Prompting method and prompting system

Abstract

A presentation method in which a first prediction value is obtained and a first reliability of the first prediction value obtained for a second process is calculated (S102), the first prediction value being a prediction value of a timing at which the second process ends, which is output by inputting a parameter value related to the second process executed in the past into a prediction model that takes a parameter value related to a first process as input and outputs a timing at which the first process ends, an actual measurement value of the timing at which the second process ends is obtained (S103), a determination of a threshold value related to the first reliability is performed using the first prediction value for the second process, the first reliability, and the actual measurement value (S104), a second prediction value showing a timing at which a third process ends is obtained by inputting a parameter value related to the third process into the prediction model and a second reliability of the second prediction value is calculated (S106), presentation control of presentation of the second prediction value is performed, and notification control for notification is also performed in a case where the second reliability is less than the threshold value (S109).

Description

Technical Field

[0001] This disclosure relates to prompting methods and prompting systems. Background Technology

[0002] There is a technique for predicting the time required for a dryer to dry laundry (see, for example, Patent Document 1).

[0003] (Existing technical literature)

[0004] (Patent Documents)

[0005] Patent Document 1: Japanese Patent Application Publication No. 2021-154049 Summary of the Invention

[0006] This disclosure provides a method for appropriately indicating when a device finishes processing.

[0007] The notification method disclosed herein is a notification method for indicating the timing of the device's completion of processing. In this notification method, a first predicted value is obtained. The first predicted value is a predicted value for the timing of the device's completion of each of the multiple processes of the second process, which is performed by the device in the past, by inputting parameter values ​​related to each of the second processes into a prediction model. The prediction model is a prediction model that takes the parameter values ​​of the device related to the first process as input and the timing of the device's completion of the first process as output, calculates a first reliability of the first predicted value obtained for each of the multiple processes of the second process, obtains a measured value of the timing of the device's completion of each of the multiple processes of the second process, and uses the first predicted value, the first reliability, and the measured value for each of the multiple processes of the second process to determine a threshold related to the first reliability, thereby obtaining a second predicted value. The second predicted value is obtained by inputting parameter values ​​related to each of the multiple processes of the second process, which is performed by the device in the past, by inputting parameter values ​​related to the first process, and outputting a prediction model for the timing of the device's completion of the first process. The relevant parameter values ​​are input into the prediction model, thereby outputting a predicted value indicating the timing at which the device ends the third process. A second reliability of the second predicted value is calculated, prompting control is applied to the second predicted value, and notification control is applied if the second reliability is less than a threshold. In determining the threshold, a threshold is determined where at least one of a first condition and a second condition is satisfied. The first condition is a condition regarding a first value, which is the proportion of the second processes in which the difference between the first predicted value and the measured value is less than an allowable value more than once, in N second processes where the first reliability is lower than the threshold, where N is an integer greater than or equal to 1. The second condition is a condition regarding a second value, which is the proportion of the second processes in which the difference between the first predicted value and the measured value is higher than the threshold more than once, in M ​​second processes in which the difference between the first predicted value and the measured value is greater than the allowable value, where M is an integer greater than or equal to 1.

[0008] The notification method in this disclosure can appropriately provide notifications when the device finishes processing. Attached Figure Description

[0009] Figure 1 This is a schematic diagram illustrating the configuration of the prompting system in the implementation method.

[0010] Figure 2 This is a block diagram illustrating the function of the dryer in the embodiment.

[0011] Figure 3 This is a block diagram illustrating the functionality of the terminal in the implementation embodiment.

[0012] Figure 4This is a block diagram illustrating the functionality of the server in the implementation embodiment.

[0013] Figure 5 This is an explanatory diagram illustrating the progression of learning levels in the implementation method.

[0014] Figure 6 This is an explanatory diagram showing the progression of the degree of change over time in the implementation method.

[0015] Figure 7 This is an explanatory diagram illustrating the shift in reliability in the implementation method.

[0016] Figure 8 This is an explanatory diagram illustrating the progression of unreliability in the implementation method.

[0017] Figure 9 This is an explanatory diagram showing whether the notifications for the respective threshold candidates in the implementation are appropriate.

[0018] Figure 10 This is an explanatory diagram showing the false alarm rate and false negative rate of each threshold candidate in the implementation method.

[0019] Figure 11 This is a flowchart illustrating the processing of the prompting system in the implementation embodiment.

[0020] Figure 12 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0021] Figure 13 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0022] Figure 14 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0023] Figure 15 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0024] Figure 16 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0025] Figure 17 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0026] Figure 18 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0027] Figure 19 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0028] Figure 20 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0029] Figure 21 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0030] Figure 22 This is an explanatory diagram showing an example of an image displayed by the prompting system in the embodiment.

[0031] Figure 23 This is a schematic diagram showing the configuration of the prompting system in Modified Example 1 of the implementation.

[0032] Figure 24 This is a schematic diagram showing the configuration of the prompting system in Modified Example 2 of the implementation. Detailed Implementation

[0033] Regarding the prompting method described in the "Background Art" section, the inventors of this invention have discovered the following problems.

[0034] There is a technique for predicting the time required for a dryer to dry laundry (see, for example, Patent Document 1). This time prediction is achieved, for example, by inputting information such as the pre-drying temperature into a learned model (also called a machine learning model) generated through machine learning.

[0035] Generally speaking, predictions that utilize machine learning can, in principle, provide prediction times that are appropriate for the specific conditions of each device.

[0036] However, predictions using machine learning are not always possible with sufficiently high accuracy over the long lifespan of equipment. Predicted times obtained with low accuracy may differ from the actual drying time required by the equipment.

[0037] The main reasons for the decline in prediction accuracy include, for example, changes in the learning progress or the time-related changes in the equipment.

[0038] For example, if the input and output information related to the device used as teacher data is not provided in sufficient quantity for learning, the accuracy of predictions will decrease due to insufficient learning.

[0039] Furthermore, rapid changes in the device over time (e.g., deterioration of the hardware used for processing or decrease in sensor sensitivity) can lead to a decrease in prediction accuracy.

[0040] Therefore, when providing suggestions on predictions, if a general baseline of the prediction's accuracy can be provided, users can understand to what extent the prediction is appropriate or reliable, and such suggestions are useful.

[0041] This disclosure provides a method for appropriately indicating when a device finishes processing.

[0042] One aspect of this disclosure relates to a prompting method for indicating the timing of a device's completion of processing. In this method, a first predicted value is obtained. The first predicted value is a predicted value for the timing of the device's completion of each of the multiple processes of the second process, output by inputting parameter values ​​related to each of the second processes previously executed by the device into a prediction model. The prediction model is defined as follows: taking the parameter values ​​of the device related to the first process as input and the timing of the device's completion of the first process as output, it calculates a first reliability of the first predicted value obtained for each of the multiple processes of the second process, obtains a measured value for the timing of the device's completion of each of the multiple processes of the second process, and uses the first predicted value, the first reliability, and the measured value for each of the multiple processes of the second process to determine a threshold related to the first reliability, thereby obtaining a second predicted value. The second predicted value is obtained by inputting parameter values ​​related to each of the third processes previously executed by the device into a prediction model. The relevant parameter values ​​are input into the prediction model, and a predicted value indicating the timing of the device ending the third process is output. A second reliability of the second predicted value is calculated, and prompting control is performed on the second predicted value. Furthermore, notification control is performed if the second reliability is less than a threshold. In determining the threshold, a threshold is determined where at least one of a first condition and a second condition is satisfied. The first condition is a condition regarding a first value, which is the proportion of second processes in which the difference between the first predicted value and the measured value is less than an allowable value more than once, in N second processes where the first reliability is lower than the threshold, where N is an integer greater than or equal to 1. The second condition is a condition regarding a second value, which is the proportion of second processes in which the difference between the first predicted value and the measured value is higher than the threshold more than once, in M ​​second processes in which the difference between the first predicted value and the measured value is greater than the allowable value, where M is an integer greater than or equal to 1.

[0043] In this manner, the aforementioned prompting method, based on the device's parameter values ​​using a machine learning model, notifies the user when the predicted time for the device to finish processing is low, provided the predicted value has low reliability. Reliability refers to the degree of certainty of the predicted value; in other words, it indicates how trustworthy the predicted value is. The user can visually confirm the predicted end-time value and, upon further notification, understand that the predicted value has low reliability. Thus, the aforementioned prompting method can appropriately indicate the time for the device to finish processing.

