JP2021154049A - Drying device - Google Patents

Abstract

To predict the time required for laundry hung in a bathroom to be dried by a dryer with good accuracy.SOLUTION: A drying device includes: a learning device 32 which generates a learned model 13 for predicting prediction drying time tp required for a dryer 1 to dry an object, by performing machine-learning using, as input data 38 for learning, a temperature before drying, a humidity before drying, and additional information including at least one of an image in which the object is imaged, a volume of a space and a weight before drying which is the weight of the object before drying, and using actual drying time td as teacher data; a storage part 35 for storing the learned model 13; and a prediction part 33 for predicting the prediction drying time tp by inputting, as input data 42 for prediction, the temperature before drying, the humidity before drying, and the additional information in the learned model 13, and outputting them. The learned model 13 is generated for each dryer 1.SELECTED DRAWING: Figure 2

Description

The present invention relates to a drying device including a dryer for drying laundry dried in a predetermined indoor space such as a bathroom.

As shown in Patent Document 1, in a bathroom dryer, a drying means (dryer) is provided in the bathroom, and air is blown out (blower) from the drying means (dryer) to dry the laundry dried in the bathroom. .. The bathroom dryer disclosed in Patent Document 1 detects that the laundry has dried due to changes in the temperature outside the bathroom and the humidity inside the bathroom.

Japanese Unexamined Patent Publication No. 2012-21520 JP-A-2018-102681 JP-A-2018-134143

However, the bathroom dryer disclosed in Patent Document 1 can detect that the laundry has dried during drying, but when the drying is started, it is necessary for the laundry to dry, that is, to dry the laundry. I can't predict the time.

Bathroom dryers generally have a timer function and can set the drying time. However, since the time required for drying is unknown, the person performing the drying empirically predicts the time and sets the drying time longer than the predicted time in order to surely dry the laundry. rice field.

Therefore, the dryer may be operated for an unnecessarily long time to perform inefficient operation, or conversely, the drying may be insufficient and re-drying may be required, so that efficient drying work may not be performed. In this way, it is required to predict the time required for drying in advance at the start of drying, but since various factors affect each other in a complicated manner in the process of drying the laundry, the time required for drying is predicted. It was difficult to do. For example, attempts have been made to predict the time required for drying depending on the weight of the laundry and the surrounding environment, but it has not been sufficiently predicted.

An object of the present invention is to accurately predict the time required for laundry dried in a bathroom to be dried by a dryer.

In order to achieve the above object, the characteristic configuration of the drying device according to the embodiment of the present invention is a drying device that dries an object dried in a predetermined space, and is a dryer that blows air into the space. A temperature sensor that measures the temperature in the space, a humidity sensor that measures the humidity in the space, a pre-drying temperature that is the pre-drying temperature measured by the temperature sensor, and a pre-drying temperature that is measured by the humidity sensor. An information acquisition unit that acquires information including pre-drying humidity, which is the humidity of the object, the pre-drying temperature and the pre-drying humidity, an image of the object, the volume of the space, and the pre-drying of the object. Predictive drying required for the dryer to dry the object by machine learning using at least one of the pre-drying weights, which is the weight, as input data for learning and the actual drying time as teacher data. A learning device that generates a trained model for predicting time, a storage unit that stores the trained model, and input for prediction of the pre-drying temperature, the pre-drying humidity, and the additional information in the trained model. It is provided with a prediction unit that inputs as data to predict and output the predicted drying time, and the trained model is generated for each dryer.

The time required to dry an object such as laundry (drying time) is difficult to predict before drying because it is affected by a combination of various factors. According to the above configuration, the predicted drying time is calculated using a machine-learned trained model that takes into account additional information that has a large effect on the drying time, in addition to the pre-drying temperature and pre-drying humidity in the space at the start of drying. Predict. Therefore, a trained model in which the interrelationships of various factors are taken into consideration is generated, and the predicted drying time can be predicted with high accuracy using this trained model.

Further, the dryer is often used in a certain usage environment and conditions so that it can be used in the same household. By specifically generating the trained model for a predetermined dryer, it is possible to generate a trained model with high accuracy corresponding to the usage environment and usage conditions. As a result, it is possible to accurately predict the predicted drying time in consideration of the usage environment and usage conditions of the dryer.

Further, an imaging unit for photographing the object may be further provided, and the image may be photographed by the imaging unit before drying and passed to the information acquisition unit.

An image of the object can be used as a candidate for additional information. By providing the image pickup unit, such an image can be easily taken. In addition, the drying time of the object is affected by the type and size of the object. By generating a trained model that includes an image of the object as input data and predicting the predicted drying time, an accurate predicted drying time that takes into account the type and size of the object, etc. You can make predictions.

Further, a holding portion provided in the space for drying the object and a weight sensor for measuring the weight of the object dried on the holding portion are further provided, and the weight before drying is measured by the weight sensor. The actual drying time is the elapsed time from the start of ventilation to the time when it is estimated from the transition of the weight of the object that the drying is completed. You may.

The pre-drying weight of the object can be used as a candidate for additional information. By providing the weight sensor in the holding portion, the weight of the object before drying can be easily measured. In addition, the drying time of the object may be affected by the weight of the object before drying. By generating a trained model including the pre-drying weight of the object as input data and predicting the predicted drying time, the predicted drying time can be predicted with high accuracy.

