Mold development estimation device, mold suppression assistance system, mold development estimation method, and program

Abstract

A mold development estimation device (1) according to the present disclosure comprises an acquisition unit (11) for acquiring input data including at least object surface temperatures in a plurality of locations within an area of interest for which mold development is to be estimated, an estimation unit (13) for generating, from the input data, mold development risk information associating estimation results of mold development risks in the plurality of locations with positions, by executing an estimation process using machine learning for estimating the mold development risks in the plurality of locations from the input data, and an output unit (14) for outputting the mold development risk information

Description

Mold Growth Estimation Apparatus, Mold Growth Suppression Support System, Mold Growth Estimation Method and Program

[0001] The present disclosure relates to a mold growth estimation apparatus that estimates the risk of mold growth, a mold growth suppression support system, a mold growth estimation method, and a program.

[0002] Mold may grow in buildings depending on the environment. Since the growth of mold is not preferable from the viewpoints of hygiene and designability, it is desired to suppress the growth of mold. Patent Document 1 discloses a mold suppression system that suppresses the growth of indoor mold. The mold suppression system described in Patent Document 1 predicts the indoor temperature and absolute humidity after a predetermined time by inputting the indoor temperature, absolute humidity, outside air temperature, and forecast temperature after the predetermined time into a trained prediction model, and estimates a mold index after the predetermined time from the prediction result. The mold suppression system described in Patent Document 1 prevents the growth of mold in indoor environments such as houses and commercial facilities by determining an operation schedule for humidity adjustment according to the transition of the predicted mold index.

[0003] Japanese Unexamined Patent Application Publication No. 2021-4703

[0004] Since the system described in Patent Document 1 targets small spaces such as living rooms and bedrooms, the entire indoor space is set as one prediction target, and only one mold index is output for each prediction time of the target. On the other hand, the risk of mold growth is generally not uniform indoors, and even in a small space, it is highly likely that there is some deviation in the mold growth risk. Furthermore, in large spaces such as stores, factories, conference rooms, hospitals, public facilities, apartment buildings, and schools, there is a higher possibility that there is deviation in the mold growth risk within the building. In order to take measures to suppress mold growth, it is desirable to enable a user to grasp areas where mold is estimated to be likely to grow. However, in the mold suppression system described in Patent Document 1, since only one mold index is output for each prediction time of the target, the user cannot grasp areas where mold is estimated to be likely to grow.

[0005] The present disclosure has been made in view of the above, and an object thereof is to obtain a mold growth estimation apparatus that can output information that allows a user to grasp areas where mold is estimated to be likely to grow.

[0006] To solve the above-mentioned problems and achieve the objective, the mold growth estimation device according to this disclosure includes: an acquisition unit that acquires input data including at least the surface temperature of articles at multiple locations within a target area for which mold growth is to be estimated; an estimation unit that generates mold growth risk information from the input data, which is information in which the estimated results of the mold growth risk at multiple locations are associated with locations, by performing an estimation process using machine learning to estimate the risk of mold growth at multiple locations from the input data; and an output unit that outputs the mold growth risk information.

[0007] The mold growth estimation device described herein has the effect of being able to output information that allows users to understand areas where mold is likely to grow.

[0008] Figure showing an example configuration of the mold growth estimation device according to Embodiment 1. Flowchart showing an example of the procedure for the learning process (learning method) executed by the mold growth estimation device of Embodiment 1. Flowchart showing an example of the procedure for the mold growth estimation process of Embodiment 1. Figure showing an example of the input reception screen for measurement data of Embodiment 1. Figure showing an example of the display screen for mold growth risk information of Embodiment 1. Figure showing an example of the display screen for the surface temperature distribution of Embodiment 1. Figure showing an example of the display screen for the moisture content distribution of Embodiment 1. Figure showing an example of the display screen for the temperature distribution of Embodiment 1. Figure showing an example of the display screen for the humidity distribution of Embodiment 1. Figure showing another example of the display screen for the mold growth risk distribution of Embodiment 1. Figure showing an example of the display screen showing two distributions of Embodiment 1. Edited layout diagram of Embodiment 1. Figure 1 shows an example of a surface. Figure 2 shows an example of the configuration of a computer system that realizes the mold growth estimation device according to Embodiment 1. Figure 3 shows an example of the configuration of the mold growth estimation device according to Embodiment 3. Figure 4 shows an example of the configuration of the mold growth estimation device according to Embodiment 4. Figure 4 shows an example of the countermeasure support processing procedure of Embodiment 4. Figure 5 shows an example of the configuration of the mold suppression support system according to Embodiment 5. Figure 6 shows an example of the configuration of the mold growth estimation device according to Embodiment 6. Figure 7 shows an example of the configuration of the mold growth estimation device according to Embodiment 6.

[0009] The mold growth estimation device, mold suppression support system, mold growth estimation method, and program according to the embodiment will be described in detail below with reference to the drawings.

[0010] Embodiment 1. Figure 1 shows an example of the configuration of the mold growth estimation device 1 according to Embodiment 1. The mold growth estimation device 1 of this embodiment comprises an acquisition unit 11, an information storage unit 12, an estimation unit 13, and an output unit 14. The mold growth estimation device 1 of this embodiment estimates the mold growth risk (hereinafter also referred to as mold growth risk) of a target area. The target area is, for example, a store such as a supermarket, but it may also be a factory, a conference room, a hospital, a public facility, an apartment building, a school, a room in a private house, etc., and can be any space. Furthermore, the target area may be the entire building, room, etc., or a part of it.

[0011] The acquisition unit 11 acquires input data for an estimation model, described later, for estimating the risk of mold growth. As will be described later, the input data itself may be input to the estimation model, or the input data may be input to the estimation model after preprocessing, as will be described later. The input data (or preprocessed input data, if preprocessing is performed) can also be described as explanatory variables or features in machine learning. The acquisition unit 11 acquires input data by, for example, receiving input from a user such as an operator. Alternatively, the acquisition unit 11 may acquire input data by receiving input data from another device such as a terminal device 2. The input data includes at least the surface temperature of items at multiple locations within the target area for mold growth estimation. Hereinafter, the surface temperature of an item will also be simply referred to as surface temperature. The surface temperature is, for example, measurement data, as will be described later, but is not limited to this. Items include, for example, ceilings. Items may also include walls, electrical equipment, furniture, goods, interiors, etc. Items are not limited to these, and may include any item present in the target area. Details of the input data will be described later.

[0012] The information storage unit 12 stores data used in the processing of the mold growth estimation device 1, processing results, etc. The information storage unit 12 may also store input data acquired by the acquisition unit 11, for example.

[0013] The estimation unit 13 performs an estimation process using machine learning to estimate the risk of mold growth at multiple locations from the input data acquired by the acquisition unit 11. This process generates mold growth risk information from the input data, where the estimated risk of mold growth at multiple locations is associated with their respective locations. The mold growth risk information may, for example, represent the distribution of mold growth risk, but is not limited to this; it may also represent the mold growth risk for each location within the multiple locations. The following describes an example where the mold growth risk information represents the distribution of mold growth risk, but is not limited to this. The estimation unit 13 outputs the generated mold growth risk information to the output unit 14. The estimation unit 13 generates the mold growth risk information using input data corresponding to multiple locations within the target area.

[0014] The estimation unit 13 includes, for example, a learning unit 131, a model storage unit 132, an inference unit 133, and a distribution generation unit 134. The learning unit 131 uses the learning data, which is the input data for learning, to generate an estimation model, which is a trained model, by machine learning, and stores the generated estimation model in the model storage unit 132. The estimation model is a model for estimating the risk of mold growth from the input data. Furthermore, the estimation model in this embodiment is a model for predicting the risk of mold growth a first period after the date and time corresponding to the input data, i.e., the point in time corresponding to the input data. This embodiment assumes, for example, the estimation of the risk of mold growth in the short term. The first period is, for example, about two weeks, but is not limited to this. The learning unit 131 generates the estimation model by, for example, supervised learning or semi-supervised learning, but the machine learning algorithm is not limited to supervised learning.

[0015] When the learning unit 131 generates an estimation model by supervised learning, it uses multiple training datasets containing input data and corresponding ground truth data as training data to generate the estimation model by supervised learning. The ground truth data is the risk of mold growth a first period after the date and time corresponding to the input data. For example, it may be determined by a person evaluating the actual situation of mold growth, or it may be determined using a trained model generated by machine learning that determines the risk of mold growth from images showing the situation of mold growth. Alternatively, the ground truth data may be determined by extracting features from images using image analysis techniques and using the extracted features and defined rules. The method for determining the ground truth data is not limited to these examples. The training data may be acquired by the acquisition unit 11, or by a transmission / reception unit or input reception unit other than the acquisition unit 11. Furthermore, the training dataset may include not only the facility to be estimated, but also input data and corresponding ground truth data accumulated in other countries, other regions, or other facilities. Alternatively, a trained model may be generated using only a training dataset containing input data and corresponding ground truth data accumulated in other countries, other regions, or other facilities.