[0044] Alternatively, the first condition may include a condition that the first value is below a specified value, or the second condition may include a condition that the second value is below a specified value.

[0045] By utilizing the condition that either the first or second value is below a predetermined value, the aforementioned prompting method more easily determines the reliability threshold for the predicted value used to determine the end timing. Therefore, the aforementioned prompting method can appropriately prompt the device to end processing by easily determining the reliability threshold.

[0046] Alternatively, the first condition may include a condition that the first value is the minimum of a plurality of candidates for a first value calculated using a plurality of candidates for the threshold, or the second condition may include a condition that the second value is the minimum of a plurality of candidates for a second value calculated using a plurality of candidates for the threshold.

[0047] By utilizing the condition that the first or second value is the minimum, the above-described prompting method makes it easier to determine the reliability threshold for the predicted value used to determine the end timing. Therefore, the above-described prompting method can appropriately prompt the device to end processing by easily determining the reliability threshold.

[0048] Alternatively, the prompt control may include control to display the second predicted value on the display screen, and the notification control may include control to change the display method of the second predicted value.

[0049] In this way, the above-described notification method notifies the user by changing the way the predicted value is displayed on the screen. By visually confirming the change in the display method, the user can easily grasp that the reliability of the predicted value is low. Therefore, by changing the display method to notify the user of the decrease in reliability, the above-described notification method can appropriately indicate the time when the device finishes processing.

[0050] Alternatively, when determining the threshold, the user may receive an instruction to raise or lower the determined threshold, and the determined threshold may be raised or lowered according to the received instruction. In the notification control, the raised or lowered threshold may be used to perform the notification control.

[0051] By means of the above method, since the reliability threshold of the predicted value used to determine the end timing is adjusted according to the instruction received from the user, notifications can be delivered at a quantity or frequency corresponding to the instruction. The user can receive notifications at a quantity or frequency corresponding to the instruction. Therefore, the above method can appropriately prompt the device to end processing at the appropriate time using notifications of an appropriate quantity or frequency.

[0052] Alternatively, when determining the threshold, the user may receive an instruction to set the threshold, and the threshold may be set according to the received instruction. In the notification control, the set threshold may be used to perform the notification control.

[0053] By means of the above method, since the aforementioned prompting method sets a threshold for determining the reliability of the predicted value for ending the processing time based on the instruction received from the user, it is possible to provide notifications at a quantity or frequency corresponding to the instruction. The user can receive notifications at a quantity or frequency corresponding to the instruction. Therefore, the aforementioned prompting method can appropriately prompt the user regarding the timing of the device's processing termination using notifications of an appropriate quantity or frequency.

[0054] Furthermore, it is also possible that, in obtaining the first reliability, the accuracy of the sensor is obtained, and the first reliability is obtained such that the higher the accuracy of the obtained sensor, the higher the reliability is, wherein the sensor is a sensor that has obtained parameter values ​​of the device related to each step of the second process; in obtaining the second reliability, the accuracy of the sensor is obtained, and the second reliability is obtained such that the higher the accuracy of the obtained sensor, the higher the reliability is, wherein the sensor is a sensor that has obtained parameter values ​​of the device related to the third process.

[0055] In this way, the above-described prompting method easily achieves reliability based on the accuracy of the sensor that obtains the parameter values. Therefore, the above-described prompting method can appropriately indicate the timing of the device's completion of processing.

[0056] Alternatively, the device may be a dryer, and the process may be the process of drying the laundry in the dryer.

[0057] In this way, the above-described prompting method can appropriately remind users when the dryer has finished drying the laundry.

[0058] One aspect of this disclosure relates to a prompting system that provides a prompting system for indicating the time when a device finishes processing. The prompting system includes a prediction unit, a reliability calculation unit, a threshold determination unit, and a prompting control unit. In this system, the prediction unit obtains a first prediction value, which is a predicted value for the time when the device finishes processing multiple times, by inputting parameter values ​​related to each of the second processes (which are previously executed by the device) into a prediction model. The prediction model is a prediction model that takes the parameter values ​​of the device related to the first process as input and the time when the device finishes the first process as output. The reliability calculation unit calculates a first reliability of the first prediction value obtained for each of the multiple second processes. The threshold determination unit obtains a measured value for the time when the device finishes processing multiple times, and uses the first prediction value, the first reliability, and the measured value for each of the multiple second processes to determine a threshold related to the first reliability. The prediction unit obtains a second prediction value. The second predicted value is a predicted value that is output by inputting parameter values ​​related to the third process as said process into the prediction model, thereby indicating the timing of the device ending the third process. The reliability calculation unit calculates the second reliability of the second predicted value. The prompt control unit performs prompt control on the second predicted value and also performs notification control for notification if the second reliability is less than the threshold. In the determination of the threshold, the threshold is determined to be the threshold in which at least one of the first condition and the second condition is satisfied. The first condition is a condition about a first value, which is the proportion of the second process in which the difference between the first predicted value and the measured value is less than the allowable value more than once in N second processes in multiple second processes where the first reliability is lower than the threshold, where N is an integer greater than or equal to 1. The second condition is a condition about a second value, which is the proportion of the second process in which the difference between the first predicted value and the measured value is higher than the threshold more than once in M ​​second processes in multiple second processes where the difference between the first predicted value and the measured value is greater than the allowable value, where M is an integer greater than or equal to 1.

[0059] The same effect as the above prompting method can be achieved by using the above method.

[0060] Hereinafter, various embodiments will be described in detail with appropriate reference to the accompanying drawings. However, unnecessary detailed descriptions may be omitted. For example, detailed descriptions of well-known matters or repeated descriptions of substantially the same structures may be omitted. This is to avoid making the following description excessively lengthy and to facilitate understanding by those skilled in the art.

[0061] Furthermore, the accompanying drawings and the following description are provided by the inventors of the present invention to enable those skilled in the art to fully understand this disclosure, and are not intended to limit the scope of the technical solutions described herein.

[0062] (Implementation Method)

[0063] In this embodiment, a method and system for appropriately indicating the timing of the device's completion of processing will be described.

[0064] Figure 1 This is a schematic diagram showing the configuration of the prompting system 1 in this embodiment.

[0065] like Figure 1 As shown, the notification system 1 includes a server 10, dryers 20A, 20B, and 20C (also referred to as dryer 20A, etc.), and terminals 30A, 30B, and 30C (also referred to as terminal 30A, etc.). The devices included in the notification system 1 are connected via a network N to enable communication.

[0066] Server 10 calculates a predicted value for the timing of the end of processing of the dryer 20A (also called the end timing) and the reliability of that predicted value. The prediction of the end timing is made, for example, by utilizing machine learning methods, although this is used as an example for illustration, but is not a limitation. Server 10 provides the calculated predicted value and reliability to terminal 30A.

[0067] Here, the terminal to which the server 10 provides predicted values ​​and reliability information for the dryer 20A, etc., can be a terminal such as terminal 30A that has established a correspondence with the dryer 20A, etc. The establishment of a correspondence between the dryer 20A, etc., and the terminal 30A, etc., is performed by the server 10.

[0068] Dryer 20A is a device that performs a processing procedure. Dryer 20A is, for example, a dryer for drying washed clothes and other laundry (or a washing machine with a dryer), although this is used as an example, it is not a limitation. The processing performed by dryer 20A is the drying of laundry (also called drying treatment). Dryer 20A can also be a freezer or air conditioner, other than those mentioned above. In the case of a freezer, the processing is ice-making. In the case of an air conditioner, the processing is heating (more specifically, for example, heating to raise the temperature of a bedroom to 25°C).

[0069] Dryers 20B and 20C are the same equipment as dryer 20A, and operate independently of dryer 20A.

[0070] Terminal 30A provides information prompts. Terminal 30A provides prompts for the predicted value of the end timing calculated by server 10. Furthermore, terminal 30A issues a notification if the reliability of the predicted value is less than a threshold.

[0071] For example, while a notification may refer to displaying an image containing information on the display screen of terminal 30A, it is not limited to this and may also be output as sound. Furthermore, while a notification may include displaying an image containing information on the display screen of terminal 30A, or changing the display method of the information displayed on the display screen of terminal 30A, it is not limited to these and may also be output as sound or have its sound output method changed. Changes to the sound output method may include, for example, changes to volume, playback speed, the presence or absence of effects, or changes to the applied effects.