The proportion of clothing types contained in the laundry may tend to be similar for each household, and the effect of clothing types on the drying time may be small. As a result, the pre-drying weight of the object has a relatively large effect on the drying time. In such a case, the predicted drying time can be predicted more accurately by generating a trained model including the pre-drying weight of the object as input data and predicting the predicted drying time.

Further, as the object dries, the water contained in the object evaporates and the weight of the object decreases. Therefore, it is possible to detect that the drying of the object has been completed due to the change in the weight of the object. By providing a weight sensor capable of measuring the weight of the object, it is possible to measure the weight change of the object during the drying process and easily detect that the drying of the object has been completed.

Further, a holding portion provided in the space for drying the object, a weight sensor for individually measuring the weight of each of the objects dried in the holding portion, and an individual object from the image. Further provided with a determination unit for determining the type, each of the pre-drying weights is measured by the weight sensor and passed to the information acquisition unit, and the learning device uses the trained model as the type of the object. The prediction unit may predict the time required for drying each of the objects, and the time with the longest predicted time may be the predicted drying time.

When the object is dried under the same conditions, the drying time differs depending on the type of the object. With the above configuration, a trained model can be generated for each type of object, and the dry weight for each type of object can be used as input data for each trained model. Therefore, it is possible to more accurately predict the individual drying time for each type of object. Then, by setting the longest drying time as the predicted drying time of the entire object, the predicted drying time can be predicted more accurately.

Further, the dryer can select the air volume and the blast temperature within a predetermined range, further includes the selected air volume and the blast temperature as the learning input data, and further includes the selected air volume and the blast temperature as the prediction input data. The blast temperature may be further included.

The dryer may be able to set the air volume and air temperature to be blown. In this case, the drying time differs depending on the strength of the air volume and the blowing temperature. According to the above configuration, it is possible to accurately predict the predicted drying time by generating a trained model including the air volume and the blowing temperature set by the dryer as input data and predicting the predicted drying time. ..

Further, each time the learning device performs drying, the prediction input data input to the prediction unit is newly added as the learning input data, and the actual drying time at the time of drying is newly added as the teacher data. It is preferable to generate a trained model and update the trained model stored in the storage unit with the newly generated trained model.

With such a configuration, it is expected that machine learning using new input data and teacher data will progress each time the drying process is performed, and the accuracy of the generated trained model will be improved each time the drying process is performed. can. As a result, by using such a trained model, it can be expected that the accuracy of the predicted drying time is improved every time the drying process is performed.

Further, the dryer may receive the predicted drying time from the prediction unit and automatically stop the blowing when the predicted drying time elapses after the blowing is started.

With such a configuration, the object can be dried efficiently and easily.

Further, a server for performing data communication is further provided, the learning device is mounted on the server, the information acquisition unit transmits the acquired information to the learning device, and the server transmits the generated learned model to the learning device. It may be stored in the storage unit.

With such a configuration, it is possible to appropriately generate a trained model on the server while simplifying the configuration of the dryer.

Further, a server for performing data communication is further provided, the learning device and the prediction unit are mounted on the server, the information acquisition unit transmits the acquired information to the learning device and the prediction unit, and the server predicts. The predicted drying time may be transmitted to the dryer.

With such a configuration, it is possible to appropriately generate a trained model on the server and predict the predicted drying time while simplifying the configuration of the dryer.

It is the schematic which illustrates the structure of the drying apparatus. It is a block diagram which illustrates the structure of the drying apparatus. It is a figure which illustrates the data flow in machine learning. It is a figure explaining the method of estimating the completion of drying from the weight change. It is a figure which illustrates the processing flow of a drying apparatus. It is a figure explaining the structure which obtains the predicted drying time in another embodiment. It is a figure which illustrates the processing flow of the drying apparatus in another embodiment.

[Construction of drying device]
The configuration of the drying device 10 (drying device system) according to the present embodiment will be described with reference to FIGS. 1 and 2.

The dryer 1 is provided behind the ceiling of the bathroom 2 and can blow warm air or the like into the bathroom 2. The bathroom 2 can be provided with a clothesline 3 (corresponding to a "holding portion") over the facing walls, and the clothesline 3 can hang laundry 4 (corresponding to an "object"). The dryer 1 can dry (dry) the laundry 4 by blowing warm air or the like into the bathroom 2.

The dryer 1 includes a display panel 5 that functions as an operation unit 12. The display panel 5 is provided on the wall of the bathroom 2 or the like, and is connected to the main body of the dryer 1 in a state where data communication is possible. The display panel 5 operates the operation of the dryer 1 and displays various information. The dryer 1 can set the blowing intensity (air volume), whether the blown air is warm air or room temperature (blower temperature), the sending time, and the like. The display panel 5 can operate such setting and start / stop of ventilation, and can display the set ventilation time, the elapsed time of ventilation, various environmental information, the current time, and the like. The dryer 1 starts blowing air after setting the blowing temperature, the air volume, the blowing time, and the like on the display panel 5, and dries the laundry 4 dried in the bathroom 2. A remote control means (not shown) such as a remote controller may be provided as the operation unit 12, and the remote control means may be used to set the blowing temperature, the air volume, the blowing time, the start and stop of blowing, and the like. The display panel 5 may be provided in a dressing room or the like outside the bathroom 2. Further, the blowing temperature is not limited to the configuration in which the temperature of the blowing air is set in two stages of warm air and room temperature, and may be configured in which the temperature of the warm air is further set in a plurality of stages.