[0016] Examples of supervised learning algorithms that can be applied include neural networks, deep learning, genetic programming, functional logic programming, and support vector machines, but specific supervised learning algorithms are not limited to these.

[0017] The risk of mold growth can be represented, for example, by a mold index, but the specific indicators used to show the risk of mold growth are not limited to the mold index. Furthermore, the risk of mold growth may be information on whether or not mold will grow on the surface of an item after a first period, or it may be information indicating the probability of mold growing on the surface of an item after a first period. In addition, by learning the timing of mold growth using a regression model, the risk of mold growth may include information on when mold is likely to grow. In the example shown in Figure 1, the estimation model is a model for estimating the risk of mold growth at one location, and the input data for the estimation model is the input data corresponding to that location. However, as input data for estimating the risk of mold growth at one location, not only data corresponding to the location itself but also data from the surrounding area may be used.

[0018] The inference unit 133 estimates the risk of mold growth by performing inference on the risk of mold growth using the input data acquired by the acquisition unit 11 and the estimated model stored in the model storage unit 132. When data corresponding to the current time or a time around the current time is used as input data, the estimation of the risk of mold growth becomes an estimation of the risk of future mold growth, i.e., a prediction of the risk of mold growth. However, the inference unit 133 may also use input data corresponding to past dates and times to estimate the risk of mold growth a certain period after a past date and time. When semi-supervised learning is used, the configuration, learning, and inference procedures of the estimation unit 13 are the same as when supervised learning is used. Examples of semi-supervised learning algorithms that can be applied include Graph Convolutional Networks, label diffusion, label propagation, Self-Training, Mean Teacher, and variational autoencoders, but the specific algorithms for semi-supervised learning are not limited to these.

[0019] The distribution generation unit 134 generates mold growth risk information showing the distribution of mold growth risk using the mold growth risk estimated by the inference unit 133, and outputs the generated mold growth risk information to the output unit 14. Alternatively, the distribution generation unit 134 may store the generated mold growth risk information in the information storage unit 12.

[0020] Mold growth risk information is information that associates the estimated risk of mold growth at multiple locations with their respective locations. As described above, when using an estimation model that estimates the risk of mold growth at one location, the inference unit 133 will estimate the risk of mold growth for each input data at each location. For example, the distribution generation unit 134 may manage the input data and the corresponding location for each location where the risk of mold growth is calculated, and associate the mold growth risk inferred by the inference unit 133 with the location. Alternatively, the inference unit 133 may receive location information indicating the location of the input data along with the input data from the acquisition unit 11, and output the mold growth risk estimated using the input data along with the corresponding location information to the distribution generation unit 134. Alternatively, the input data acquired by the acquisition unit 11 may be associated with the location and temporarily stored in the information storage unit 12, and the inference unit 133 may sequentially read the input data from the information storage unit 12 to estimate the risk of mold growth, and store the estimation results in the information storage unit 12 in association with the input data. This allows the distribution generation unit 134 to associate the estimation results with locations by referring to the information storage unit 12. The method for associating the estimated mold growth risk with locations is not limited to these examples.

[0021] Furthermore, the distribution generation unit 134 may associate the estimated mold risk of areas where mold growth risk is estimated with the location itself as mold growth risk information, or it may generate information showing a contour map with contour lines as mold growth risk information by interpolating between the areas where estimation results are obtained. Any general method can be used to create the contour map. Similarly, the distribution generation unit 134 may also generate information showing a contour map from the input data and output the generated information to the output unit 14 or store it in the information storage unit 12.

[0022] Furthermore, the order of processing between distribution generation and mold growth risk estimation is not limited to the example described above. For example, input data from multiple locations stored in the information storage unit 12 may be input to the distribution generation unit 134, the distribution generation unit 124 may generate a distribution of input data using the input data from multiple locations, and the inference unit 133 may obtain an inference result of mold growth risk using the distribution of input data. This makes it possible to estimate the mold growth risk between measurement points. For example, suppose a trained model that determines that mold is likely to grow at temperatures between 35°C and 40°C has been generated through training, and the input data for a certain location A includes a temperature of 33°C, and the input data for the adjacent location B includes a temperature of 42°C. If the mold growth risk is estimated using the input data for both location A and location B, the estimated result will be that mold will not grow at location A and location B. On the other hand, by first obtaining the distribution of input data, interpolated data can be obtained between location A and location B, so an estimated result can be obtained that there is a mold growth risk between location A and location B.

[0023] The output unit 14 outputs mold growth risk information received from the distribution generation unit 134. The output unit 14 may also read mold growth risk information from the information storage unit 12 and output the read mold growth risk information. The output unit 14 may also output information showing a contour plot of the input data received from the distribution generation unit 134 or read from the information storage unit 12. The output from the output unit 14 may be a display or transmitted to another device. When the output unit 14 performs a display, it displays the distribution of mold growth risk by performing a display based on the mold growth risk information. The output unit 14 may also output mold growth risk information by transmitting the mold growth risk information to the terminal device 2 or a display device (not shown). In this case, the terminal device 2 or the display device (not shown) displays the distribution of mold growth risk by performing a display based on the mold growth risk information.

[0024] Terminal device 2 is a terminal device that can be operated by the user. When the output unit 14 transmits mold growth risk information to terminal device 2, as described above, terminal device 2 displays the distribution of mold growth risk by displaying information based on the mold growth risk information. For example, the distribution shows which stage each location in the target area is at by dividing the mold growth risk value into multiple stages. Terminal device 2 may also accept input data from the user and transmit the received input data to the mold growth estimation device 1. Note that although terminal device 2 is shown in Figure 1, terminal device 2 does not need to be provided if the output unit 14 or a display device (not shown) displays the distribution of mold growth risk.

[0025] In the example described above, the estimation model estimated the risk of mold growth at one location, but it is not limited to this, and an estimation model that estimates the risk of mold growth at multiple locations may be used. In this case, for example, the input data includes not only data such as surface temperatures at multiple locations, but also location information indicating the location corresponding to that data. The learning unit 131 uses, for example, the data such as surface temperatures at multiple locations and the location information as features, and uses the risk of mold growth at multiple locations as ground truth data to generate an estimation model that estimates the risk of mold growth at multiple locations, for example, by supervised learning or semi-supervised learning. The inference unit 133 obtains the inference result of the risk of mold growth at multiple locations, i.e., the estimation result of the risk of mold growth at multiple locations, by inputting the features into the estimation model. In this case, the mold growth estimation device 1 may not have a distribution generation unit 134, and the inference unit 133 may generate mold growth risk information by associating the estimation result of the risk of mold growth at each location with the location information. Alternatively, the mold growth estimation device 1 may include a distribution generation unit 134, and the distribution generation unit 134 may generate information showing a contour map with contour lines as mold growth risk information by performing interpolation processing using the estimated mold growth risk results and location information.

[0026] Furthermore, although Figure 1 illustrates an example in which the mold growth estimation device 1 includes a learning unit 131, a learning device having the functions of the learning unit 131 may be provided separately from the mold growth estimation device 1. In this case, the estimation unit 13 does not need to include the learning unit 131, and the learning device generates the estimation model, which is then stored in the model storage unit 132. The estimation model may be updated after it has been generated using newly acquired input data and corresponding ground truth data.

[0027] Next, the operation of this embodiment will be described. Figure 2 is a flowchart showing an example of the steps of the learning process (learning method) performed by the mold growth estimation device 1 of this embodiment. As shown in Figure 2, the mold growth estimation device 1 acquires learning data (step S1). In detail, as described above, the learning unit 131 acquires multiple learning datasets, for example, which include input data and correct answer data, as learning data.

[0028] The mold growth estimation device 1 generates an estimation model (step S2), stores the generated estimation model (step S3), and terminates the learning process. Specifically, the learning unit 131 generates an estimation model using machine learning with training data and stores the generated estimation model in the model storage unit 132. As described above, the learning process shown in Figure 2 may be performed by a learning device other than the mold growth estimation device 1.

[0029] Next, the mold occurrence estimation process (mold occurrence estimation method) performed by the mold occurrence estimation device 1 of this embodiment will be described. Figure 3 is a flowchart showing an example of the procedure for the mold occurrence estimation process of this embodiment. The mold occurrence estimation device 1 acquires input data (step S11). In detail, for example, the acquisition unit 11 receives input data from a user such as an operator.