[0072] Terminals 30B and 30C are the same devices as terminal 30A, and operate independently of terminal 30A.

[0073] Furthermore, although this example uses three units each of dryer 20A and terminal 30A, it is not a limitation. The number of dryers 20A and terminal 30A can be arbitrary. Moreover, the number of dryers 20A and terminal 30A can be different.

[0074] Network N can be composed of any kind of communication lines or networks, such as the Internet, mobile phone operator networks, Internet service provider access networks, or public access networks.

[0075] The functions of dryer 20A, terminal 30A, and server 10 will be described in detail below. Although this description focuses on dryer 20A, the same description applies to dryers 20B and 20C.

[0076] Figure 2 This is a block diagram illustrating the function of dryer 20A in this embodiment. Furthermore, since dryers 20B and 20C also have the same function, descriptions of dryers 20B and 20C are omitted.

[0077] like Figure 2 As shown, the dryer 20A includes a communication unit 21, a processing unit 22, and a display unit 23 as functional units.

[0078] The communication unit 21 is a communication interface capable of connecting to the network N. When the functional units of the dryer 20A communicate with other devices via the network N, the communication unit 21 transmits and receives communication packets with those devices. The communication unit 21 may be, for example, a Wi-Fi (registered trademark) communication interface.

[0079] The processing unit 22 controls the drying process performed by the dryer 20A. Specifically, the processing unit 22 has a control unit that controls the operation of the tub (generally also used as a washing tub) into which the laundry is to be processed is placed, the heat pump that draws air from the tub and generates hot air to be delivered into the tub, and the operation of the tub and the heat pump. Furthermore, the processing unit 22 has one or more sensors that detect information related to the drying process. These sensors may include, for example, a weight sensor that detects the weight of the laundry in the tub, a temperature sensor that detects the temperature of the air drawn from the tub into the heat pump (also called the intake temperature), or a temperature sensor that detects the temperature of the air delivered from the heat pump into the tub (also called the exhaust temperature). The processing unit 22 determines the drying time, which is the length of the drying process. This drying time is determined, for example, by selecting a drying time suitable for the weight of the laundry to be dried, based on the weight of the laundry at the start of drying, which is also detected by the weight sensor. The processing unit 22 begins the drying process and ends the drying process when the drying time has elapsed since the start of the drying process. Alternatively, as the drying process progresses, the processing unit 22 can continuously adjust the drying time determined at the start of the drying process based on the temperature detected by the temperature sensor.

[0080] Furthermore, in order for the server 10 to calculate the predicted value for the end timing, the processing unit 22 sends information detected by one or more sensors to the server 10. The information detected by the one or more sensors includes, for example, the weight of the laundry detected by the weight sensor, and the suction temperature or discharge temperature detected by the temperature sensor.

[0081] Display unit 23 is a display device that displays information. For example, display unit 23 displays information obtained from server 10, including a predicted value for the end timing, as an image. Furthermore, display unit 23 provides notification by changing the display of the image or the display mode. Alternatively, display unit 23 may also be a speaker that outputs the information as sound. Furthermore, display unit 23 may be a notification unit that includes at least one of the aforementioned display device and speaker.

[0082] Figure 3 This is a block diagram illustrating the functions of terminal 30A in this embodiment. Furthermore, since terminals 30B and 30C also have the same functions, descriptions of terminals 30B and 30C are omitted.

[0083] like Figure 3 As shown, the terminal 30A includes a communication unit 31, a control unit 32, and a display unit 33 as functional units.

[0084] The communication unit 31 is a communication interface capable of connecting to the network N. When the functional units of the terminal 30A communicate with other devices via the network N, the communication unit 31 performs the sending and receiving of communication packets with those devices. The communication unit 31 may be, for example, a Wi-Fi communication interface or a mobile network communication interface.

[0085] The control unit 32 controls the operation of the terminal 30A. The control unit 32 includes a processor and a memory, and controls the operation of the terminal 30A by executing a predetermined program through the processor and memory.

[0086] Display unit 33 is a display device that displays information. For example, display unit 33 displays information obtained from server 10, including a predicted value for the end timing, as an image. Display unit 33 notifies users by changing the display of the image or the display mode. Alternatively, display unit 33 may be a speaker that outputs the information as sound. Furthermore, display unit 33 may be a notification unit that includes at least one of the aforementioned display device and speaker.

[0087] Figure 4 This is a block diagram illustrating the functionality of server 10 in this embodiment.

[0088] like Figure 4 As shown, server 10 includes a communication unit 11, a training unit 12, a prediction unit 13, a reliability calculation unit 14, a threshold determination unit 15, and a display control unit 16 as functional units. The functional units of server 10 can be implemented by a processor of server 10 executing a predetermined program using memory.

[0089] Communication unit 11 is a communication interface capable of connecting to network N. When the functional units of server 10 communicate with other devices via network N, communication unit 11 performs the sending and receiving of communication packets with other devices. Communication unit 11 is, for example, an Ethernet (registered trademark) communication interface.

[0090] Training unit 12 possesses a predictive model, which serves as a machine learning model for predicting the end time, and trains the predictive model. The predictive model takes the parameter values ​​of the dryer 20A related to the drying process (equivalent to the first process) as input and the end time of the dryer 20A's completion of the drying process as output. The predictive model is constructed using machine learning training with teacher data that takes the parameter values ​​of the dryer 20A related to the drying process as input and the end time of the dryer 20A's completion of the drying process as output. The teacher data can be either teacher data that takes the parameter values ​​of the dryer 20A related to the actual drying process as input and the end time of the dryer 20A's completion of the drying process as output, or teacher data that takes the parameter values ​​of equipment of the same type as the dryer 20A as input and the end time of the aforementioned equipment's completion of the drying process as output. The predictive model is, for example, a neural network model.

[0091] The prediction unit 13 uses a prediction model to obtain a predicted value for the end timing of the actual drying process performed by the dryer 20A. Specifically, for each of the multiple drying processes previously performed by the dryer 20A (equivalent to the second process), the prediction unit 13 obtains information detected by one or more sensors of the dryer 20A, and obtains a predicted value for the end timing (also called a first predicted value) that is output by inputting the obtained information as parameter values ​​into the prediction model. Furthermore, for the drying process performed by the dryer 20A (equivalent to the third process), the prediction unit 13 obtains information detected by one or more sensors of the dryer 20A, and obtains a predicted value for the end timing (also called a second predicted value) that is output by inputting the obtained information as parameter values ​​into the prediction model.

[0092] The reliability calculation unit 14 calculates the reliability (first reliability) of the end time of the first predicted value obtained by the prediction unit 13. Furthermore, the reliability calculation unit 14 calculates the reliability (second reliability) of the end time of the second predicted value obtained by the prediction unit 13. Reliability is the degree to which a predicted value is appropriate; in other words, it represents the degree to which the predicted value can be relied upon.

[0093] The reliability calculation unit 14 obtains, for example, the accuracy of the sensor and the higher the accuracy of the sensor, the higher the reliability. This sensor is the one that obtains parameter values ​​of the dryer 20A related to the drying process.

[0094] Furthermore, when calculating reliability, the reliability calculation unit 14 can obtain, for example, an indicator showing the progress of the prediction model's learning, namely the learning degree, and an indicator showing the degree of change of the dryer 20A over time, namely the degree of change over time. Reliability is calculated using at least one of the obtained learning degree and degree of change over time. The method for calculating reliability using the learning degree and the degree of change over time will be explained later.

[0095] The threshold determination unit 15 uses the first predicted value, the first reliability, and the measured value for each of the multiple second processes to determine a threshold related to the first reliability. In determining the threshold, the threshold determination unit 15 determines a threshold in which at least one of the following first condition and the following second condition is satisfied.

[0096] The first condition relates to the false alarm rate. The false alarm rate is an indicator used in the confusion matrix and is the proportion of cases where a notification should not be made based on the difference between the predicted and measured values ​​at the end of the processing time, out of cases where a notification is made due to reliability being lower than a threshold. More specifically, the first condition is a condition concerning the false alarm rate (equivalent to the first value), which is the proportion of second processes where the difference between the first predicted value and the measured value is less than the tolerance at least once, out of N second processes where the reliability is lower than the threshold. Here, N is an integer greater than or equal to 1.

