainwater infiltration, and determines whether to increase or decrease the reward based on a predetermined increase or decrease in the amount of rainwater infiltration.

If the amount of rainwater infiltration decreases, the result in step S22 is "rainwater infiltration amount decreased." In this case, the reward calculation unit 232a determines that the reward should be increased, and so increases the reward in step S23. On the other hand, if the amount of rainwater infiltration increases, the result in step S22 is "rainwater infiltration amount increased." In this case, the reward calculation unit 232a determines that the reward should be decreased, and so decreases the reward in step S24.

In step S25, the function update unit 232b updates the action value function Q(s _t , a _{t )} represented by equation (1) stored in the learned model storage unit 24, based on the reward calculated by the reward calculation unit 232a.

The learning device 23 repeatedly executes the above steps S21 to S25, and stores the generated action value function Q(s _t , a _t ) as the learned model 241.

In addition, in generating the trained model 241, the best action a _t in state s _t at time t is associated with the weather forecast data at the same time. That is, the best action a _t for the weather forecast data at time t is trained.

The learning device 23 according to this embodiment stores the learned model 241 in a learned model storage unit 24 provided outside the learning device 23, but the learned model storage unit 24 may be provided inside the learning device 23.

FIG. 13 is a diagram showing the configuration of an inference device for a ventilation system according to embodiment 9. The inference device 25 includes an inference data acquisition unit 251 and an inference unit 252.

The inference data acquisition unit 251 acquires weather forecast data. The weather forecast data consists of one or more of precipitation, temperature, humidity, air pressure, solar radiation, wind direction, wind speed, etc.

The inference unit 252 infers ventilation control conditions using the trained model 241. That is, by inputting the weather forecast data acquired by the data-for-inference acquisition unit 251 to the trained model 241, it is possible to infer ventilation control conditions suitable for the weather forecast data.

At this time, the weather forecast data was associated with the amount of rainwater infiltration when the trained model 241 was generated.

In the present embodiment, the ventilation control conditions are output using the trained model 241 trained by the model generation unit 232. However, the trained model 241 may be acquired from another ventilation system 100, and the ventilation control conditions may be output based on this trained model 241. Also, the trained model 241 may be generated by a device other than the ventilation system 100. In other words, the trained model 241 may be generated by a learning device 23 provided separately from the ventilation system 100.

Next, we will explain the process for obtaining ventilation control conditions using the inference device 25. Figure 14 is a flowchart showing the inference process of the inference device of the ventilation system according to embodiment 9.

In step S31, the inference data acquisition unit 251 acquires weather forecast data, which is inference data.

In step S32, the inference unit 252 inputs the weather forecast data into the trained model 241 stored in the trained model storage unit 24 to obtain ventilation control conditions.

In step S33, the inference unit 252 outputs the data on the ventilation control conditions obtained by the learned model 241 to the air supply volume management unit 22.

In step S34, the air supply volume management unit 22 controls the ventilation device 1 using the output ventilation control conditions. This makes it possible to reduce the amount of rainwater entering through the ventilation device 1. The ventilation device 1 is controlled by the speed of the ventilation blower and the opening/closing angle of the damper 13.

In the present embodiment, a case has been described in which reinforcement learning is applied to the learning algorithm used by the model generation unit 232, but this is not limited to this. As for the learning algorithm, it is also possible to apply supervised learning, unsupervised learning, semi-supervised learning, etc. in addition to reinforcement learning.

The learning algorithm used in the model generation unit 232 can be deep learning, which learns to extract the features themselves, or machine learning can be performed according to other known methods, such as neural networks, genetic programming, functional logic programming, and support vector machines.

The learning device 23 and the inference device 25 may be separate devices from the ventilation system 100 and connected to the ventilation system 100 via a network. Furthermore, the learning device 23 and the inference device 25 may exist on a cloud server.

The model generation unit 232 may learn the ventilation control conditions using learning data acquired from multiple ventilation systems 100. The model generation unit 232 may acquire learning data from multiple ventilation systems 100 used in the same area, or may learn the ventilation control conditions using learning data collected from multiple ventilation systems 100 operating independently in different areas. It is also possible to add or remove a ventilation system 100 from which learning data is collected during the process. Furthermore, the learning device 23 that has learned the ventilation control conditions for a certain ventilation system 100 may be applied to another ventilation system 100, and the ventilation control conditions for the other ventilation system 100 may be re-learned and updated.