[0030] The input data includes, for example, measurement data related to the target area. The measurement data related to the target area may be measurement data within the target area, or it may include both measurement data within the target area and measurement data related to the target area outside the target area. As described above, the input data includes at least the surface temperature of multiple locations of an object in the target area, where the surface temperature of multiple locations of an object is, for example, measurement data of the surface temperature of multiple locations of an object. The surface temperature of multiple locations may be the surface temperature of different locations on a single object, or it may be the surface temperature of one or more locations on multiple objects. For example, the surface temperature of multiple locations may be the surface temperature of multiple different locations on a ceiling. The input data may also include the moisture content of the object, where this moisture content is, for example, measurement data. The moisture content is, for example, the water content.

[0031] The measurement data may include, for example, surface temperature of an object, moisture content of an object, and temperature (indoor temperature, room temperature) in the target area. The measurement data may also include data measured by sensors embedded in equipment such as air conditioning systems installed in the target area. Furthermore, the measurement data may include humidity at measurement points within the target area, wind speed at measurement points within the target area, and outside temperature. Outside temperature is an example of measurement data related to the target area outside the target area, representing the temperature around the target area. Note that outside temperature is not limited to examples obtained as measurement data, as described later, but may also be data provided as meteorological data. Similarly, humidity and wind speed may also be data provided as meteorological data. The measurement data may also include at least one of the temperature and humidity in the attic. Meteorological data may be obtained, for example, from an external meteorological data provision system.

[0032] Furthermore, instead of measurement data, the input data may include data predicted based on past measurement data, data estimated from measurement data from another location, etc.

[0033] Measurement data is acquired, for example, by a worker and a robot moving within the target area while using sensors, which are measuring devices. For example, when measuring the surface temperature and moisture content of a ceiling, a worker supports a pole to which a temperature sensor, such as a radiation thermometer for measuring surface temperature, and a moisture meter for measuring moisture content are attached. The temperature sensor and moisture meter are then brought closer to the ceiling to measure the surface temperature and moisture content of the ceiling. Surface temperature may also be measured using a thermal camera. When using a thermal camera, the surface temperature within the thermal camera's field of view can be measured in a single measurement. However, even in this case, the area that can be measured in a single measurement is limited to the thermal camera's field of view. Therefore, even when using a thermal camera, measurements are taken at multiple locations, i.e., by changing the area within the thermal camera's field of view. By having at least one of the worker and a robot move within the target area while taking measurements, measurement data from multiple locations can be acquired without having to place sensors at each of the multiple locations. Note that the sensors used for measurement and the method of acquiring measurement data are not limited to the examples described above. Also, the locations to be measured (hereinafter referred to as measurement points) can be changed as appropriate. If there are not many measurement points, measurement data may be acquired using a fixed sensor.

[0034] The measurement data may be input to the mold occurrence estimation device 1 by an operator, or, if the sensor has a function to output the measurement data as electronic data, the acquisition unit 11 may function as a sensor interface capable of acquiring electronic data from the sensor and acquire the electronic data from the sensor. In this case as well, if the operator is moving while taking measurements, the operator will associate each electronic data with the position of the measurement point. For example, for each measurement point, the operator may input the location to be measured to the mold occurrence estimation device 1, and the electronic data will be acquired by the mold occurrence estimation device 1, thereby associating the position of the measurement point with the measurement data. Alternatively, the measurement data, which is electronic data, may include the date and time of measurement, and the operator may record the date and time of measurement or the order of measurement, and after the measurement is performed, the operator may associate the measurement data with the position of the measurement point based on the recorded information. Furthermore, if a robot is moving while taking measurements, the association between the position of the measurement point and the measurement data may be performed in the same way as when an operator takes measurements, or the robot's position information may be transmitted to the mold occurrence estimation device 1 along with the corresponding time. As a result, the acquisition unit 11 or the estimation unit 13 may perform a correspondence between the measurement data and the location of the measurement point. The method by which the measurement data is input to the mold growth estimation device 1 is not limited to the examples described above.

[0035] Figure 4 shows an example of the input reception screen for measurement data in this embodiment. The input reception screen shown in Figure 4 is displayed by the output unit 14, for example, when the output unit 14 is performing the display. Alternatively, information for displaying the input reception screen shown in Figure 4 may be transmitted from the mold occurrence estimation device 1 to the terminal device 2, and the terminal device 2 may display the input reception screen.

[0036] The display screen shown in Figure 4 has instructions for inputting measurement data in the upper left corner. In this example, first, a layout diagram (labeled "Diagram" in Figure 4) containing the target area is selected, and then the boundary vertices are input (selected). The layout diagram is a diagram that shows the layout of a building, room, etc., containing the target area, and the arrangement of equipment, furnishings, etc., to be installed. Here, the layout diagram is pre-stored in the information storage unit 12, and an example is shown in which a layout diagram is selected from among multiple layout diagrams stored in the information storage unit 12. However, this is not limited to this, and a new layout diagram may be input. The boundary vertices refer to the vertices that form the boundary of the target area. In the example shown in Figure 4, four selected vertices 202 are selected, and the rectangular area enclosed by the lines connecting the selected vertices 202 is selected as the target area. In Figure 4, the target area is shown on a planar layout diagram, so the target area specified by the selected vertices 202 is shown in two dimensions, but in reality, the target area is a three-dimensional space that also includes the height direction. Furthermore, the number of vertices that can be selected is not limited to four; it is possible to specify any shape as the target area using multiple vertices.

[0037] The display screen shown in Figure 4 serves as both an input screen for displaying mold growth risk information and a display screen for showing mold growth risk and measurement data as a distribution. In the example shown in Figure 4, it is possible to display the distributions of mold growth risk, surface temperature, and moisture content superimposed on the layout diagram, and the user can select which distributions to display or whether to display any distributions at all.

[0038] When inputting measurement data, for example, in the lower left of Figure 4, a selection diagram 201, which is a selected layout diagram, and an area indicating the target area specified by the selected vertices 202 are shown. On the selection diagram 201, the corresponding position on the selection diagram 201 is input for each measurement point number. In Figure 4, the state after the input of the positions for each measurement point number is performed, and a circular mark 203 is displayed on the selection diagram 201 at the position corresponding to the measurement point, along with the measurement point number. In the figure, a code is attached to the corresponding mark 203 for each measurement point, but similar circular marks without codes also indicate measurement points. The data input area 204 on the right side of Figure 4 is an area for receiving measurement data input, and the rectangles displayed next to the surface temperature, moisture content, and measurement time (measurement date and time) within the data input area 204 are input fields for receiving input. This measurement data and measurement date and time may be entered by an operator as described above, or they may be acquired from a sensor. Note that Figure 4 is an example, and the method of receiving measurement data input is not limited to the example shown in Figure 4. In this way, by specifying the position of the measurement point in the selection diagram 201, or layout diagram, the user can easily specify the position of the measurement point corresponding to the measurement data. Furthermore, it becomes easier to understand the position of the measurement point corresponding to the measurement data on the layout.

[0039] Returning to the explanation of Figure 3, the input data acquired in step S11 may include the meteorological data described above, in addition to the measurement data. The input data may include at least one of the following: whether or not there is insulation around the item (e.g., whether or not there is insulation in the ceiling), whether or not mold prevention treatment such as mold-resistant coating has been applied, information about the surrounding equipment, meteorological data at that time, measurement date and time, type of mold-resistant coating agent (company, product name), type of ceiling material, and wall material. Information about the surrounding equipment may include, for example, information indicating the location of equipment such as entrances, vents, air conditioners, and refrigerated display cases, as well as the operating status and settings of equipment such as entrances, vents, air conditioners, and refrigerated display cases. Information about the surrounding equipment may include not only information at the time of measurement, but also schedules for defined periods such as annual schedules. The meteorological data may include at least one of the following: temperature (outside temperature), humidity, atmospheric pressure, precipitation, wind speed, and wind direction. As described above, the input data may include, for example, at least one of the following: temperature within the target area, ambient temperature around the target area, presence or absence of insulation around the item, presence or absence of mold prevention treatment, humidity within the target area, operating status of equipment in the target area, and location of equipment. The same type of input data acquired in step S11 will be used as input data in the training data during training.

[0040] After step S11, the mold growth estimation device 1 estimates the risk of mold growth using the input data and the estimation model (step S12). Specifically, the inference unit 133 obtains the inference result obtained by inputting the input data into the estimation model stored in the model storage unit 132 as the estimated risk of mold growth.

[0041] Furthermore, the input data may be preprocessed as described above. Preprocessing may include, for example, calculating the difference between the average temperature at all measurement points in the room and the temperature at the measurement point in question, or calculating the difference between the average surface temperature at all measurement points in the room and the surface temperature at the measurement point in question. Preprocessing may also include operations such as arithmetic operations, exponential operations, logarithmic operations, and normalization between explanatory variables, or a combination of two or more of these. If preprocessing is performed, the learning unit 131 preprocesses the input data and uses the preprocessed input data as training data, while the inference unit 133 preprocesses the input data and inputs the preprocessed input data into the estimation model.