[0097] The first condition may also include a condition that the false alarm rate is below a specified value. Furthermore, the first condition may also include a condition that the false alarm rate is the minimum among multiple candidate false alarm rates calculated using multiple candidate thresholds.

[0098] Furthermore, the second condition relates to the false negative rate. The false negative rate is an indicator used in the confusion matrix, and it represents the proportion of cases where notification was not issued due to reliability exceeding a threshold, out of the cases deemed to require notification based on the difference between the predicted and measured values ​​at the end of the processing time. More specifically, the second condition (equivalent to the second value) is the proportion of the difference between the first predicted and measured values ​​in multiple processing steps where the reliability is higher than the threshold at least once, out of M processing steps where the difference exceeds the tolerance. Here, M is an integer greater than or equal to 1.

[0099] The second condition may also include a condition that the false negative rate is below a specified value. Furthermore, the second condition may also include a condition that the false negative rate is the minimum among multiple false negative rate candidates calculated using multiple candidates for the threshold.

[0100] For example, the first condition can be set to "false alarm rate is below 0.3" and the second condition can be set to "false negative rate is below 0.3".

[0101] Furthermore, for example, the first condition can be set to "false alarm rate is below 0.5", and the second condition can be set to "minimum among the candidates for false negative rate".

[0102] Furthermore, when the false positive rate and false negative rate are expressed using precision and recall, which are commonly used in confusion matrices, they can be represented as follows:

[0103] False alarm rate = 1 - precision

[0104] Missed call rate = 1 - recall

[0105] In addition to the above, a method that uses a confusion matrix to determine an appropriate threshold can also be adopted.

[0106] The display control unit 16 controls the display of a prompt message, which includes a second predicted value obtained by the prediction unit 13. This display control is equivalent to prompt control, which controls the display of the prompt message. Furthermore, the display control unit 16 not only performs prompt control but also performs notification control when the second reliability is less than a threshold. The display control unit 16 is equivalent to a prompt control unit.

[0107] Specifically, in display control, the display control unit 16 sends a prompt message to the terminal 30A, causing the terminal 30A to display the prompt message on the display unit 33. Furthermore, in notification control, the display control unit 16 controls the terminal 30A to change its display mode.

[0108] Additionally, the display control unit 16 possesses corresponding information and selects the terminal to provide the prompt information by referring to this information. This corresponding information establishes a correspondence between the dryer and the terminal that provides prompts and notifications based on the predicted value and reliability of the dryer's end time. In this corresponding information, it is assumed that dryers 20A, 20B, and 20C are corresponding to terminals 30A, 30B, and 30C, respectively. For example, the terminal corresponding to dryer 20A could be a terminal owned by the user who owns dryer 20A.

[0109] Alternatively, the display control unit 16 can send a prompt message to the dryer 20A, so that the dryer 20A can display the prompt message on the display unit 23 for control.

[0110] Alternatively, the threshold determination unit 15 may receive instructions from the user to raise or lower the determined threshold when determining the threshold, and raise or lower the determined threshold according to the received instructions. In this case, the display control unit 16 uses the raised or lowered threshold to perform notification control in notification control.

[0111] Alternatively, the threshold determination unit 15 may receive instructions from the user to set the threshold when determining the threshold, and set the threshold according to the received instructions. In this case, the display control unit 16 uses the set threshold to perform notification control in notification control.

[0112] Next, we will explain the reliability.

[0113] Here, we will explain the reliability calculated using the degree of learning and the degree of change over time.

[0114] Figure 5 This is an explanatory diagram showing the progression of the learning level α in this embodiment.

[0115] exist Figure 5 In the diagram, the horizontal axis represents the number of processing steps, and the vertical axis represents the learning level α. The learning level α is an indicator that increases as the predictive model learns through repeated processing. Figure 5 The learning level α shown is an indicator of the learning progress of the prediction model, which is a range of 0 to 1. Although this is used as an example, the range of the learning level α is not limited to the above.

[0116] like Figure 5 As shown, the learning level α has a lower value when the number of processing times is small, and a higher value when the number of processing times is large. For example, the learning level α2 in processing times P2 is higher than the learning level α1 in processing times P1.

[0117] Furthermore, as the number of processing iterations progresses to a certain level, the learning level α takes a value of 1 or close to 1, and then continues to take a value of 1 or close to 1. For example, the learning level α is approximately 1 in the number of processing iterations P3, and the learning level α is also approximately 1 when the number of processing iterations exceeds P3.

[0118] As an example, the learning level α is calculated by the reliability calculation unit 14 using at least one of the following pieces of information.

[0119] (a) is the value obtained by dividing the training error by the validation error in the cross-validation of the prediction model.

[0120] (b) is the amount of data used in training the prediction model.

[0121] (c) is the convergence rate of the prediction error of the first prediction value in the multiple processing performed by the device.

[0122] Figure 6 This is an explanatory diagram showing the progression of the degree of change over time in this embodiment.

[0123] exist Figure 6In the diagram, the horizontal axis represents the elapsed time since the manufacturing time of dryer 20A, and the vertical axis represents the degree of change over time, β. The degree of change over time, β, is an indicator showing the extent of change in the elapsed time of dryer 20A since the manufacturing time. It is 0 at the manufacturing time and increases as time goes on after the manufacturing time.

[0124] like Figure 6 As shown, the degree of change over time β has a lower value when the elapsed time is short, and a higher value when the elapsed time is long. For example, the degree of change over time β3 in elapsed time T3 is higher than the degree of change over time β2 in elapsed time T2.

[0125] As an example, the degree of change over time, β, is calculated by the reliability calculation unit 14 using at least one of the following pieces of information. Here, N is an integer greater than or equal to 2.

[0126] (d) is the average prediction error of the first predicted value in the most recent N processes among the multiple processes performed by dryer 20A.

[0127] (e) is the value obtained by dividing (d) above by the average of the prediction errors of the first prediction value in the processing of multiple devices of the same type as dryer 20A.

[0128] (f) is the standard deviation of the timing of the most recent N processing cycles performed by the dryer 20A when the dryer 20A finishes the processing.

[0129] Figure 7 This is an explanatory diagram illustrating the shift in reliability in this embodiment.

[0130] exist Figure 7 In the diagram, the horizontal axis represents the elapsed time since the manufacturing of dryer 20A, and the vertical axis represents reliability δ. Reliability δ is an indicator showing the reliability of the predicted end time of dryer 20A's processing, and it is an indicator that ranges from 0 to 1.

[0131] Reliability δ is calculated using the degree of learning α and the degree of change over time β. More specifically, the higher the degree of learning α, the higher the reliability δ, or the higher the degree of change over time β, the lower the reliability δ.

[0132] Furthermore, for ease of explanation, the following explanation will use unreliability ε as an indicator of reliability. Unreliability is an indicator whose magnitude is the reverse of reliability δ. Unreliability ε is an indicator that moves within the range of 0 to 1. The lower the unreliability ε, the more reliable the predicted value of the end time of the dryer 20A's end processing (refer to Equation 1 below).

[0133] ε=1-δ (Equation 1)

[0134] Next, the method for determining the threshold will be explained. The threshold determination unit 15 determines the threshold by selecting one threshold from a plurality of threshold candidates (also called threshold candidates).

[0135] Figure 9 This is an explanatory diagram showing whether the notifications for the respective threshold candidates in this embodiment are appropriate.

[0136] Figure 9 The diagram shows the drying time, predicted drying time, difference between the predicted and actual drying times (also referred to as "difference") for each of 10 previous drying processes performed on the dryer 20A, notification signs, and unreliability information. Figure 9 This demonstrates whether unreliable flags and notifications are appropriate for each of the aforementioned 10 drying processes, assuming there are 3 threshold candidates.

[0137] The drying time is the time required for the drying process calculated at the moment the drying process begins in dryer 20A. The drying time is an example of a measured value at the end of the drying time.

[0138] The predicted drying time is a forecast of the time required for the drying process, calculated by dryer 20A using the parameter values ​​of dryer 20A at the moment the drying process begins. The predicted drying time is an example of a predicted value for the end timing.

[0139] The difference between the drying time and the predicted drying time is the value obtained by subtracting the above-mentioned drying time from the above-mentioned predicted drying time.