The ventilation system 100 according to embodiment 9 learns the ventilation control conditions based on learning data including the ventilation control conditions, weather forecast data, and the amount of rainwater infiltration to generate a trained model 241, and infers the ventilation control conditions using the trained model 241. Therefore, even in weather conditions where it is difficult for a human to determine whether or not rainwater is likely to infiltrate, the possibility of rainwater infiltration can be reduced.

Embodiment 10.
15 is a diagram showing the configuration of a ventilation system according to a tenth embodiment. The ventilation system 100 according to the tenth embodiment is a system in which a trained model 241 is shared among a plurality of buildings. In the ventilation system 100 according to the tenth embodiment, a plurality of buildings 30 are the targets of ventilation, and each ventilation control device 2 communicates with a server 50 through a network 40. Data to be communicated includes ventilation control conditions, weather forecast data, and the amount of rainwater infiltration, as well as information on the size of each building 30 and the installation position of the ventilation device 1, etc.

The learning device 23, the learned model storage unit 24, and the inference device 25 are provided in the server 50.

In the ventilation system 100 according to the tenth embodiment, the learning device 23 generates a learned model 241 for each building 30. The learned model 241 varies depending on the location of the building 30, the size of the building 30, and the installation position of the ventilation blower 11, and therefore the learned model 241 created by the ventilation control device 2 of one building 30 cannot be applied as is to the ventilation control device 2 of another building 30. Therefore, the learning device 23 learns the learned models 241 of multiple buildings 30 using learning data to which building information has been added. The inference device 25 infers ventilation control conditions based on the weather forecast data and the building information.

The building information includes the location of the building 30, the size of the building 30, the location of the ventilation devices 1, the location of the rainwater infiltration measurement devices 8, and the ventilation control conditions of each ventilation device 1.

The learning device 23 learns using learning data that includes building information, making it possible to use the learned model 241 to perform ventilation control for different buildings.

In addition, by installing rainwater infiltration amount measuring devices 8 in only some of the buildings 30 out of the multiple buildings 30 and generating a trained model 241, the ventilation device 1 can be controlled using inference data including the trained model 241 and building information, even if rainwater infiltration amount measuring devices 8 are not installed in the other buildings 30.

In addition, when this system is introduced into a certain building 30, the loaned rainwater infiltration amount measuring device 8 is installed for a certain period of time after the introduction to generate a trained model 241, and after a certain period of time has passed, the rainwater infiltration amount measuring device 8 is used sequentially in other buildings 30 to generate trained models 241, thereby minimizing the burden on equipment and generating trained models 241 for each building 30.

By managing the trained model 241 on the server 50, it is also possible to update the trained model 241 at any time.

The hardware configuration of the learning device 23 and the inference device 25 according to the first to tenth embodiments will be described. FIG. 16 is a diagram showing the hardware configuration of the learning device according to the ninth and tenth embodiments. The learning device 23 is realized by a computer system including a processor 91 that executes various processes, a memory 92 that is a main memory, and a storage device 93 that stores information.

The processor 91 may be a calculation means such as an arithmetic unit, a microprocessor, a microcomputer, a CPU (Central Processing Unit), or a DSP (Digital Signal Processor). The memory 92 may be a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable Read Only Memory), or an EEPROM (registered trademark) (Electrically Erasable Programmable Read Only Memory). The storage device 93 stores a program for performing a process of learning ventilation control conditions.

The above computer system realizes the functions of the model generation unit 232 by the processor 91 reading into the memory 92 the programs stored in the storage device 93 and corresponding to the processing of each component, and executing them. The memory 92 is also used as a temporary memory for each process executed by the processor 91. The programs executed by the processor 91 may be provided in a state stored in a storage medium, or may be provided via a network.

Similar to the learning device 23, the inference device 25 is realized by a computer system including a processor 91 that executes various processes, a memory 92 that is a main memory, and a storage device 93 that stores information. In the computer system that realizes the inference device 25, the storage device 93 stores a program for performing a process of inferring ventilation control conditions using a learned model.

The configurations shown in the above embodiments are merely examples of the content, and may be combined with other known technologies. Parts of the configurations may be omitted or modified without departing from the spirit of the invention.

1 ventilation device, 2 ventilation control device, 3 west side ventilation device, 4 north side ventilation device, 5 ventilation control history storage device, 6 display device, 7 ventilation control condition input device, 8 rainwater infiltration amount measurement device, 11 ventilation blower, 12 speed regulator, 13 damper, 21 rainwater infiltration possibility determination unit, 22 air supply amount management unit, 23 learning device, 24 trained model storage unit, 25 inference device, 30 building, 31 west wall, 32 north wall, 40 network, 50 server, 91 processor, 92 memory, 93 storage device, 100 ventilation system, 231 learning data acquisition unit, 232 model generation unit, 232a reward calculation unit, 232b function update unit, 241 trained model, 251 inference data acquisition unit, 252 inference unit.