[0042] The mold growth estimation device 1 outputs mold growth risk information that shows the estimated mold growth risk as a distribution (step S13). Specifically, the distribution generation unit 134 uses the mold growth risk of each location estimated by the inference unit 133 to generate mold growth risk information that shows a distribution of mold growth risk associated with location, and outputs the generated mold growth risk information to the output unit 14. The output unit 14 outputs the mold growth risk information. The output unit 14 may also output the layout diagram and the mold growth risk information linked together. Specifically, the output unit 14 may output the layout diagram and the mold growth risk information linked together by, for example, displaying the mold growth risk information superimposed on the layout diagram. Alternatively, the output unit 14 may transmit the layout diagram and the mold growth risk information linked together to the terminal device 2 or display device, so that the terminal device 2 or display device displays the mold growth risk information superimposed on the layout diagram.

[0043] FIG. 5 is a diagram showing an example of a display screen for mold growth risk information according to the present embodiment. In the example shown in FIG. 5, the input reception screen shown in FIG. 4 also serves as the display screen for mold growth risk information. When measurement data is input and there is input data other than the measurement data, after the input data is entered, the radio button near the center on the left side of FIG. 5 selects mold growth risk as a display target, whereby the distribution of mold growth risk is displayed superimposed on a selection diagram 201 (layout diagram) in the lower left. This allows the mold growth risk to be visually presented to the user in association with indoor positions. Therefore, the user can visually and easily grasp regions where mold is estimated to be likely to grow. As described above, the mold growth risk information can be used as information that allows the user to grasp regions where mold is estimated to be likely to grow.

[0044] Further, in the example shown in FIG. 5, an example in which the distribution of mold growth risk is displayed in the lower left is described, but the present invention is not limited thereto. According to a user's selection, a surface temperature distribution or a moisture content distribution is displayed in the lower left. This allows the user to visually and easily grasp measurement data such as surface temperature and moisture content as a distribution associated with positions.

[0045] Note that FIG. 5 is an example, and the distribution of mold growth risk and the distribution of measurement data may be displayed independently. For example, which distribution is to be displayed on the screen may be switchable, and two or more distributions may be displayed simultaneously. FIGS. 6 to 9 are diagrams each showing an example of a display screen for a distribution of measurement data according to the present embodiment. FIG. 10 is a diagram showing another example of a display screen for a distribution of mold growth risk according to the present embodiment. FIG. 11 is a diagram showing an example of a display screen on which two distributions are displayed according to the present embodiment. As shown in FIGS. 6 to 11, the measurement data and the distribution of mold growth risk according to the present embodiment may be displayed as contour diagrams.

[0046] Figure 6 shows an example of a display screen for surface temperature distribution, Figure 7 shows an example of a display screen for moisture content distribution, Figure 8 shows an example of a display screen for air temperature distribution, and Figure 9 shows an example of a display screen for humidity distribution. Figure 6 shows the surface temperature distribution of a ceiling, which is an example of an object. In the display screens illustrated in Figures 6 to 11, a screen switching button 205 is displayed. For example, selecting the screen switching button 205 displays a menu screen 206, and by selecting the distribution to switch to on the menu screen 206, the displayed distribution can be switched. Note that the method of switching screens is not limited to this example. As shown in Figures 4 and 5, it may be possible to select which distribution to display using radio buttons, or it may be possible to select by other methods. As illustrated in Figures 6 to 11, by displaying the distribution superimposed on the layout diagram, the user can easily grasp the situation at each location in the target area. The display information for displaying the display screens illustrated in Figures 4 to 11 may be generated by the acquisition unit 11, the distribution generation unit 134, or a display control unit (not shown). The generated display information is displayed by the output unit 14, the terminal device 2, or a display device (not shown).

[0047] In the example shown in Figure 11, the distribution of surface temperature and the distribution of mold growth risk are displayed simultaneously, one above the other. By displaying the measurement data and mold growth risk side by side on a single screen in this way, users can easily understand the relationship between the measurement data and the mold growth risk. Note that Figure 11 is an example, and two distributions may be displayed side by side, or three or more distributions may be displayed on a single screen. The display screens shown above are examples, and the methods for displaying each distribution and switching between display screens are not limited to the examples described above.

[0048] Although FIGS. 4 to 11 show an example in which a plan view is used as a layout drawing, the present invention is not limited thereto, and an image obtained by capturing the target area may be used. In addition, the layout drawing is not limited to planar drawings and planar images, and may be three-dimensional drawings, three-dimensional point cloud data, three-dimensional images, or the like. When the layout drawing is shown three-dimensionally, that is, in three dimensions, measurement points may be specified by three-dimensional positions, and various distribution diagrams such as mold growth risk may also be displayed as three-dimensional distribution diagrams. In FIGS. 4 to 11, the layout drawing shows the layout of a portion below the ceiling, but the layout drawing may also show the arrangement of lighting fixtures, ventilation equipment, fire alarms, decorations, and the like on the ceiling. In addition, a user may be allowed to additionally add the arrangement of articles such as lighting fixtures, ventilation equipment, fire alarms and decorations on the ceiling to the layout drawing of the portion below the ceiling afterward.

[0049] FIG. 12 is a diagram showing an example of an editing screen for a layout drawing according to the present embodiment. In the example shown in FIG. 12, the current layout drawing is displayed on the left side, and graphics of articles to be arranged on the ceiling are shown on the right side. A user can add an article to the layout drawing by selecting the graphic corresponding to the article the user wants to add and arranging it on the layout drawing. An input field for a user to draw an arbitrary graphic is provided at the lower part of the right side of FIG. 12, and the user may input a graphic in this portion. In addition, an article (a graphic of an article) that has been arranged once may be allowed to be deleted from the layout drawing, or the position thereof on the layout drawing may be changed.

[0050] It should be noted that by changing the display mode between the portion below the ceiling and the articles arranged on the ceiling, it may be possible to distinguish whether an article is arranged on the ceiling or not. For example, at least one of the color and the line type may be made different between the portion below the ceiling and the articles arranged on the ceiling. The display mode is not limited to this example. Although FIG. 12 shows an example of adding articles on the ceiling to a plan view, as described above, articles on the ceiling may be similarly added to two-dimensional or three-dimensional images, three-dimensional drawings, three-dimensional point cloud data, and the like. Graphics of articles represented in three dimensions may be allowed to be added to a layout drawing shown in three dimensions.

[0051] Furthermore, while Figure 12 shows an example of adding items from the ceiling to a layout diagram of the area below the ceiling, items located below the ceiling may also be placed on the layout diagram by the user by selecting or inputting shapes. Figure 12 is just one example, and the method for accepting changes to the layout diagram, i.e., the editing method, is not limited to the example shown in Figure 12. As shown in Figure 12, by accepting additions, deletions, and movements of items on the layout diagram while it is displayed, the user can easily edit the layout diagram.

[0052] Furthermore, if the layout diagram is shown in two dimensions, the user may be able to switch between displaying a layout diagram below the ceiling and a layout diagram of the ceiling. In this case, the system may accept input of the measurement data position as a three-dimensional position, making it possible to represent various distributions such as mold growth risk in three dimensions, and allowing the user to specify the height corresponding to the distribution to be superimposed on the layout diagram. For example, by superimposing various distributions such as mold growth risk at ceiling height onto the ceiling layout diagram, the user can more easily grasp the level of mold growth risk in areas prone to mold growth, such as the ceiling.

[0053] Returning to the explanation of Figure 3, after step S13, the mold growth estimation device 1 decides whether or not to terminate the mold growth estimation process (step S14). For example, the mold growth estimation device 1 decides to terminate the mold growth estimation process if the user instructs it to terminate. If the mold growth estimation process is to be terminated (step S14 Yes), the mold growth estimation device 1 terminates the mold growth estimation process. If the mold growth estimation process is not to be terminated (step S14 No), the process from step S11 is repeated.

[0054] Next, the hardware configuration of the mold growth estimation device 1 of this embodiment will be described. In the mold growth estimation device 1 of this embodiment shown in Figure 1, the computer system functions as the mold growth estimation device 1 when a program (computer program) describing the processing in the mold growth estimation device 1 is executed on the computer system. Figure 13 is a diagram showing an example of the configuration of a computer system that realizes the mold growth estimation device 1 according to this embodiment. As shown in Figure 13, this computer system comprises a control unit 101, an input unit 102, a storage unit 103, a display unit 104, a communication unit 105, and an output unit 106, which are connected via a system bus 107. The control unit 101 and the storage unit 103 constitute a processing circuit.