[0140] The notification flag is for each of the 10 drying processes described above, indicating whether a notification should be issued based on the difference between the actual drying time and the predicted drying time. The notification flag is set to 1 (i.e., a notification should be issued) if the absolute value of the difference exceeds the tolerance, and set to 0 (i.e., no notification should be issued) if the absolute value of the difference is below the tolerance. A tolerance of 15 minutes is used as an example for explanation. Furthermore, for drying processes where the notification flag is 1, notification is not issued regardless of whether one actually receives a notification.

[0141] The unreliability refers to the unreliability of the predicted drying time. The unreliability can be calculated using the parameter values ​​of dryer 20A at the moment when the drying process begins.

[0142] For example, in drying process #1, the drying time is 90 minutes, and the predicted drying time is 95 minutes, a difference of 5 minutes. The notification flag is 0 because the absolute value of the difference, 5 minutes, is less than the tolerance of 15 minutes. The unreliability is 10.

[0143] For example, in drying process #2, the drying time is 100 minutes, and the predicted drying time is 120 minutes, a difference of 20 minutes. The notification flag is 1 because the absolute value of the difference, 20 minutes, exceeds the tolerance of 15 minutes. The unreliability is 20.

[0144] For example, in drying process #3, the drying time is 80 minutes, and the predicted drying time is 70 minutes, with a difference of -10 minutes. The notification flag is 0 because the absolute value of the difference, 10 minutes, is less than the allowable deviation of 15 minutes. The unreliability is 25.

[0145] Next, we will explain whether the notifications for each threshold candidate are appropriate. Here, we will use the case where there are three threshold candidates: 20, 50, and 80, as examples. However, threshold candidates are not limited to these three; multiple threshold candidates can be used, ranging from the minimum unreliable value to the maximum value. For example, threshold candidates can also be all integer values ​​ranging from the minimum value of 0 to the maximum value of 100.

[0146] The unreliability flag is used for each of the 10 drying processes, indicating whether the unreliability is above the threshold when the threshold candidate is used as the threshold. The unreliability flag is set to 1 if the unreliability in the drying process is above the threshold, and set to 0 otherwise.

[0147] Regarding the drying process when the unreliability flag is 1, the "whether the notification is appropriate" flag indicates whether the threshold determination unit 15 has made a judgment on whether to issue a notification. Regarding the drying process when the unreliability flag is 1, the "whether the notification is appropriate" flag indicates whether issuing a notification based on the difference is appropriate. Regarding the drying process when the unreliability flag is 1, the "whether the notification is appropriate" flag is set to 1 if the notification flag is 1, and set to 0 otherwise. Furthermore, the "whether the notification is appropriate" flag is not defined for the drying process when the unreliability flag is 0, and is indicated by "-" in the figure.

[0148] For example, in drying process #1, the unreliability flag is 0 when 20, which is a threshold candidate, is used as the threshold. This is because 10, which is unreliable, is less than 20, which is a threshold candidate. Furthermore, in drying process #1, the cases where 50 or 80, which are threshold candidates, are used as the threshold are the same as the cases where 20, which is a threshold candidate, is used as the threshold.

[0149] For example, in drying process #2, the unreliability flag is 1 when 20, which is a threshold candidate, is used as the threshold. This is because 20, which is unreliable, is greater than 20, which is a threshold candidate. Furthermore, in drying process #2, since the unreliability flag is 1 and the notification flag is 1, the appropriateness of the notification is 1.

[0150] Furthermore, in drying process #2, the unreliability flag is 0 when 50 or 80, which are candidates for the threshold, are used as the threshold. This is because 20, which is unreliable, is less than 50 or 80, which are candidates for the threshold.

[0151] For example, in drying process #3, the unreliability flag is 1 when 20, which is a threshold candidate, is used as the threshold. This is because 25, which is unreliable, is greater than 20, which is a threshold candidate. Furthermore, in drying process #3, since the unreliability flag is 1 and the notification flag is 0, the appropriateness of the notification is 0.

[0152] Furthermore, in drying process #3, the unreliability flag is 0 when 50 or 80, which are candidates for the threshold, are used as the threshold. This is because 25, which is unreliable, is less than 50 or 80, which are candidates for the threshold.

[0153] Figure 10 This is an explanatory diagram showing the false alarm rate and false negative rate of each threshold candidate in this embodiment.

[0154] Figure 10 The number of times each threshold candidate was notified, the number of times the notification was deemed appropriate, the false alarm rate, and the false negative rate are shown.

[0155] The number of notifications indicates the number of drying processes that should be notified, assuming that this threshold candidate is used as the threshold. Figure 9 The number of drying processes when the unreliable flag is 1 in the threshold candidate case shown.

[0156] The number of times that a notification is deemed appropriate is, regarding the drying process that the threshold determination unit 15 determines should be notified, is assumed to be... Figure 9 The number of drying processes when the notification is appropriate in the threshold candidate case shown is 1.

[0157] The false alarm rate is the false alarm rate when the threshold candidate is used as the threshold, and is calculated by the following (Equation 2).

[0158] False alarm rate = ((number of times a notification was processed) - (number of times a notification was processed when it was appropriate)) / (number of times a notification was processed) (Equation 2)

[0159] For example, for 20 as a threshold candidate, the calculated false alarm rate is (9-4) / 9≒0.56.

[0160] The false negative rate, shown as the rate of false negatives when the threshold candidate is used as the threshold, is calculated using Equation 3 below. Furthermore, the number of processes for which notification should be made is the number of drying processes when the notification flag is 1, which is 4 in this case.

[0161] The underreporting rate = (number of times notifications should have been processed) - (number of times notifications were appropriate) / (number of times notifications should have been processed) (Equation 3)

[0162] For example, for 50 as a candidate threshold, the calculated false negative rate is (4-3) / 4 = 0.25.

[0163] in addition, Figure 10 The multiple threshold candidates shown each have their own false alarm rate and false negative rate, which are equivalent to multiple candidates for false alarm rate and false negative rate, respectively.

[0164] The processing of the prompting system 1 as described above will be explained.

[0165] Figure 11 This is a flowchart illustrating the processing of the prompting system 1 in this embodiment.

[0166] In step S101, the prediction unit 13 obtains parameter values ​​related to each of the multiple drying processes (equivalent to the second process) previously performed by the dryer 20A.

[0167] In step S102, the prediction unit 13 obtains a predicted value for the end time of the drying process previously performed by the dryer 20A by inputting the parameter values ​​obtained in step S101 into the prediction model. The reliability calculation unit 14 obtains the reliability of the predicted value obtained by the prediction unit 13 in step S101.

[0168] In step S103, the threshold determination unit 15 obtains the measured value of the end time of the end drying process previously performed by the dryer 20A.

[0169] In step S104, the threshold determination unit 15 determines the threshold for reliability.

[0170] In step S105, the prediction unit 13 obtains parameter values ​​related to each drying process (equivalent to the third process) performed by the dryer 20A.

[0171] In step S106, the prediction unit 13 obtains a predicted value for the end timing of the drying process performed by the dryer 20A by inputting the parameter values ​​obtained in step S105 into the prediction model. The reliability calculation unit 14 obtains the reliability of the predicted value obtained by the prediction unit 13 in step S106.

[0172] In step S107, the display control unit 16 controls the display of the predicted value obtained in step S106.

[0173] In step S108, the display control unit 16 determines whether the reliability obtained in step S106 is higher than the threshold determined in step S104. If the reliability is higher than the threshold ("Yes" in step S108), the process proceeds to step S110; otherwise ("No" in step S108), the process proceeds to step S109.

[0174] In step S109, the display control unit 16 performs notification control, and then ends. Figure 11 The series of processes shown.

[0175] In step S110, the display control unit 16 does not perform notification control (in other words, it suppresses the execution of notification control), and then ends. Figure 11 The series of processes shown.

[0176] Furthermore, when the prediction unit 13 obtains parameter values ​​in step S101, it can also obtain parameter values ​​related to multiple drying processes previously performed by the dryer 20A, which are also drying processes with the same operating mode as the processes to be performed in the future. In this way, the threshold can be determined based on the drying processes with the same operating mode (see step S104), and the appropriateness of the threshold determined for the drying processes with the aforementioned operating mode can be improved.