Claims (13)

A ventilation system installed in the building;
A ventilation control device that controls the ventilation device based on weather forecast data,
The ventilation control device includes:
a rainwater infiltration possibility determination unit that determines whether or not there is a possibility of rainwater infiltration into the ventilation device based on the weather forecast data;
A ventilation system characterized by having an air supply volume management unit that manages the amount of air supplied by the ventilation device by stopping the air supply by the ventilation device or reducing the amount of air supplied by the ventilation device when there is a possibility of rainwater entering the ventilation device. A plurality of the ventilation devices are provided,
The plurality of ventilation devices are installed in different locations of the building,
The rainwater infiltration possibility determination unit determines whether or not there is a possibility of rainwater infiltration for each of the plurality of ventilation devices,
The ventilation system according to claim 1, characterized in that the air supply volume management unit stops air supply to only the ventilation device into which rainwater may enter or reduces the amount of air supply by the ventilation device. The ventilation system according to claim 2, characterized in that the rainwater infiltration possibility determination unit obtains the weather forecast data individually for each of the installation locations of the multiple ventilation devices and determines whether or not there is a possibility of rainwater infiltrating into the multiple ventilation devices. A ventilation control history storage device that stores a ventilation control history, which is a control history of the ventilation device by the ventilation control device;
The ventilation system according to any one of claims 1 to 3, characterized in that it is provided with a display that displays weather forecast data, actual weather data which is measured weather data, and the ventilation control history together with time data. The ventilation system according to any one of claims 1 to 4, characterized in that the ventilation control device is provided with a ventilation control condition input device for inputting ventilation control conditions, which are control conditions for the ventilation device. The ventilation system according to any one of claims 1 to 5, characterized in that the air supply volume management unit stops the air supply by the ventilation device or reduces the air supply volume by the ventilation device based on the amount of pollen or fine particulate matter dispersed or the predicted amount of dispersion. The ventilation system according to any one of claims 1 to 5, characterized in that the air supply volume management unit stops the air supply by the ventilation device or reduces the amount of air supply by the ventilation device based on predicted concentration data of carbon monoxide or carbon dioxide in the outside air. A rainwater infiltration amount measuring device that measures the amount of rainwater infiltration into the ventilation device;
A learning data acquisition unit that acquires learning data including the rainwater infiltration amount, the weather forecast data, and a ventilation control condition that is a control condition of the ventilation device;
The ventilation system described in any one of claims 1 to 7, characterized in that it is provided with a learning device having a model generation unit that uses the learning data to generate a learned model for inferring the ventilation control conditions from the weather forecast data. An inference data acquisition unit that acquires the weather forecast data;
The ventilation system described in any one of claims 1 to 8, characterized in that it is provided with an inference device having an inference unit that outputs ventilation control conditions, which are control conditions of the ventilation device, from the weather forecast data acquired by the inference data acquisition unit using a learned model for inferring ventilation control conditions, which are control conditions of the ventilation device, from the weather forecast data. A learning data acquisition unit that acquires learning data including an amount of rainwater infiltrating into a ventilation device installed in a building, weather forecast data for a location where the ventilation device is installed, and ventilation control conditions for the ventilation device;
and a model generation unit that uses the learning data to generate a trained model for inferring the ventilation control conditions from the weather forecast data. The model generation unit
A reward calculation unit that calculates a reward based on a reward standard for the amount of rainwater infiltration into the ventilation device;
11. The learning device according to claim 10, further comprising a function updating unit that updates a function for determining the ventilation control condition in accordance with the reward calculated by the reward calculation unit. an inference data acquisition unit that acquires weather forecast data for a location where a ventilation device is installed in a building;
an inference unit that infers ventilation control conditions, which are control conditions of the ventilation device, from the weather forecast data acquired by the inference data acquisition unit using a learned model for inferring the ventilation control conditions, which are control conditions of the ventilation device, from the weather forecast data. a learning data acquisition unit that acquires learning data including the amount of rainwater infiltration into each of the ventilation devices installed in each of a plurality of buildings, weather forecast data for the locations where the ventilation devices of each of the plurality of buildings are installed, ventilation control conditions for the ventilation devices installed in each of the plurality of buildings, and location information for each of the plurality of buildings;
and a model generation unit that uses the learning data to generate a trained model for inferring the ventilation control conditions from the weather forecast data.

Images