The drying device 10 controls the operation of the dryer 1 to dry the laundry 4 dried in the bathroom 2. In addition to the dryer 1, the dryer includes a detection device 6 that detects the environment in the bathroom 2 and the state of the laundry 4, and a server 7 that is communicably connected to the dryer 1 via a network or the like. The server 7 stores the trained model 13 and inputs various data transmitted from the control device 14 mounted on the dryer 1 into the trained model 13 to predict the time required to dry the laundry 4. do.

The detection device 6 includes at least a temperature sensor 17 for measuring the temperature in the bathroom 2 and a humidity sensor 18 for measuring the humidity in the bathroom 2. The detection device 6 can further include another sensor group 20, and the sensor group 20 includes an image pickup device 21 for photographing the laundry 4 dried in the bathroom 2 and the laundry 4 dried in the bathroom 2. A weight sensor 22 or the like for measuring the weight of the bathroom. The image pickup device 21 is supported on the side wall or ceiling of the bathroom 2 toward the clothesline 3. The weight sensor 22 is installed on the clothesline 3 and measures the total weight of the laundry 4 dried on the clothesline 3.

The control device 14 of the dryer 1 includes a control unit 25 that controls the operation of the control device 14, a communication unit 26, an information acquisition unit 27, a storage unit 28, and a dryness detection unit 29. The communication unit 26 transmits / receives data to / from the server 7 via a network line such as the Internet according to the control of the control unit 25. The information acquisition unit 27 acquires various information acquired by the detection device 6 via the communication unit 26 in response to the control of the control unit 25. The storage unit 28 stores various types of information as described later. The dryness detection unit 29 detects that the laundry 4 has been dried, as will be described later with reference to FIG.

The server 7 is equipped with AI (Artificial Intelligence) 40, generates and updates the trained model 13 using various information transmitted from the dryer 1, and generates various information transmitted from the dryer 1. The predicted drying time tp, which is the time predicted to be required to dry the laundry 4 by inputting it into the trained model 13 as input data, is predicted. The server 7 includes a communication unit 31, a learning device 32, a prediction unit 33, and a storage unit 35.

The communication unit 31 transmits / receives data to / from the communication unit 26 of the dryer 1. The learning device 32 generates the trained model 13 by inputting the learning input data 38 and the teacher data 39 into the AI and performing machine learning under the control of the server 7. The generated trained model 13 is stored in the storage unit 35. The learning input data 38 includes the pre-drying temperature T0 of the bathroom 2 before the start of drying measured by the temperature sensor 17 and the pre-drying humidity of the bathroom 2 before the start of drying measured by the humidity sensor 18. This is information including H0 and adding additional information 41 described later. As will be described later, the teacher data 39 is the time required when the laundry 4 is actually dried under the conditions corresponding to the learning input data 38, that is, the time from the start of drying to the drying of the laundry 4. Is the actual drying time td.

The prediction unit 33 inputs the prediction input data 42 into the trained model 13 and outputs the prediction drying time tp according to the control of the server 7. The output predicted drying time tp is stored in the storage unit 35 and transmitted to the dryer 1 via the communication unit 31. The prediction input data 42 is information including the pre-drying temperature T0, the pre-drying humidity H0, and the additional information 41, which are acquired when the drying is started.

[Machine learning]
Next, machine learning for generating the trained model 13 will be described with reference to FIGS. 1 to 4.

As described above, the trained model 13 is generated by inputting the training input data 38 as input data and the actual drying time td as teacher data into AI 40. By machine learning a large amount of combinations of the training input data 38 and the actual drying time td, a trained model 13 with higher accuracy is generated. The generated trained model 13 is used to predict the predicted drying time tp, which is the time required for the laundry 4 to be dried by the dryer 1. At the time of prediction, the predicted drying time tp is predicted by inputting the prediction input data 42 into the trained model 13 as described later.

The drying device 10 is often provided in the bathroom of a general household, and naturally the volume of the bathroom is constant and does not change. In addition, depending on the family structure, the behavior pattern of the family, and the like, a certain regularity is often seen in the laundry 4 to be dried for each household. Therefore, even if a predetermined trained model 13 is used at the start of use of the drying device 10, each time the drying device 10 is used thereafter, the results of use for each of the drying devices 10 in each household are fed back. Update model 13. As a result, it is possible to update to a more optimized trained model 13 for each drying device 10 installed in each home. As a result, by using the optimized trained model 13, the predicted drying time tp can be predicted more accurately.

In such machine learning performed each time the dryer 1 is used, the pre-drying temperature T0, the pre-drying humidity H0, and the additional information 41 are used as the learning input data 38. The pre-drying temperature T0 is the temperature of the bathroom 2 at the start of drying. Since the ambient temperature of the laundry 4 affects the drying time, the pre-drying temperature T0 is important as input data for machine learning. Similarly, the pre-drying humidity H0 is the humidity of the bathroom 2 at the start of drying. Since the laundry 4 is difficult to dry when the humidity is high, the humidity H0 before drying is important as input data for machine learning.