[0177] pass Figure 11 The processing shown indicates that system 1 can appropriately prompt when the device finishes processing.

[0178] The following is a detailed explanation of the display performed by prompting system 1.

[0179] The display made by the prompting system 1 is, for example, the display made by the display unit 33 provided by the terminal 30A. Although this case is described as an example, the display may also be made by the display unit 23 of the dryer 20A.

[0180] Terminal 30A receives a prompt message from server 10, including a second predicted value calculated by server 10, and displays the second predicted value as an image on the display screen. When making a notification, terminal 30A will also display the notification on the display screen, or change the display method corresponding to the notification.

[0181] Figure 12 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0182] Figure 12The image shown is a predicted value for the end timing of the drying process performed by dryer 20A. Additionally, Figure 12 It is an image that does not include notifications.

[0183] Figure 12 The images shown include image 51.

[0184] Image 51 is an image that displays the predicted end time of the drying process performed by dryer 20A as the remaining time from the current moment. Specifically, image 51 includes a rectangular solid line border and the string "1:40 remaining." within the box, indicating that the predicted end time of the drying process is 1 hour and 40 minutes from the current moment.

[0185] I was notified Figure 12 The image shown allows users to grasp the predicted value of the end timing.

[0186] Figure 13 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment. Furthermore, images identical to those already described are assigned the same symbols, and detailed descriptions may be omitted.

[0187] Figure 13 The image shown illustrates the predicted end time of the drying process performed by dryer 20A, and further shows that the predicted value is obtained through AI (Artificial Intelligence) based prediction.

[0188] Figure 13 The images shown include image 51 (with) showing the predicted value of the end timing of the drying process performed by dryer 20A. Figure 12 (Same), and image 52 showing the situation where the predicted value is obtained through AI-based prediction.

[0189] Specifically, image 52 includes a circular border and a string such as "AI" within the border.

[0190] I was notified Figure 13 The image shown allows users to understand the predicted value of the end timer and that this prediction is obtained through AI-based forecasting.

[0191] Figure 14 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0192] Figure 14 The image shown illustrates the predicted end time of the drying process performed by dryer 20A, and further includes a notification.

[0193] Figure 14 The images shown include image 51A showing the predicted value of the end timing of the drying process performed by dryer 20A, and image 52A showing that the predicted value is obtained through AI-based prediction.

[0194] Image 51A is identical to Image 51, displaying a predicted value of the end time of the drying process performed by dryer 20A as the remaining time from the current moment. However, the rectangular border included in Image 51A is a lighter color than the border included in Image 51. This lighter border color is an example of a change in the display method corresponding to a notification.

[0195] Image 52A is identical to Image 52, illustrating a situation where the predicted value is obtained through AI-based prediction. However, the circular border included in Image 52A is a lighter color than the border included in Image 52. This lighter border color is an example of a change in the display method corresponding to a notification.

[0196] I was notified Figure 14 The image shown allows users to grasp the situation regarding the predicted value at the end of the timing and the low reliability of that prediction.

[0197] Figure 15 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0198] Figure 15 The image shown illustrates the predicted end time of the drying process performed by dryer 20A, and further includes notification.

[0199] Figure 15 The images shown include image 51B, which shows the predicted value of the end timing of the drying process performed by dryer 20A, and image 52B, which shows that the predicted value is obtained through AI-based prediction.

[0200] Image 51B is identical to Image 51, displaying a predicted value of the end time of the drying process performed by dryer 20A as the remaining time from the current moment. However, the rectangular border included in Image 51B is represented by dashed lines. The dashed lines indicate that Image 51B is blinking, serving as an example of a change in display method corresponding to notification.

[0201] Image 52B is identical to image 52, illustrating a situation where the predicted value is obtained through AI-based prediction. However, the circular border included in image 52B is represented by dashed lines. The dashed lines indicate that image 51B is blinking, serving as an example of a change in the display method corresponding to a notification.

[0202] I was notified Figure 15 The image shown allows users to grasp the situation regarding the predicted value at the end of the timing and the low reliability of that prediction.

[0203] Figure 16 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0204] Figure 16 The image shown includes a notification of the predicted value for the end timing of the drying process performed by dryer 20A.

[0205] Figure 16 The images shown include image 51C and image 52C. Image 51C is positioned to display the predicted value of the end timing of the drying process performed by the dryer 20A (e.g., image 51), and image 52C is positioned to display the predicted value obtained through AI-based prediction (e.g., image 52).

[0206] Image 51C includes a rectangular solid-line border and symbols such as "..." within that border. This is because the predicted value of the end timing of the drying process is unreliable, so the above symbols are displayed instead of the predicted value. Displaying symbols such as "..." is an example of a change in the display method corresponding to a notification. Alternatively, symbols such as "..." can also be animated by repeatedly displaying and not displaying each of multiple "..." symbols in sequence to make the "..." flow in the left and right directions.

[0207] Image 52C includes a circular border and symbols such as "..." within that border. This is because the predicted value of the end time of the drying process is of low reliability, so the above symbols are displayed instead of "AI". The explanation of the symbols such as "..." is the same as that for image 51B above.

[0208] I was notified Figure 16 The image shown allows users to grasp the low reliability of the predicted values ​​at the end timing.

[0209] Figure 17 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0210] Figure 17 The image shown illustrates the predicted end time of the drying process performed by dryer 20A, and further includes notification.

[0211] Figure 17The images shown include image 51D showing the predicted value of the end timing of the drying process performed by dryer 20A, and image 52D showing that the predicted value is obtained through AI-based prediction.

[0212] Image 51D is identical to Image 51, displaying the predicted end time of the drying process performed by dryer 20A as the remaining time from the current moment. However, in Image 51D, an "×" symbol is also displayed along with the predicted value. Displaying the "×" symbol is an example of a display method corresponding to changes and notifications. Furthermore, the "×" symbol can be displayed across the entire... Figure 17 The area inside the box shown can also be displayed as a portion of the area inside the box instead of spanning the entire area inside the box.

[0213] Image 52D is identical to Image 52, illustrating a situation where the predicted value is obtained through AI-based prediction. However, in Image 52D, a "?" symbol is displayed alongside the string "AI". The display of the "?" symbol is an example of a change in the display method corresponding to a notification. Additionally, the "?" symbol is displayed... Figure 17 The portion within the box shown can also be displayed as spanning the entire box instead of just a portion within the box.

[0214] I was notified Figure 17 The image shown allows users to grasp the situation regarding the predicted value at the end of the timing and the low reliability of that prediction.

[0215] Figure 18 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0216] Figure 18 The image shown illustrates the predicted end time of the drying process performed by dryer 20A, and further includes notification.

[0217] Figure 18 The images shown include image 51 showing the predicted value of the end timing of the drying process performed by dryer 20A, image 52 showing that the predicted value is obtained through AI-based prediction, and image 53 notifying the low reliability of the predicted value.

[0218] Image 53 is an image that notifies users of the low reliability of the predicted value. Image 53 includes the string "AI recalculating." The string "AI recalculating" indirectly indicates that the predicted value is unreliable. Using a direct string like "Low reliability of predicted value" could mislead users into believing that the AI ​​technology itself is unreliable; therefore, this image avoids that misunderstanding. Alternatively, the string could also be a direct statement like "Low reliability of predicted value."

[0219] I was notified Figure 18 The image shown allows users to grasp the situation regarding the predicted value at the end of the timing and the low reliability of that prediction.

[0220] Figure 19 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0221] Figure 19 The image shown illustrates the predicted end time of the drying process performed by dryer 20A, and further includes notification.

[0222] Figure 19 The images shown include image 51 showing the predicted value of the end timing of the drying process performed by dryer 20A, image 52 showing that the predicted value is obtained through AI-based prediction, and image 54 notifying the low reliability of the predicted value.

[0223] Image 54 is an image of an icon displayed in the notification area (so-called status bar) of terminal 30A. Imagine a user who has visually recognized image 54 performs an operation on terminal 30A to browse the notification corresponding to that image, such as image 53 (see reference). Figure 18 ) is displayed.

[0224] I was notified Figure 19 The image shown allows users to grasp the situation regarding the predicted value at the end of the timing and the low reliability of that prediction.