[0055] In Figure 13, the control unit 101 is a processor such as a CPU (Central Processing Unit), which executes a program describing the processing in the mold growth estimation device 1 of this embodiment. Note that a part of the control unit 101 may be implemented by dedicated hardware such as a GPU (Graphics Processing Unit) or FPGA (Field-Programmable Gate Array). The input unit 102 is a button, keyboard, mouse, touchpad, etc. The storage unit 103 includes various types of memory such as RAM (Random Access Memory) and ROM (Read Only Memory), as well as storage devices such as a hard disk, and stores the program to be executed by the control unit 101, necessary data obtained during the processing, etc. The storage unit 103 is also used as a temporary storage area for the program. The display unit 104 is a display, such as an LCD (Liquid Crystal Display). Note that the display unit 104 and the input unit 102 may be integrated and implemented as a touch panel or the like. The communication unit 105 is a receiver and transmitter that perform communication processing. The output unit 106 is a speaker or the like. Note that Figure 13 is just one example, and the configuration of the computer system is not limited to the example in Figure 13. For example, the computer system that implements the mold growth estimation device 1 does not need to have an output unit 106.

[0056] Here, an example of the operation of the computer system until the program of this embodiment becomes executable will be described. In a computer system with the above configuration, for example, a computer program is installed in the storage unit 103 from a CD-ROM or DVD-ROM set in a CD (Compact Disc)-ROM drive or DVD (Digital Versatile Disc)-ROM drive (not shown). When the program is executed, the program read from the storage unit 103 is stored in the main memory area of ​​the storage unit 103. In this state, the control unit 101 performs processing as the mold growth estimation device 1 of this embodiment according to the program stored in the storage unit 103.

[0057] In the above explanation, a program describing the processing in the mold growth estimation device 1 is provided using a CD-ROM or DVD-ROM as the recording medium. However, the explanation is not limited to this, and depending on the configuration of the computer system, the capacity of the program to be provided, a program provided via a transmission medium such as the Internet may be used.

[0058] The program of this embodiment, for example, causes a computer system to perform the following steps: acquire input data including at least the surface temperature of multiple items in a target area where mold growth is to be estimated; generate mold growth risk information from the input data by performing an estimation process using machine learning to estimate the mold growth risk of multiple locations from the input data, thereby associating the estimated mold growth risk of multiple locations with their locations; and output the mold growth risk information.

[0059] The learning unit 131, inference unit 133, and distribution generation unit 134 shown in Figure 1 are realized by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to realize the learning unit 131, inference unit 133, and distribution generation unit 134. The acquisition unit 11 shown in Figure 1 is realized by at least one of the input unit 102 and communication unit 105 shown in Figure 13. Furthermore, when the acquisition unit 11 generates display information, the control unit 101 and storage unit 103 are also used to realize the acquisition unit 11. The information storage unit 12 and model storage unit 132 shown in Figure 1 are part of the storage unit 103 shown in Figure 13. Note that the mold occurrence estimation device 1 may be realized by multiple computer systems. Also, for example, the mold occurrence estimation device 1 may be realized by a cloud computer system. Also, the terminal device 2 shown in Figure 1 can be realized by, for example, the computer system illustrated in Figure 13.

[0060] As described above, in this embodiment, the mold growth estimation device 1 performs estimation processing using machine learning with input data that includes at least the surface temperature of multiple locations on an item in the target area. This generates mold growth risk information that shows the distribution of mold growth risk, with the estimated risk of mold growth at multiple locations associated with their locations. As a result, the mold growth estimation device 1 can output the mold growth risk information as information that allows the user to understand areas where mold is estimated to be likely to grow. By displaying the mold growth risk information, the user can easily identify areas where mold is likely to grow.

[0061] Embodiment 2. Figure 14 shows an example of the configuration of the mold growth estimation device 1a according to Embodiment 2. The mold growth estimation device 1a of this embodiment is the same as the mold growth estimation device 1 of Embodiment 1, except that it includes an estimation unit 13a instead of an estimation unit 13. Components having the same functions as in Embodiment 1 are denoted by the same reference numerals as in Embodiment 1, and redundant explanations are omitted. The differences from Embodiment 1 will be mainly described below.

[0062] While Embodiment 1 assumed the estimation of short-term mold growth risk (hereinafter also referred to as short-term estimation), the mold growth estimation device 1a of this embodiment can perform not only short-term mold growth risk estimation but also long-term (long-term) mold growth risk estimation (hereinafter also referred to as long-term estimation). Long-term refers to, for example, a second period after the date and time corresponding to the input data, where the second period is, for example, two or three years, but is not limited to this. The second period just needs to be longer than the first period.

[0063] As shown in Figure 14, the estimation unit 13a comprises a learning unit 131a, a model storage unit 132a, an inference unit 133a, and a distribution generation unit 134. The learning unit 131a has the same functions as the learning unit 131 described in Embodiment 1, and, similar to Embodiment 1, uses learning data, i.e., learning data for estimating the short-term mold growth risk (hereinafter also referred to as the first learning data), to generate a first estimation model, which is an estimation model for estimating the short-term mold growth risk. Hereinafter, the input data for the first estimation model will also be referred to as the first input data. The learning unit 131a stores the generated first estimation model in the model storage unit 132a.

[0064] The learning unit 131a further generates a second estimation model, which is an estimation model for estimating the long-term risk of mold growth, using learning data for estimating the long-term risk of mold growth (hereinafter also referred to as the second learning data) through machine learning, and stores the generated second estimation model in the model storage unit 132a. Hereinafter, the input data for the second estimation model will also be referred to as the second input data. As for the machine learning used by the learning unit 131a to generate the second estimation model, supervised learning or semi-supervised learning may be used, as in Embodiment 1, or other methods may be used. As for the supervised learning algorithm, as in Embodiment 1, for example, neural networks, deep learning, genetic programming, functional logic programming, support vector machines, etc. can be applied, but the specific algorithm for supervised learning is not limited to these. As for the semi-supervised learning algorithm, as in Embodiment 1, for example, graph convolutional networks, label diffusion, label propagation, self-training, mean teacher, variational autoencoder, etc. can be applied, but the specific algorithm for semi-supervised learning is not limited to these. The machine learning algorithm used to generate the first estimation model and the machine learning algorithm used to generate the second estimation model may be the same or different.

[0065] The input data (explanatory variables, features) in the second training data is different from the first input data. The second input data includes, for example, the surface temperature of objects at multiple locations in the target area and the ambient temperature around the target area at the date and time corresponding to the second input data, and further includes the ambient temperature around the target area for a certain period prior to the date and time corresponding to the second input data. Specifically, for example, the second input data includes measurement data of the surface temperature of objects in the target area, measurement data of the ambient temperature in the target area, measurement data or meteorological data of ambient temperature at locations surrounding the target area, and meteorological data including ambient temperature for a certain period prior to the date and time corresponding to the measurement data at locations surrounding the target area. The input data in the second training data may also include information indicating the moisture content of objects such as ceilings and the current mold growth status.

[0066] Historical weather data must include data from at least the same season, month, or day as the forecast period. The forecast period is the period for which mold growth risk is estimated. Historical weather data may also be time-series data covering more than one year in the past. Using time-series data allows for learning of environmental changes such as temperature or humidity, which occur day or night or seasonally, enabling highly accurate estimation of mold growth. For example, the time-series data may consist of four or more data points per day with intervals of less than six hours.

[0067] Furthermore, using weather data from the past year or more allows for highly accurate estimation that reflects seasonal changes. Using weather data from the past two years or more enables a more average estimation of mold growth, unaffected by trends in the most recent year. Using weather data over an even longer period allows for mold growth estimation that also considers recent climate change. Also, similar to Embodiment 1, the mold growth risk may be information indicating whether or not mold will grow on the surface of the item after the second period, or information indicating the probability of mold growth on the surface of the item after the second period. Additionally, by learning the timing of mold growth using a regression model, the mold growth risk information may include information indicating when mold is likely to grow.

[0068] Furthermore, the second estimation model may be generated for each length of the prediction period. For example, a second estimation model that performs estimation (prediction) up to six months ahead and a second estimation model that performs estimation (prediction) up to two years ahead may be generated. Also, as in Embodiment 1, preprocessing may be applied to the second input data.

[0069] The inference unit 133a has the same functions as in Embodiment 1, and estimates the short-term mold growth risk in the same way as in Embodiment 1, using the first input data acquired by the acquisition unit 11 and the first estimation model stored in the model storage unit 132a. The inference unit 133a further estimates the long-term mold growth risk using the second input data acquired by the acquisition unit 11 and the second estimation model stored in the model storage unit 132a.