In addition to that, as the additional information 41, various information according to the situation of the bathroom 2 itself, the laundry 4 dried at the time of drying, and the like is input. For example, the additional information 41 includes an image P obtained by photographing the laundry 4 being dried at the start of drying, a volume V of the bathroom 2, a pre-drying weight W0 which is the total weight of the laundry 4 at the start of drying, and the like. , At least one. Since the air volume and the blowing temperature of the dryer 1 also affect the drying time, these may be appropriately selected and added to the additional information 41.

The drying time of the laundry 4 varies depending on the type of the laundry 4. For example, jeans and sweatshirts take longer to dry than T-shirts. Also, even with the same T-shirt, a large T-shirt takes longer to dry than a small T-shirt. Therefore, by adding the image P obtained by photographing the laundry 4 to the input data of machine learning, it is expected that an accurate trained model 13 in consideration of features such as the type and size of the laundry 4 is generated. NS. In particular, clothes and the like used at home are limited to some extent for each household, and the type of laundry 4 at home tends to be constant within a certain range. Therefore, learning using the image P obtained by photographing the laundry 4 for each drying device 10 is useful for predicting the predicted drying time tp.

The drying time of the laundry 4 is also affected by the volume V of the bathroom 2. For example, when the amount of laundry 4 is the same, the rate at which the humidity in the bathroom 2 decreases becomes slower as the volume V in the bathroom 2 increases because the decrease in the humidity in the bathroom 2 becomes slower. Further, since it is machine learning of the specific drying device 10, the same bathroom 2 is targeted. Therefore, the volume V of the bathroom 2 is useful as input data for machine learning. Since the volume V of the bathroom 2 does not change, the volume V is stored in the storage unit 28 in advance, and is read out and used at the time of machine learning and at the time of predicting the predicted drying time tp.

The drying time of the laundry 4 tends to be longer as the total weight of the laundry 4 is heavier. For example, if the total weight of the laundry 4 is heavy, there is a high possibility that the amount of the laundry 4 is large. Further, if the total weight of the laundry 4 is heavy, there is a high possibility that the thick laundry 4 that is difficult to dry is often used. Further, if the total weight of the laundry 4 is heavy, the amount of water contained in the laundry 4 may be large. Therefore, it is expected that the trained model 13 with high accuracy will be generated by adding the total weight of the laundry 4 to the input data of machine learning.

Further, when the laundry 4 is dried under such conditions, the actual drying time td required to actually dry the laundry 4 is input as the teacher data. The laundry 4 before drying contains water. As shown in FIG. 4, since the water content evaporates as the drying progresses, the weight of the laundry 4 becomes lighter. Then, when the weight reduction rate of the laundry 4 approaches 0, it can be determined that the moisture of the laundry 4 has almost disappeared and the laundry 4 has dried. For example, the dryness detection unit 29 acquires the weight of the laundry 4 for each predetermined unit time Δt from the weight sensor 22, and when the amount of change ΔW of the weight per unit time Δt becomes smaller than the predetermined weight Wr, the laundry is washed. It is determined that the object 4 is dry.

As described above, each time the drying process is performed, the information acquired during the drying process is used as the learning input data 38, and the actual drying time td actually required is used as the teacher data 39, and the machine learning is repeated. Therefore, the accuracy of the trained model 13 is improved every time the drying process is performed. Further, a dedicated trained model 13 is generated for each dryer 1 installed in each home. Therefore, the trained model 13 that reflects the usage environment of the dryer 1 is generated.

Further, since the additional information 41 reflecting the environment and tendency of each household is added to the input data for learning to perform machine learning, the learned model 13 reflecting the environment and tendency of each household can be generated. As a result of the above, since the predicted drying time tp is predicted using the trained model 13 with such high accuracy, the predicted drying time tp can be predicted with high accuracy.

[Drying process]
Next, an example of a process of drying the laundry 4 in the bathroom 2 will be described with reference to FIGS. 1 to 3.

First, the dryer 1 is started by operating the display panel 5 in a state where the laundry 4 is dried on the clothes drying rod 3 in the bathroom 2. At this time, the blowing temperature, the air volume, and the like of the dryer 1 may be set. When the dryer 1 is operated, the control unit 25 controls the communication unit 26, and the pre-drying temperature T0, which is the temperature of the bathroom 2 measured by the temperature sensor 17, and the humidity sensor 18 are measured via the information acquisition unit 27. The measured humidity H0 before drying, which is the humidity of the bathroom 2, is acquired and stored in the storage unit 28 (step # 1).

The control unit 25 controls the communication unit 26, acquires an image P of the laundry 4 in the bathroom 2 photographed by the image pickup device 21 via the information acquisition unit 27, and stores the image P in the storage unit 28 (step #). 2). Similarly, the control unit 25 controls the communication unit 26, acquires the pre-drying weight W0, which is the total weight of the laundry 4 measured by the weight sensor 22, via the information acquisition unit 27, and stores it in the storage unit 28. (Step # 3). Further, the volume V in the bathroom 2 is acquired in advance and stored in the storage unit 28 (step # 4).

Here, the image P, the weight W0 before drying, and the volume V are additional information 41. In the present embodiment, the image P, the weight W0 before drying, and the volume V are all acquired, but at least one of these may be acquired. Further, in addition to at least one of these, the additional information 41 may include the air blowing temperature, the air volume, and the like set on the display panel 5.