[0225] The following explains the threshold for adjusting the reliability of the predicted value. Although the threshold for the reliability of the predicted value is calculated by the threshold determination unit 15 as described above, the threshold calculated by the threshold determination unit 15 can be adjusted according to the user's specifications. The adjustment for accepting the image specified by the user and corresponding to the user's instructions will be explained.

[0226] Figure 20 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0227] Figure 20 The image shown is a specified image displayed by system 1 to accept the threshold set by the user.

[0228] Figure 20 The image shown includes image 61 and a specified image 62 for accepting the threshold, image 61 including the message.

[0229] Image 61 includes a message for receiving a specified instruction, which is an instruction for adjusting a threshold. The message includes a string such as "Adjust recalculation sensitivity", a string describing the adjustment of the sensitivity such as "Adjust the sensitivity for recalculating the remaining drying time", and a string describing the effect of the adjustment such as "The higher the sensitivity, the more frequently recalculated the data, and the more notifications are sent".

[0230] Image 62 includes a slider 63 and a handle 64. Regarding the slider 63, the right end corresponds to a higher threshold, while the left end corresponds to a lower threshold. The user can specify the threshold by moving the handle 64 left or right along the slider 63.

[0231] Alternatively, the threshold can be specified in a relative manner, such as by raising or lowering only a predetermined value portion of the threshold calculated by the threshold determination unit 15. Furthermore, the threshold can also be specified absolutely, such as by setting a predetermined value regardless of how the threshold calculated by the threshold determination unit 15 is set.

[0232] I was notified Figure 20 The images shown allow users to not only understand what adjusting the recalculated sensitivity means and what its effects are, but also to specify the sensitivity adjustment.

[0233] Figure 21 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0234] Figure 21 The image shown is a prompt image displayed by system 1 indicating acceptance of the threshold specified by the user.

[0235] Figure 21 The image shown includes image 71 and a specified image 72 for accepting a threshold, image 71 including a message.

[0236] Image 71 includes a message for receiving a specified information, which is a specification for adjusting a threshold. The message includes a string such as "Adjust the frequency of accuracy reduction notifications" and a question such as "How many notifications would you like to receive regarding the remaining drying time?" as a query to the user regarding this frequency.

[0237] Image 72 includes a slider 73 and a handle 74. Regarding the slider 73, the right end corresponds to increasing the frequency of receiving notifications ("want to receive many"), and the left end corresponds to decreasing the frequency of receiving notifications ("don't want to receive many"). The slider 73 accepts five levels from 1 to 5. The user can adjust the notification frequency by moving the handle 74 left or right along the slider 73.

[0238] In addition, the specification of the threshold and Figure 20 The same applies to the case of [the previous sentence], which can be either a relative specification or an absolute specification.

[0239] I was notified Figure 21 The image shown allows users to specify the frequency at which they wish to receive notifications.

[0240] Figure 22 This is an explanatory diagram showing an example of an image displayed by the prompting system 1 in this embodiment.

[0241] Figure 22 The image shown is displayed by system 1 to receive user feedback on the notification.

[0242] Figure 22 The images shown include image 81 and image 82 for receiving feedback on the notification, image 81 including the message.

[0243] Image 81 includes a specified message for receiving a notification, which is a specification for evaluating the notification (more specifically, evaluating the quantity of the notification). The message includes a string such as "Please evaluate the notification," a string as a question to the user making the evaluation such as "Was the notification about the remaining drying time appropriate?", and a string as a description of the effect of the specified evaluation such as "AI learns and makes an appropriate notification."

[0244] Image 82 includes a slider 83 and a handle 84. Regarding the slider 83, the right end corresponds to a notification amount less than the user expects ("too little"), and the left end corresponds to a notification amount more than the user expects ("too much"). The slider 83 accepts five rating options. The user can rate the notification by moving the handle 84 left and right along the slider 83.

[0245] In addition, the specification of the threshold and Figure 20 The same applies to the case of [the previous sentence], which can be either a relative specification or an absolute specification.

[0246] I was notified Figure 22 The image shown allows users to rate the notification.

[0247] In addition, some or all of the functions of the server 10 in the prompting system 1 may also be provided by the dryer 20A or the terminal 30A.

[0248] (Modification 1 of the implementation method)

[0249] In this variation, a method and system for appropriately indicating the timing of the device's completion of processing will be described in a manner different from the embodiments described above.

[0250] Figure 23 This is a schematic diagram showing the configuration of the prompting system 1D in this modified example.

[0251] like Figure 23 As shown, the system 1D includes a server 10D, a dryer 20A, and a terminal 30D.

[0252] Server 10D is a server used in place of server 10 in embodiment 1. Server 10D includes a communication unit 11 and a training unit 12. The communication unit 11 and training unit 12 of server 10D are the same as those of server 10 in embodiment 1.

[0253] The dryer 20A is the same as the dryer 20A in Embodiment 1.

[0254] Terminal 30D is a terminal used to replace terminal 30A in Embodiment 1. Terminal 30D includes a communication unit 31, a control unit 32, a display unit 33, a prediction unit 34, a reliability calculation unit 35, a threshold determination unit 36, and a display control unit 37.

[0255] The communication unit 31, control unit 32, and display unit 33 are the same as those of the terminal 30A in Embodiment 1.

[0256] Furthermore, the prediction unit 34, reliability calculation unit 35, threshold determination unit 36, and display control unit 37 are the same as the prediction unit 13, reliability calculation unit 14, threshold determination unit 15, and display control unit 16 provided by the server 10 in the embodiment.

[0257] Furthermore, the number of dryers or terminals can be arbitrary. Also, the number of dryers and terminals can be different.

[0258] Alternatively, the prompting system 1D in this variation can be considered as a prompting system in which the functions of the prediction unit 13, reliability calculation unit 14, threshold determination unit 15, and display control unit 16 of the prompting system 1 in the embodiment are transferred from the server to the terminal.

[0259] By doing so, the terminal 30D can perform the calculation of the predicted values, reliability and thresholds using the prediction model, as well as the display control, thereby reducing the processing load on the server 10D.

[0260] (Modification 2 of the implementation method)

[0261] In this variation, a method and system for appropriately indicating the timing of the device's completion of processing will be described in a manner different from the embodiments described above.

[0262] Figure 24 This is a schematic diagram showing the configuration of the prompting system 1E in this modified example.

[0263] like Figure 24 As shown, system 1E includes dryer 20A and terminal 30E. System 1E does not include a server.

[0264] The dryer 20A is the same as the dryer 20A in Embodiment 1.

[0265] Terminal 30E is a terminal used to replace terminal 30A in Embodiment 1. Terminal 30E includes a communication unit 31, a control unit 32, a display unit 33, a prediction unit 34, a reliability calculation unit 35, a threshold determination unit 36, a display control unit 37, and a training unit 38.

[0266] The communication unit 31, control unit 32, and display unit 33 are the same as those of the terminal 30A in Embodiment 1.

[0267] Furthermore, the prediction unit 34, reliability calculation unit 35, threshold determination unit 36, display control unit 37, and training unit 38 are the same as the prediction unit 13, reliability calculation unit 14, threshold determination unit 15, display control unit 16, and training unit 12 provided by the server 10 in the embodiment.

[0268] Furthermore, the number of dryers or terminals can be arbitrary. Also, the number of dryers and terminals can be different.

[0269] Alternatively, the prompting system 1E in this modified example can be considered as a prompting system in which the functions of the training unit 12, prediction unit 13, reliability calculation unit 14, threshold determination unit 15, and display control unit 16 of the prompting system 1 in the embodiment are transferred from the server to the terminal.

[0270] By doing so, the terminal 30E can perform the training of the prediction model, the calculation of the predicted values ​​of the prediction model, the calculation of reliability and thresholds, and the display control, thereby achieving the effect of not needing the server 10.

[0271] As described above, the prompting method of the above embodiments or variations, based on the parameter values ​​of the device utilizing a machine learning model, notifies the user when the predicted value of the device's end-of-process timing is low. Reliability refers to the degree of appropriateness of the predicted value; in other words, it indicates how much the predicted value can be trusted. The user can visually confirm the predicted end-of-process timing and, upon further notification, understand that the reliability of the predicted value is low. Thus, the above prompting method can appropriately indicate the end-of-process timing of the device.

[0272] Furthermore, by utilizing the condition that the first or second value is below a predetermined value, the aforementioned prompting method makes it easier to determine the reliability threshold for the predicted value used to determine the end timing. Therefore, the aforementioned prompting method can appropriately prompt the device to end processing by easily determining the reliability threshold.