[0070] The acquisition unit 11 can acquire first input data similar to that in Embodiment 1, and can also acquire second input data. The acquisition unit 11 can also accept a selection of whether to perform short-term estimation or long-term estimation, or both short-term and long-term estimation, and outputs the selection result to the inference unit 133a. The inference unit 133a performs short-term estimation, long-term estimation, or both short-term and long-term estimation according to the received selection result. The method by which the acquisition unit 11 acquires the second input data is the same as the method of acquiring input data described in Embodiment 1. If the second input data includes measurement data, the acquisition unit may accept the position of the measurement points on the layout diagram and accept the input of measurement data for each measurement point, as illustrated in Figure 4 of Embodiment 1.

[0071] The distribution generation unit 134 generates information indicating the distribution of measurement data of the first input data, and mold growth risk information based on short-term estimation results, as first mold growth risk information, similar to the first embodiment. The distribution generation unit 134 further generates information indicating the distribution of measurement data of the second input data, and second mold growth risk information indicating the distribution of estimation results, which is information where long-term estimation results are associated with location. The distribution generation method is the same as in the first embodiment.

[0072] The method for displaying the distribution corresponding to the long-term estimation results is the same as in Embodiment 1. Furthermore, the distribution of mold growth risk corresponding to the short-term estimation results and the distribution of mold growth risk corresponding to the long-term estimation results may be displayed on the same screen. Additionally, the mold growth estimation device 1a may estimate the mold growth risk for multiple different prediction periods and display the distribution of mold growth risk for each prediction period sequentially, for example, in order of shortest to longest prediction period, thereby displaying the mold growth risk distribution as a video.

[0073] The mold growth estimation device 1a of this embodiment is implemented, similar to the mold growth estimation device 1 of Embodiment 1, by, for example, the computer system illustrated in Figure 13. The learning unit 131a and inference unit 133a shown in Figure 14 are implemented by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the learning unit 131a and inference unit 133a. The model storage unit 132a shown in Figure 14 is a part of the storage unit 103 shown in Figure 13. The mold growth estimation device 1a may be implemented by multiple computer systems. Also, for example, the mold growth estimation device 1a may be implemented by a cloud computing system.

[0074] As described above, in this embodiment, the mold growth estimation device 1a performs the same processing as in Embodiment 1, and further generates and outputs second mold growth risk information that shows the distribution of the estimation results, with the long-term estimation results associated with location. This allows the user to grasp both short-term and long-term mold growth risks. For example, when formulating long-term equipment plans or budget plans, the user can refer to the distribution of long-term mold growth risks, and when considering the purchase of materials in the near future, they can refer to the distribution of short-term mold growth risks, thus grasping mold growth risks according to their purpose.

[0075] Embodiment 3. Figure 15 shows an example of the configuration of the mold growth estimation device 1b according to Embodiment 3. The mold growth estimation device 1b shown in Figure 15 is the same as the mold growth estimation device 1 of Embodiment 1, except that it includes an estimation unit 13b instead of an estimation unit 13. Components having the same functions as in Embodiment 1 are denoted by the same reference numerals as in Embodiment 1, and redundant explanations are omitted. The differences from Embodiment 1 will be mainly described below.

[0076] In this embodiment, in addition to the measurement data described in Embodiment 1, measurement data is acquired from one or more sensors 3 permanently installed in the target area. Although Figure 15 shows one sensor 3, there may be multiple sensors 3, and if multiple sensors 3 are used, the multiple sensors 3 may include sensors of different types. Sensor 3 is, for example, at least one of a temperature sensor that detects temperature and a humidity sensor that detects humidity, but is not limited to these. The acquisition unit 11 acquires measurement data from the sensor 3 and outputs the measurement data from the sensor 3 to the estimation unit 13b. Sensor 3 transmits measurement data to the mold growth estimation device 1b at a predetermined period, for example, but sensor 3 may transmit measurement data to the mold growth estimation device 1b when the mold growth estimation device 1b requests the sensor 3 to transmit measurement data. Alternatively, for example, sensor 3 may transmit measurement data to another device (not shown) at a predetermined interval, the other device may store the measurement data from sensor 3, and when mold growth estimation device 1b requests the other device to transmit the measurement data from sensor 3, the other device may transmit the measurement data from sensor 3 to mold growth estimation device 1b. The method by which mold growth estimation device 1b acquires the measurement data from sensor 3 is not limited to these examples. Hereinafter, the measurement data from sensor 3 will also be referred to as reference data.

[0077] The estimation unit 13b includes a learning unit 131, a model storage unit 132, an inference unit 133, and a distribution generation unit 134, similar to those in Embodiment 1, and further includes a correction unit 135. The correction unit 135 corrects at least a portion of the input data using measurement data acquired from the sensor 3. Specifically, the correction unit 135 receives reference data from the acquisition unit 11, corrects the input data used for input to the estimation model using the reference data, and passes the corrected input data to the inference unit 133. The inference unit 133 estimates the risk of mold growth using the corrected input data and the estimation model, similar to Embodiment 1. If preprocessing is performed, the inference unit 133 applies preprocessing to the corrected input data and inputs the preprocessed input data to the estimation model. Alternatively, the order of preprocessing and correction may be reversed, with the correction unit 135 correcting the preprocessed input data using the reference data, and the inference unit 133 inputting the corrected input data to the estimation model.

[0078] For example, if the reference data is temperature measurement data within the target area, the correction unit 135 uses the temperature from the reference data to correct the temperature-related measurement data among the input data used for input to the estimation model. The temperature-related measurement data is air temperature measurement data, but may also include other measurement data such as surface temperature and attic temperature. Also, if the reference data is humidity measurement data within the target area, the correction unit 135 uses the humidity from the reference data to correct the humidity-related measurement data among the input data used for input to the estimation model. The humidity-related measurement data is humidity, but is not limited to humidity. It is sufficient that at least a portion of all types of measurement data in the input data is corrected based on the reference data.

[0079] Figure 16 is a schematic diagram showing the correction in this embodiment. In Figure 16, the horizontal axis represents the date and time, and the vertical axis represents the temperature. The curve 301 shown in Figure 16 approximates the reference data with a curve. As described in Embodiment 1, the measurement data used as input data is acquired at multiple locations while a worker or robot moves, for example. Acquiring measurement data at multiple locations while a worker or robot moves is also called full-surface data measurement. In general, it is difficult to perform full-surface data measurement at all times, and the timing of measurements is limited to some extent. In contrast, since the reference data is measurement data from a permanently installed sensor 3, the measurement frequency can be increased compared to full-surface data measurement. That is, the acquisition cycle of the reference data is shorter than the acquisition cycle of the measurement data by full-surface data measurement.

[0080] In this embodiment, when the mold growth estimation device 1b estimates the risk of mold growth at a time when a certain amount of time has elapsed since the time of full-surface data measurement, it corrects the measurement data measured by the full-surface data measurement using reference data. As shown in Figure 16, if the time at which the risk of mold growth is estimated is the correction target time, the correction unit 135 corrects the temperatures 302 and 303, such as surface temperature and air temperature, obtained from the most recent full-surface data measurement at the correction target time, using reference data. Specifically, for example, it calculates the difference by subtracting the temperature value shown by the reference data at the time of full-surface data measurement from the temperature value shown by the reference data at the correction target time, and calculates the corrected temperatures 304 and 305 by adding this difference to temperatures 302 and 303, respectively. Note that the difference can be positive or negative, so adding the difference may result in a decrease in temperature.

[0081] Alternatively, instead of directly adding the difference value to the measurement data to be corrected, the difference may be multiplied by a predetermined weighting coefficient and then added to the measurement data to be corrected. The weighting coefficient may be determined according to the position and type of each measurement data. For example, the weighting coefficient may be determined in advance through tests or experiments. Furthermore, the method of correction based on reference data is not limited to the examples described above. For example, a correction formula may be determined by regression analysis using previously obtained data, or a trained model for correction may be generated using machine learning, and the correction may be performed using the trained model.

[0082] Furthermore, the method is not limited to using reference data at two points: the time of full-surface data measurement and the time of correction. Correction may also be performed using reference data from time-series data from the time of full-surface data measurement to the time of correction. Additionally, correction may be performed by mixing different types of data, for example, by acquiring temperature and humidity as reference data and representing the air temperature, surface temperature, and humidity at the time of full-surface data measurement as functions of both, the air temperature, surface temperature, and humidity may be corrected. Furthermore, if multiple sensors of the same type are provided, such as when multiple temperature sensors 3 are provided, the average value of the measured values ​​of the multiple sensors 3 at each point in time may be used as reference data, or the measured values ​​of the multiple sensors 3 may be used as they are. The time at which the mold growth risk is estimated may be the present, the future, or the past. The operation of this embodiment other than that described above is the same as in Embodiment 1.

[0083] In the example above, the input data was corrected with reference data before being input to the estimation model. However, the model is not limited to this; the reference data can also be included as input data during machine learning, allowing the estimation model to output estimation results that reflect the correction made by the reference data. In this case, using the estimation model is equivalent to correcting the input data with the reference data.