Next, the predicted drying time tp is predicted. At this time, the trained model 13 is stored in the storage unit 35 of the server 7. The trained model 13 reflects the drying process up to the previous time. When the predicted drying time tp is predicted, first, the control unit 25 sends the communication unit 31 of the server 7 via the communication unit 26 as the prediction input data 42, the pre-drying temperature T0, the pre-drying humidity H0, and so on. Image P, pre-drying weight W0, and volume V are transmitted. The server 7 stores the acquired input data 42 for prediction in the storage unit 35.

Then, the server 7 stores the pre-drying temperature T0, the pre-drying humidity H0, the image P, and the pre-drying weight W0 stored in the storage unit 35 as the prediction input data 42 in the learned model 13 stored in the storage unit 35. And the volume V is input, and the predicted drying time tp is output (step # 5). The server 7 temporarily stores the output predicted drying time tp in the storage unit 35, and transmits the output predicted drying time tp to the communication unit 26 of the dryer 1 via the communication unit 31.

The control unit 25 of the dryer 1 stores the received predicted drying time tp in the storage unit 28. Further, the control unit 25 automatically sets the received predicted drying time tp or the time obtained by adding a predetermined time to the predicted drying time tp as the blowing time of the dryer 1 and starts blowing (step # 6). As a result, the laundry 4 which is the object to be dried is started to be dried. The blown air time is not automatically set, the received predicted drying time tp is displayed on the display panel 5, the operator confirms the predicted drying time tp, and the display panel 5 is operated to manually set the blown time. You may set it. In the following description, it is assumed that the predicted drying time tp is automatically set as the blowing time.

Next, it is determined whether or not the blowing time has passed the predicted drying time tp (step # 7). Blasting is continued until the predicted drying time tp elapses (No in step # 7). When it is determined that the predicted drying time tp has elapsed (Yes in step # 7), the control unit 25 stops blowing air and ends drying (step # 8). Although an example in which the blast is automatically stopped has been described here, the operator who confirmed the predicted drying time tp displayed on the display panel 5 did not set the blast time, set the blast to continuous operation, and blasted. The operator may manually stop the ventilation after a time equal to or longer than the predicted drying time tp has elapsed from the start.

During the drying process, the control unit 25 controls the communication unit 26 and continuously acquires the total weight of the laundry 4 measured by the weight sensor 22 via the information acquisition unit 27. The acquired total weight is stored in the storage unit 28 in association with the passage of time. Then, during the drying process or after the drying is completed, the control unit 25 controls the drying detection unit 29 to calculate the actual drying time td actually required for drying from the time change of the total weight as described above (the actual drying time td). Step # 9). The calculated actual drying time td is stored in the storage unit 28.

Next, the trained model 13 in consideration of the drying process this time is generated, and the trained model 13 stored in the storage unit 35 of the server 7 is updated. Specifically, first, the control unit 25 of the dryer 1 controls the communication unit 26 to transmit the actual drying time td stored in the storage unit 28 to the server 7. The server 7 receives the actual drying time td transmitted via the communication unit 31 and stores it in the storage unit 35. Then, the server 7 inputs the input data 42 for prediction corresponding to the received actual drying time td into the AI 40 as input data and the received actual drying time td as teacher data to perform machine learning, and performs machine learning in this drying process. The trained model 13 to which the achievements are added is generated (step # 10), and the trained model 13 stored in the storage unit 35 is updated (step # 11).

As described above, since the trained model 13 that reflects the usage environment of the specific dryer 1 and reflects the environment and tendency of each household in which the dryer 1 is installed is used, the predicted drying time tp can be accurately calculated. Can be predicted.

Further, every time the dryer 1 is used, the input data 42 for prediction of the drying process is used as the input data 38 for learning, and the actually measured actual drying time td is used as the teacher data for superimposing machine learning. Each time, the learning model 13 can be updated to reflect the usage environment of the dryer 1 and the environment and tendency of each household. As a result, the accuracy of the trained model 13 is improved with each use, and the predicted drying time tp can be predicted more accurately.

Further, when the time required for drying is unknown, it is necessary to set a longer blowing time in order to surely dry the laundry 4. As a result, energy may be wasted by operating the dryer 1 for a long time. Since the drying device 10 according to the present embodiment can predict the predicted drying time tp with high accuracy, it is possible to set the blowing time more appropriately and without waste, and it is also possible to suppress the waste of energy. ..

When the image P is captured by the image pickup device 21, the control device 14 may further include a determination unit 34. The determination unit 34 analyzes the image P to determine the type of the laundry 4. When the determination unit 34 is provided, the image P may be divided according to the type of the laundry 4 and may be used as additional information 41 of the learning input data 38 and the prediction input data 42, and the determined laundry 4 may be used as the additional information 41. Types may be included.

[Another Embodiment]
(1) In the above embodiment, the weight sensor 22 has been described as one sensor for measuring the total weight of the laundry 4 dried on the clothesline 3, but a plurality of weight sensors 22 are provided and the weight of each laundry 4 is provided. It can also be configured to measure. In this case, as the learning input data 38 and the prediction input data 42, the total weight obtained by totaling the weights measured by the individual weight sensors 22 may be the pre-drying weight W0, or the individual weight sensors 22 may be used. Each of the measured weights may be a plurality of pre-drying weights W0.