[0273] Furthermore, by utilizing the condition that the first or second value is the minimum, the aforementioned prompting method makes it easier to determine the reliability threshold for the predicted value used to determine the end timing. Therefore, the aforementioned prompting method can appropriately prompt the device to end processing by easily determining the reliability threshold.

[0274] Furthermore, the aforementioned notification method notifies the user by changing the way the predicted value is displayed on the screen. By visually confirming the change in the display method, the user can easily grasp that the reliability of the predicted value is low. Therefore, by changing the display method to notify the user of the decrease in reliability, the aforementioned notification method can appropriately indicate the time when the device finishes processing.

[0275] Furthermore, since the aforementioned notification method adjusts the reliability threshold of the predicted value used to determine the end timing based on the instruction received from the user, it can provide notifications at a quantity or frequency corresponding to the instruction. The user can receive notifications at a quantity or frequency corresponding to the instruction. Therefore, the aforementioned notification method can appropriately prompt the device to end processing at the appropriate time using notifications of an appropriate quantity or frequency.

[0276] Furthermore, since the aforementioned notification method sets a threshold for determining the reliability of the predicted value for ending the processing time based on the instruction received from the user, it can provide notifications at a quantity or frequency corresponding to the instruction. The user can receive notifications at a quantity or frequency corresponding to the instruction. Therefore, the aforementioned notification method can appropriately prompt the user regarding the timing of the device's processing termination using notifications of an appropriate quantity or frequency.

[0277] Furthermore, the aforementioned notification method readily achieves reliability based on the accuracy of the sensor used to obtain the parameter values. Therefore, the aforementioned notification method can appropriately indicate the timing of the device's completion of processing.

[0278] Furthermore, the above-mentioned prompting method can appropriately remind you of the time when the dryer finishes drying the laundry.

[0279] As described above, embodiments serving as technical examples in this disclosure have been presented. Drawings and detailed descriptions are provided for this purpose.

[0280] Therefore, the components described in the accompanying drawings and detailed description include not only those necessary for solving the problem, but also, for the purpose of illustrating the above-described technology, components that are not essential for solving the problem. Therefore, one should not immediately assume that these non-essential components are essential simply because they are described in the accompanying drawings or detailed description.

[0281] Furthermore, since the above-described embodiments are examples used to illustrate the technology of this disclosure, various changes, substitutions, additions, omissions, etc., can be made within the required scope or the equivalent scope.

[0282] This disclosure is applicable to systems that predict the end timing of processing of a device.

[0283] Symbol Explanation

[0284] 1, 1D, 1E prompting system

[0285] 10, 10D server

[0286] 11, 21, 31 Department of Communications

[0287] Training Department 12, 38

[0288] 13,34 Prediction Department

[0289] 14, 35 Reliability Calculation Department

[0290] 15, 36 Threshold Decision Section

[0291] 16, 37 Display Control Section

[0292] 20A, 20B, 20C dryers

[0293] 22 Processing Department

[0294] Display sections 23 and 33

[0295] Terminals 30A, 30B, 30C, 30D, and 30E

[0296] 32 Control Department

[0297] 51, 51A, 51B, 51C, 51D, 52, 52A, 52B, 52C, 52D, 53, 54, 61, 62,

[0298] Images 71, 72, 81, and 82

[0299] 63, 73, 83 sliders

[0300] 64, 74, 84 handles

[0301] N Network

Claims

1. A notification method for indicating when a device has finished processing. In the aforementioned prompting method, A first predicted value is obtained, which is a predicted value of the timing at which the device ends each of the multiple second processes by inputting parameter values ​​related to each of the multiple processes previously executed by the device into a prediction model. The prediction model is a prediction model that takes the parameter values ​​of the device related to the first process as input and takes the timing at which the device ends the first process as output. Calculate the first reliability of the first prediction value obtained for each of the multiple instances of the second processing. The measured time value is obtained for each time the device completes the second process multiple times. Using the first predicted value, the first reliability, and the measured value for each of the multiple second processes, a threshold related to the first reliability is determined. A second predicted value is obtained, which is a predicted value indicating the timing at which the device ends the third process, by inputting parameter values ​​related to the third process as said process into the prediction model. Calculate the second reliability of the second predicted value. The system provides prompt control for the second predicted value, and further provides notification control if the second reliability is less than the threshold. In the determination of the threshold, The threshold that determines whether at least one of the first and second conditions is satisfied. The first condition is a condition concerning a first value, which is the proportion of the second processing in which the difference between the first predicted value and the measured value is less than an allowable value more than once, out of N second processing sessions in which the first reliability is lower than the threshold, where N is an integer greater than or equal to 1. The second condition is a condition concerning a second value, which is the proportion of the difference between the first predicted value and the measured value in M ​​second processes in which the first reliability is higher than the threshold by more than one time, and M is an integer greater than 1.

2. The prompting method as described in claim 1, The first condition includes a condition where the first value is below a specified value, or the second condition includes a condition where the second value is below a specified value.

3. The prompting method as described in claim 1 or 2, The first condition includes a condition that the first value is the minimum of a plurality of candidates for a first value calculated using a plurality of candidates for the threshold, or the second condition includes a condition that the second value is the minimum of a plurality of candidates for a second value calculated using a plurality of candidates for the threshold.

4. The prompting method as described in claim 1 or 2, The prompt control includes controlling the display of the second predicted value on the display screen. The notification control includes control over how the second predicted value is displayed.

5. The prompting method as described in claim 1 or 2, When determining the threshold Furthermore, the user receives an instruction to raise or lower the determined threshold. According to the received instructions, the determined threshold is raised or lowered. In the notification control, The notification control is performed using the threshold value after it has increased or decreased.

6. The prompting method as described in claim 1 or 2, When determining the threshold Furthermore, the user accepts an instruction to set the threshold. Set the threshold according to the received instructions. In the notification control, The notification control is performed using the set threshold.

7. The prompting method as described in claim 1 or 2, Further in the aforementioned prompting method, When the first reliability is obtained, The accuracy of the sensor is obtained, and a first reliability is obtained whereby higher accuracy of the obtained sensor corresponds to higher reliability. This reliability is achieved by obtaining parameter values ​​of the device relevant to each step of the second process. When the second reliability is obtained, The accuracy of the sensor is obtained, and a second reliability is obtained, wherein the higher the accuracy of the obtained sensor, the higher the reliability is, wherein the sensor is the sensor that obtains the parameter values ​​of the device in relation to the third processing.

8. The prompting method as described in claim 1 or 2, The equipment is a dryer. The process described is the drying of the washed clothes by the dryer.

9. A notification system that provides a timed notification when a device has finished processing. The alert system includes a prediction unit, a reliability calculation unit, a threshold determination unit, and an alert control unit. In the notification system, The prediction unit obtains a first prediction value, which is a prediction value of the timing at which the device ends each of the multiple second processes by inputting parameter values ​​related to each of the multiple processes previously executed by the device into a prediction model. The prediction model is a prediction model that takes the parameter values ​​of the device related to the first process as input and takes the timing at which the device ends the first process as output. The reliability calculation unit calculates the first reliability of the first predicted value obtained for each of the multiple second processes. The threshold determination unit obtains the measured value of the timing for each of the multiple second processes completed by the device, and uses the first predicted value, the first reliability, and the measured value for each of the multiple second processes to determine a threshold related to the first reliability. The prediction unit obtains a second prediction value, which is a prediction value indicating the timing at which the device ends the third process, by inputting parameter values ​​related to the third process as the process into the prediction model. The reliability calculation unit calculates the second reliability of the second predicted value. The prompt control unit provides prompt control for the second predicted value, and also provides notification control if the second reliability is less than the threshold. In the determination of the threshold, The threshold that determines whether at least one of the first and second conditions is satisfied. The first condition is a condition concerning a first value, which is the proportion of the second processing in which the difference between the first predicted value and the measured value is less than an allowable value more than once, out of N second processing sessions in which the first reliability is lower than the threshold, where N is an integer greater than or equal to 1. The second condition is a condition concerning a second value, which is the proportion of the difference between the first predicted value and the measured value in M ​​second processes in which the first reliability is higher than the threshold by more than one time, and M is an integer greater than 1.

Images