[0084] In this way, the mold growth estimation device 1b can estimate the short-term mold growth risk, taking into account the effects of time, season, climate change, fluctuations in the number of visitors, fluctuations in business hours, and fluctuations in air conditioning settings, by correcting the measurement data obtained from full-surface data measurement using the measurement data from the permanently installed sensor 3.

[0085] Furthermore, in the example shown in Figure 15, a correction unit 135 is added to the estimation unit 13 of the mold occurrence estimation device 1 of Embodiment 1. Similarly, a correction unit 135 may be added to the estimation unit 13a of the mold occurrence estimation device 1a of Embodiment 2. In this case, the correction unit 135 corrects the first input data, which is the input data to the first estimation model for estimating the short-term mold occurrence risk, using reference data. In addition, the correction unit 135 corrects the second input data, which is the input data to the second estimation model for estimating the long-term mold occurrence risk, using reference data. The inference unit 133a inputs the corrected first input data to the first estimation model and the corrected second input data to the second estimation model. If preprocessing is performed, the inference unit 133a performs preprocessing on the corrected first input data and second input data, for example. As described above, the order of preprocessing and correction may be reversed. In this case as well, the first estimation model and the second estimation model may output estimation results that also reflect the correction using reference data.

[0086] By adding a correction unit 135 to the mold growth estimation device 1a of Embodiment 2, it is possible to estimate the long-term risk of mold growth, taking into account the effects of time, season, climate change, fluctuations in the number of visitors, fluctuations in business hours, and fluctuations in air conditioning settings.

[0087] The mold growth estimation device 1b of this embodiment is implemented, similar to the mold growth estimation device 1 of Embodiment 1, by, for example, the computer system illustrated in Figure 13. The correction unit 135 shown in Figure 15 is implemented by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the correction unit 135. The mold growth estimation device 1b may be implemented by multiple computer systems. Furthermore, for example, the mold growth estimation device 1b may be implemented by a cloud computing system.

[0088] Embodiment 4. Figure 17 shows an example of the configuration of the mold growth estimation device 1c according to Embodiment 4. The mold growth estimation device 1c shown in Figure 17 is the same as the mold growth estimation device 1 of Embodiment 1, except that a countermeasure support unit 15 is added. Components having the same functions as in Embodiment 1 are denoted by the same reference numerals as in Embodiment 1, and redundant explanations are omitted. The differences from Embodiment 1 will be mainly described below.

[0089] The countermeasure support unit 15 uses the mold growth risk information generated by the estimation unit 13 to estimate the cause of the high risk of mold growth in areas estimated to have a high risk of mold growth, determines proposed countermeasures corresponding to the cause, and outputs the estimated cause and countermeasures to the output unit 14. The countermeasure support unit 15 may also use at least a portion of the input data used to generate the mold growth risk information in addition to the mold growth risk information to estimate the cause and determine proposed countermeasures. Furthermore, the countermeasure support unit 15 may also use information other than the mold growth risk information and input data, such as information indicating the location of equipment or images of areas including areas estimated to be prone to mold growth, to estimate the cause and determine proposed countermeasures. Hereinafter, the information that becomes the input to the countermeasure support unit 15 will also be referred to as mold-related information. As described above, mold-related information includes mold growth risk information. Mold-related information may also include at least a portion of the input data used to generate the mold growth risk information and at least one of the other information.

[0090] In this example, the estimation unit 13 estimates mold growth risk information using the estimation method described in Embodiment 1. However, the method for generating mold growth risk information is not limited to the example described in Embodiment 1, and may be the method described in Embodiment 2 or Embodiment 3, or it may be generated by a method different from the methods described in Embodiments 1 to 3. Furthermore, the mold growth risk information in this embodiment only needs to be information where the estimated mold growth risk results for multiple locations are associated with their locations, and the method for estimating mold growth risk and generating its distribution is not limited to the methods described in Embodiments 1 to 3. In this case as well, the mold growth risk information may be information showing the distribution of mold growth risk.

[0091] The countermeasure support unit 15 includes a factor analysis unit 151 and a countermeasure proposal unit 152. In this embodiment, if the above-mentioned other information is used as mold-related information, the acquisition unit 11 acquires the other information and outputs it to the factor analysis unit 151.

[0092] The factor analysis unit 151 acquires mold-related information and performs factor analysis using the acquired mold-related information to estimate the causes of the high mold growth risk in areas where the estimated mold growth risk is high. Hereinafter, estimating the causes of the high mold growth risk in areas where the mold growth risk is estimated will also be referred to simply as "estimating the causes." For example, the factor analysis unit 151 can calculate the contribution of each input variable by performing a contribution analysis of each input variable in the input data using a trained model generated by supervised learning or semi-supervised learning as described in Embodiment 1 and Embodiment 2. This makes it possible to calculate how much each input variable, such as surface temperature and moisture content, contributes to the high mold growth risk in areas where the mold growth risk is estimated. Therefore, by performing a contribution analysis and identifying the input variables that influence the high mold growth risk, the causes can be estimated. Methods for contribution analysis include, but are not limited to, Permutation Importance, PDP (Partial Dependence Plot), SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model-Agnostic Explanations), and ALE (Accumulated Local Effects). Furthermore, the factor analysis unit 151 may estimate causes using, for example, a rule-based approach, or using generative AI (Artificial Intelligence). Alternatively, the factor analysis unit 151 may estimate causes using, for example, supervised or semi-supervised learning with mold-related information as explanatory variables, or using mathematical optimization techniques. The method for estimating causes is not limited to these examples.

[0093] The countermeasure proposal unit 152 uses at least one of the causes estimated by the factor analysis unit 151 and mold-related information to determine countermeasures to suppress mold growth in areas with an estimated high risk of mold growth, i.e., to propose countermeasures. The countermeasure proposal unit 152 also generates countermeasure support information including the estimated causes and countermeasures (proposed countermeasures) and outputs it to the output unit 14. Countermeasures include, but are not limited to, anti-mold coating in areas with an estimated high risk of mold growth, anti-mold coating after bleach pretreatment in areas where mold has already grown, changing the location of equipment, upgrading equipment (new installation, additional installation, replacement with high-performance equipment), changing settings of air conditioning equipment (changing airflow direction, changing set temperature, etc.), and changing settings of ventilation equipment. Furthermore, regarding anti-mold coating, the unit may also propose which type of coating agent to use. For example, if mold is growing on the ceiling, applying an anti-mold coating after bleach pretreatment can restore the aesthetic appearance while also providing mold resistance. Furthermore, countermeasures may be modified depending on the estimated high risk of mold growth in certain areas. For example, while it may not be advisable to apply anti-mold coating to areas that people frequently touch, it can be proposed for ceilings, where people do not touch, as a countermeasure specifically for such areas. Additionally, permanent measures against mold growth can be implemented by proposing measures to enhance equipment. Moreover, by reviewing the countermeasures proposed by the user before equipment installation, these measures can be incorporated into the interior design, creating an environment less prone to mold growth.

[0094] The countermeasure support information may include multiple countermeasures. The countermeasure proposal unit 152 may, for example, use the trained models described in Embodiment 1 and Embodiment 2 to solve a minimization problem (optimization problem) aimed at minimizing the mold growth risk or bringing it below a threshold, using input data representing the current state of areas with a high risk of mold growth as initial values, thereby finding the combination of input variables necessary to reduce the risk of mold growth. Specifically, optimization methods such as mathematical optimization or evolutionary algorithms can be used. The difference between the combination of input variables obtained by solving the minimization problem and the input data representing the current state of the area, which was set as the initial value, represents the content of the countermeasures to be implemented. Specifically, for example, if the input variables include whether or not an anti-mold coating has been applied, and the input data representing the current state of the area, which was set as the initial value, shows no coating application, but the combination of input variables obtained by optimization shows that a coating has been applied, then coating application should be implemented as a countermeasure. Also, if there is a difference in surface temperature, room temperature, humidity, moisture content, etc., it means that countermeasures should be taken by adding or changing the settings of equipment such as air conditioning equipment and blowers. In this case, based on the positional relationship between the location in question and existing or newly installed equipment, it is possible to request more specific countermeasures by using a physical model or another machine learning model to determine how the surface temperature, room temperature, humidity, and moisture content of the location will change due to at least one of the addition of equipment or the modification of its settings.

[0095] Furthermore, the countermeasure proposal unit 152 may determine the proposed countermeasures using a rule-based approach, or it may determine the proposed countermeasures using generative AI. Alternatively, the countermeasure proposal unit 152 may determine the proposed countermeasures by estimating them through supervised learning with mold-related information as explanatory variables, or it may determine the proposed countermeasures by estimating them using mathematical optimization techniques. The method for determining the proposed countermeasures is not limited to these examples.