Further, when the weight of each laundry 4 is measured, the trained model 13 may be generated for each type of laundry 4. Hereinafter, a specific configuration example will be described with reference to FIGS. 1 to 3 with reference to FIGS. 6 and 7.

The clothes-drying rod 3 in the bathroom 2 in the present embodiment is provided with a plurality of weight sensors 22, and the weight of each laundry 4 hung on the clothes-drying rod 3 can be measured individually. For example, each weight sensor 22 is provided on each support portion (not shown) of the clothesline 3 on which the laundry 4 is suspended.

As the laundry 4, a case where jeans 4j, a T-shirt 4T, a sweater 4s, and a towel 4t are dried will be described as an example.

The storage unit 35 individually stores the trained model 13 corresponding to each of the jeans 4j, the T-shirt 4T, the sweater 4s, and the towel 4t. The memorized trained model 13 is a trained model 13j for jeans, a trained model 13T for a T-shirt, a trained model 13s for a sweater, and a trained model 13t for a towel. Further, as described above, since the weight sensor 22 can individually measure the weight of the dried laundry 4, the pre-drying weight Wj of the jeans 4j, the pre-drying weight WT of the T-shirt 4T, and the drying of the sweater 4s The pre-weight Ws and the pre-drying weight Wt of the towel 4t are measured individually. When a plurality of laundry 4 of the same type, for example, jeans 4j are dried, the total weight of the plurality of jeans 4j measured by the weight sensor 22 may be the pre-drying weight Wj. At this time, the type of the laundry 4 can be discriminated by analyzing the image P taken by the image pickup device 21.

In the prediction input data 42 according to the present embodiment, in the prediction input data 42 in the above embodiment, the pre-drying weight W0 is replaced with the respective pre-drying weights (Wj, WT, Ws, Wt). For example, the prediction input data 42j in jeans 4j are the weight Wj before drying, the temperature T0 before drying, the humidity H0 before drying, the image P, and the volume V. The image P and the volume V do not necessarily have to be included in the prediction input data 42.

Each prediction input data (42j, 42T, 42s, 43t) is input to the corresponding trained model (13j, 13T, 13s, 13t), and the corresponding individual prediction drying time (tj, tT, ts, tt) is output. The individual predicted drying time tj is the time expected to be required for the dried jeans 4j to dry. The individual predicted drying time tT is the time expected to be required for the dried T-shirt 4T to dry. The individual predicted drying time ts is the time expected to be required for the dried sweater 4s to dry. The individual predicted drying time tt is the time expected to be required for the dried towel 4t to dry.

Then, each individual predicted drying time (tj, tT, ts, tt) is compared, and the maximum individual predicted drying time (tj, tT, ts, tt) is the time required for all the laundry 4 to dry. Is the predicted drying time tp. Such comparison is performed by the determination unit 34 provided in the server 7.

Hereinafter, a specific process example of drying the laundry 4 in the bathroom 2 will be described.
First, when the dryer 1 is operated, the control unit 25 controls the communication unit 26, and the pre-drying temperature T0, which is the temperature of the bathroom 2 measured by the temperature sensor 17, and the humidity sensor via the information acquisition unit 27. The pre-drying humidity H0, which is the humidity of the bathroom 2 measured by 18, is acquired and stored in the storage unit 28 (step # 20).

The control unit 25 controls the communication unit 26, acquires an image P of the laundry 4 in the bathroom 2 photographed by the image pickup device 21 via the information acquisition unit 27, and stores the image P in the storage unit 28 (step #). 21). At this time, the determination unit 34 analyzes the image P and determines the type of each laundry 4 according to the control of the control unit 25.

Further, the control unit 25 controls the communication unit 26, acquires the weight of each laundry 4 measured by each weight sensor 22 via the information acquisition unit 27, and stores it in the storage unit 28 (step # 22). .. At this time, the control unit 25 obtains the weight of each type of laundry 4 from the weight of each laundry 4 based on the type of each laundry 4 determined by analyzing the image P, and determines the weight of each type of laundry 4. The weight before drying (Wj, WT, Ws, Wt) for each type is associated with the type of laundry 4 and stored in the storage unit 28. The process of determining the type of laundry 4 from the image P and distributing the weight measured by the weight sensor 22 to each type of laundry 4 may be performed by the control unit 25 of the dryer 1, but the server 7 may perform the process. You may go.

Further, the volume V in the bathroom 2 is acquired in advance and stored in the storage unit 28 (step # 23).

Next, the predicted drying time tp is predicted. At this time, the trained model (13j, 13T, 13s, 13t) is stored in the storage unit 35 of the server 7 in association with the type of laundry 4 (step # 24). This trained model (13j, 13T, 13s, 13t) reflects the drying process up to the previous time. When the predicted drying time tp is predicted, first, the control unit 25 sends the communication unit 31 of the server 7 via the communication unit 26 as the prediction input data (42j, 42T, 42s, 43t) before drying. The temperature T0, the pre-drying humidity H0, the image P, the respective pre-drying weights (Wj, WT, Ws, Wt), and the volume V are transmitted. The server 7 stores these acquired data in the storage unit 35 as prediction input data (42j, 42T, 42s, 43t) associated with the type of laundry 4.

Then, the prediction unit 33 of the server 7 stores the pre-drying temperature stored in the storage unit 35 as the prediction input data 42 in each of the learned models (13j, 13T, 13s, 13t) stored in the storage unit 35. Individual predicted drying by inputting T0, pre-drying humidity H0, image P, volume V, and pre-drying weight (Wj, WT, Ws, Wt) corresponding to the trained model (13j, 13T, 13s, 13t). The time (tj, tT, ts, tt) is output (step # 25). The server 7 stores the output individual predicted drying time (tj, tT, ts, tt) in the storage unit 35.

Next, the prediction unit 33 of the server 7 compares the output individual predicted drying times (tj, tT, ts, tt) (step # 26), and the maximum, that is, the longest individual predicted drying time (tj). , TT, ts, tt) and output as the predicted drying time tp (step # 27). The server 7 temporarily stores the output predicted drying time tp in the storage unit 35, and transmits the output predicted drying time tp to the communication unit 26 of the dryer 1 via the communication unit 31.

The control unit 25 of the dryer 1 stores the received predicted drying time tp in the storage unit 28. Further, the control unit 25 automatically sets the received predicted drying time tp or the time obtained by adding a predetermined time to the predicted drying time tp as the blowing time of the dryer 1 and starts blowing (step # 28). As a result, the laundry 4 which is the object to be dried is started to be dried. The blown air time is not automatically set, the received predicted drying time tp is displayed on the display panel 5, the operator confirms the predicted drying time tp, and the display panel 5 is operated to manually set the blown time. You may set it. In the following description, it is assumed that the predicted drying time tp is automatically set as the blowing time.

Next, it is determined whether or not the blowing time has passed the predicted drying time tp (step # 29). Blasting is continued until the predicted drying time tp elapses (No in step # 29). When it is determined that the predicted drying time tp has elapsed (Yes in step # 29), the control unit 25 stops blowing air and ends drying (step # 30). Although an example in which the blast is automatically stopped has been described here, the operator who confirmed the predicted drying time tp displayed on the display panel 5 did not set the blast time, set the blast to continuous operation, and blasted. The operator may manually stop the ventilation after a time equal to or longer than the predicted drying time tp has elapsed from the start.

During the drying process, the control unit 25 controls the communication unit 26 and continuously acquires the weight of the laundry 4 measured by the respective weight sensors 22 via the information acquisition unit 27. Further, the control unit 25 calculates the total weight by adding up the weights of all the acquired laundry 4. The total weight is stored in the storage unit 28 in association with the passage of time. Then, during the drying process or after the drying is completed, the control unit 25 controls the drying detection unit 29 to calculate the actual drying time td actually required for drying from the time change of the total weight as described above (the actual drying time td). Step # 31). The calculated actual drying time td is stored in the storage unit 28.

Next, the trained model 13 in consideration of the drying process this time is generated, and the trained model 13 stored in the storage unit 35 of the server 7 is updated. Specifically, first, the control unit 25 of the dryer 1 controls the communication unit 26 to transmit the actual drying time td stored in the storage unit 28 to the server 7. The server 7 receives the actual drying time td transmitted via the communication unit 31 and stores it in the storage unit 35. Then, the server 7 inputs the individual measurement input data (42j, 42T, 42s, 43t) corresponding to the received actual drying time td into the AI 40 as input data and the received actual drying time td as teacher data to the machine. Each trained model (13j, 13T, 13s, 13t) to be trained and the result of the drying process this time is generated (step # 32), and each trained model stored in the storage unit 35 is generated. (13j, 13T, 13s, 13t) is updated (step # 33).

In the above description, the types of laundry 4 are divided into jeans 4j, T-shirt 4T, sweater 4s, and towel 4t, but even if the types of laundry 4 include these other items, these are used. May include a part of. The types of laundry 4 can be arbitrarily sorted. Also in this case, the weight is measured for each type of laundry 4, and the trained model 13 is prepared for each type of laundry 4.

(2) In each of the above embodiments, the trained model 13 is generated for each specific drying device 10, but the server 7 receives learning input data 38 and actual drying time from a plurality of dryers 1 that can be connected to the server 7. You may collect td and generate a trained model 13 which is commonly used for each dryer 1. Even in this case, even if the trained model 13 corresponding to each household is not generated, the trained model 13 corresponding to a general household in a specific range can be generated, and the environment is not far from the general household. In this home, it is possible to generate a trained model 13 with relatively high accuracy.

(3) In each of the above embodiments, the control device 14 of the dryer 1 includes a processor such as a CPU. The control device 14 may be configured by hardware, but at least a part of the functions may be configured by software. In this case, the software is stored in the storage unit 28 and is operated by the processor. Further, the configuration of the functional blocks is not limited to the above configuration, and each functional block may be appropriately aggregated or subdivided. Similarly, the server 7 includes a processor such as a CPU in addition to the AI. The server 7 may also be configured by hardware, but at least some functions may be configured by software. In this case, the software is stored in the storage unit 35 and is operated by the processor. Further, the configuration of the functional blocks is not limited to the above configuration, and each functional block may be appropriately aggregated or subdivided.

(4) In each of the above embodiments, at least one of the learning device 32 and the prediction unit 33 may be provided in the dryer 1 instead of the server 7. That is, the drying device 10 may be configured not to include the server 7. In this case, the dryer 1 generates and updates the trained model 13, and the predicted drying time tp is predicted. On the contrary, the determination unit 34 may be provided in the server 7.

(5) In each of the above embodiments, the weight sensor 22 is provided on the clothesline 3, but the weight sensor 22 can be provided at any place as long as the weight of the laundry 4 can be measured. For example, the weight sensor 22 may be provided in a washing machine (not shown), or may be provided in a basket or the like that carries the laundry 4. Similarly, the image pickup device 21 may be provided at an arbitrary position as long as the laundry 4 can be photographed.

(6) In the above embodiment, the drying device 10 in which the dryer 1 is provided in the bathroom 2 has been described as an example, but the dryer 1 is provided not only in the bathroom 2 but also in an indoor room or a predetermined space divided. It can also be applied to the drying device 10. Further, the image is not limited to a still image and may be a moving image. In the case of a moving image, the image may be taken while the laundry 4 is being carried or while the laundry 4 is being dried. Further, the object is not limited to the laundry 4, and may be an object that requires drying of an object containing water or the like.

The configuration disclosed in the above embodiment (including another embodiment, the same shall apply hereinafter) can be applied in combination with the configuration disclosed in other embodiments as long as there is no contradiction. , The embodiments disclosed in the present specification are examples, and the embodiments of the present invention are not limited thereto, and can be appropriately modified without departing from the object of the present invention.

The present invention is applicable to a drying device that dries an object using a dryer.

1 Dryer 2 Bathroom 3 Clothesline (holding part)
4 Laundry (object)
7 Server 10 Drying device 13 Learned model 14 Control device 17 Temperature sensor 18 Humidity sensor 21 Imaging device 22 Weight sensor 25 Control unit 26 Communication unit 27 Information acquisition unit 28 Storage unit 32 Learning device 33 Prediction unit 35 Storage unit 38 Learning input Data 39 Teacher data 40 AI
42 Input data for prediction H0 Humidity before drying T0 Temperature before drying P Image td Actual drying time V Volume W0 Weight before drying

Claims (9)

A drying device that dries an object that has been dried in a predetermined space.
A dryer that blows air into the space and
A temperature sensor that measures the temperature in the space,
A humidity sensor that measures the humidity in the space and
An information acquisition unit that acquires information including the pre-drying temperature, which is the pre-drying temperature measured by the temperature sensor, and the pre-drying humidity, which is the pre-drying humidity measured by the humidity sensor.
The pre-drying temperature and the pre-drying humidity, the image of the object, the volume of the space and the additional information which is at least one of the pre-drying weight which is the weight of the object before drying. A learning device that generates a trained model for predicting the predicted drying time required for the dryer to dry the object by machine learning using the actual drying time as training input data and the actual drying time as teacher data.
A storage unit that stores the trained model and
The trained model is provided with a prediction unit that inputs the pre-drying temperature, the pre-drying humidity, and the additional information as input data for prediction to predict and output the predicted drying time.
The trained model is a drying device generated for each dryer. Further provided with an imaging unit for photographing the object,
The drying device according to claim 1, wherein the image is taken by the imaging unit before drying and is delivered to the information acquisition unit. A holding portion provided in the space for drying the object,
The holding portion is further provided with a weight sensor for measuring the weight of the dried object.
The weight before drying is measured by the weight sensor and passed to the information acquisition unit.
The drying apparatus according to claim 1 or 2, wherein the actual drying time is an elapsed time from the start of blowing air to the time when it is estimated that the drying is completed from the transition of the weight of the object. A holding portion provided in the space for drying the object,
A weight sensor that individually measures the weight of each of the objects dried on the holding portion, and
A determination unit for determining the type of each object from the image is further provided.
The weight before drying is measured by the weight sensor and passed to the information acquisition unit.
The learning device generates the trained model for each type of the object.
The drying apparatus according to claim 2, wherein the prediction unit predicts the time required for drying each object, and the time with the longest predicted time is set as the predicted drying time. The air volume and air temperature of the dryer can be selected within a predetermined range, and the dryer can be selected.
The training input data further includes the selected air volume and air temperature.
The drying apparatus according to any one of claims 1 to 4, further including the air volume and the blowing temperature as the prediction input data. Each time the learning device performs drying, the learning input data input to the prediction unit is newly added as the learning input data, and the actual drying time at the time of drying is newly added as the teacher data, and the learning has been completed. The drying apparatus according to any one of claims 1 to 5, wherein a model is generated and the trained model stored in the storage unit is updated with the newly generated trained model. The drying according to any one of claims 1 to 6, wherein the dryer receives the predicted drying time from the prediction unit, and automatically stops blowing when the predicted drying time elapses after the start of blowing. Device. Equipped with a server for data communication
The learning device is mounted on the server.
The information acquisition unit transmits the acquired information to the learning device, and the information acquisition unit transmits the acquired information to the learning device.
The drying device according to any one of claims 1 to 7, wherein the server stores the generated trained model in the storage unit. Equipped with a server for data communication
The learning device and the prediction unit are mounted on the server.
The information acquisition unit transmits the acquired information to the learning device and the prediction unit.
The drying apparatus according to any one of claims 1 to 7, wherein the server transmits the predicted drying time to the dryer.

Images