[0096] When cause estimation is performed on a rule basis, for example, the factor analysis unit 151 may use mold growth risk information to identify locations estimated to have a high risk of mold growth, and then estimate the cause using the identified locations, the values ​​of the mold-related information, and the rules described above.

[0097] Possible causes of mold growth include, for example, high humidity, distance from ventilation openings, proximity to equipment that affects moisture content, and the ease with which humid air enters during the summer months. When rules are used to estimate the causes, conditions related to the values ​​of mold-related information corresponding to each cause are associated as rules. For example, for the cause of high humidity, a condition may be set that the humidity of the location with an estimated high risk of mold growth in the measurement data within the input data, or the humidity around that location, is above a threshold. Also, for example, for the cause of distance from ventilation openings, a condition may be set that the distance between the location with an estimated high risk of mold growth based on the information indicating the location and the nearest ventilation opening is above a threshold. These causes and conditions are examples, and the content and conditions of causes are not limited to the examples given above.

[0098] Furthermore, when the decision on proposed countermeasures is made on a rule basis, for example, a rule may be established that corresponds to the type of location with a high risk of mold growth, which is estimated to be the cause (factor), and the countermeasure proposal unit 152 may refer to this rule to determine the countermeasure corresponding to the cause estimated by the factor analysis unit 151. Alternatively, similar to the estimation of the cause described above, a rule may be established that corresponds to the conditions regarding the values ​​of mold-related information, and the countermeasure proposal unit 152 may use this rule and the mold-related information to determine the countermeasure.

[0099] When generating AI is used to estimate the cause (factor), it is desirable to use a model that has learned the above-mentioned factors such as high humidity, distance from ventilation openings, proximity to equipment that affects moisture content, and the fact that humid air is more likely to enter when near an entrance in the summer, or to include these factors in the prompts.

[0100] Generative AI is a functional unit that outputs data by using a model generated by machine learning (a trained model) as input to the trained model. The model of the generative AI may be, for example, a neural network model, a CNN (Convolutional Neural Network) model, a RNN (Recurrent Neural Network), a VAE (Variational AutoEncoder), a GAN (Generative Adversarial Networks), a diffusion model, a transformer model, a Large Language Model (LLM), a Visual Language Model (VLM), a Bidirectional Encoder Representations from Transformers (BERT), a Generative Pre-trained Transformer (GPT®), or a model called CLIP (Contrastive Language Image Pre-training). Note that the above models are not mutually exclusive; for example, LLM, VLM, BERT, and GPT are included in the Transformer model category. Furthermore, generative AI models also include what are called multimodal models, which are trained by combining multiple different types of data. For example, a generative AI may generate data using two or more types of data from among text data, images, and audio as input, and the output data may also be two or more types from among text data, images, and audio.

[0101] When factors are estimated by the generating AI, for example, the prompts that the factor analysis unit 151 inputs to the generating AI model include instructions to estimate the causes of areas with a high risk of mold growth using mold-related information. The prompts are information indicating instructions or questions for the generating AI, and when prompts are input to the generating AI, the generating AI generates information based on the prompts. Of the prompts used by the countermeasure support unit 15, the parts other than the mold-related information are, for example, created in advance and stored in the information storage unit 12. The generating AI model may be stored in the factor analysis unit 151, in a generation unit (not shown) within the mold growth estimation device 1c, or in a generation device (not shown) outside the mold growth estimation device 1c. In other words, the generating AI may be the factor analysis unit 151, in a generation unit (not shown) within the mold growth estimation device 1c, or in a generation device (not shown) outside the mold growth estimation device 1c. The factor analysis unit 151 inputs prompts to the generation AI model that include instructions to estimate the causes of areas with a high risk of mold growth, and obtains the generated data, which is the output of the generation AI model, as the estimated cause.

[0102] Furthermore, when a decision on countermeasures to be proposed by the generating AI is generated, for example, the prompts that the countermeasure proposal unit 152 inputs to the generating AI model include instructions to propose countermeasures using mold-related information, or instructions to propose countermeasures based on the cause estimated by the factor analysis unit 151. Also, as described above, by training the generating AI model with information such as whether an anti-mold coating is suitable for the ceiling, or by including in the prompts information such as whether an anti-mold coating is suitable for the situation, the generating AI can propose appropriate countermeasures. The generating AI model may be stored in the countermeasure proposal unit 152, in a generation unit (not shown) within the mold occurrence estimation device 1c, or in a generation device (not shown) outside the mold occurrence estimation device 1c. In other words, the generating AI may be the countermeasure proposal unit 152, in a generation unit (not shown) within the mold occurrence estimation device 1c, or in a generation device (not shown) outside the mold occurrence estimation device 1c.

[0103] If the generating AI is a generating unit (not shown) within the mold growth estimation device 1c or a generating device (not shown) outside the mold growth estimation device 1c, the same model may be used for the generating AI used to estimate the cause and the generating AI used to determine the proposed countermeasures. In this case, by including instructions in the prompt to estimate the cause using mold-related information and propose countermeasures, the generating AI may generate countermeasure support information indicating the cause and proposed countermeasures.

[0104] When cause estimation is performed by supervised learning, the factor analysis unit 151 includes, for example, a learning unit and an inference unit. The learning unit uses multiple training datasets, each containing mold-related information and corresponding ground truth data (causes of mold occurrence), to generate a trained model for inferring the cause of mold occurrence from the mold-related information using machine learning. The inference unit infers (estimates) the cause by inputting the mold-related information into the trained model. Alternatively, instead of the factor analysis unit 151 having a learning unit, a learning device outside the mold occurrence estimation device 1c may generate the trained model, and the inference unit may store the trained model generated by the learning device. The supervised learning algorithm can be any algorithm, such as a neural network or a support vector machine, similar to the supervised learning described in Embodiment 1. When semi-supervised learning is used, the configuration, learning, and inference procedures of the factor analysis unit 151 are the same as when supervised learning is used. The semi-supervised learning algorithm can be any algorithm, similar to the semi-supervised learning described in Embodiment 1, such as Graph Convolutional Networks, Label Diffusion, Label Propagation, Self-Training, Mean Teacher, or Variational Autoencoder.

[0105] When the proposed countermeasures are determined by supervised learning, the countermeasure proposal unit 152 comprises, for example, a learning unit and an inference unit. The learning unit uses multiple training datasets, each containing mold-related information and corresponding correct data of proposed countermeasures, to generate a trained model for determining proposed countermeasures from mold-related information using machine learning. The inference unit infers (determines) proposed countermeasures by inputting mold-related information into the trained model. Alternatively, instead of the countermeasure proposal unit 152 having a learning unit, a learning device outside the mold occurrence estimation device 1c may generate the trained model, and the inference unit may store the trained model generated by the learning device. The supervised learning algorithm can be any algorithm, such as a neural network or a support vector machine, similar to the supervised learning described in Embodiment 1. When semi-supervised learning is used, the configuration, learning, and inference procedures of the countermeasure proposal unit 152 are the same as when supervised learning is used. The semi-supervised learning algorithm can be any algorithm, similar to the semi-supervised learning described in Embodiment 1, such as Graph Convolutional Networks, Label Diffusion, Label Propagation, Self-Training, Mean Teacher, or Variational Autoencoder.

[0106] Furthermore, the factor analysis unit 151 and the countermeasure proposal unit 152 may be integrated. Conditions for the values ​​of mold-related information, the corresponding causes, and proposed countermeasures may be defined as a set of rules, and the countermeasure support unit 15 may use these rules and the mold-related information to determine the estimated causes and proposed countermeasures. Alternatively, the countermeasure support unit 15 may include the causes and proposed countermeasures as ground truth data, and use a trained model generated by machine learning using multiple training datasets containing mold-related information and corresponding ground truth data to obtain the estimated causes and proposed countermeasures as the output of the trained model.

[0107] Next, the countermeasure support process (countermeasure support method) of this embodiment will be described. Figure 18 is a flowchart showing an example of the countermeasure support process procedure of this embodiment. The countermeasure support unit 15 of the mold occurrence estimation device 1 acquires mold-related information (step S21). In detail, the countermeasure support unit 15 acquires information indicating mold occurrence risk information and input data from the estimation unit 13. Note that for data shown as a distribution among the input data, the estimation unit 13 may acquire information indicating the distribution of said data. In addition, if other information other than mold occurrence risk information and input data is used, the countermeasure support unit 15 acquires the other information from the acquisition unit 11. Note that the mold occurrence risk information and input data may be stored in the information storage unit 12, and the countermeasure support unit 15 may acquire this information from the information storage unit 12.

[0108] The countermeasure support unit 15 performs a factor analysis of mold occurrence and determines proposed countermeasures based on mold-related information (step S22). Specifically, the factor analysis unit 151 performs the factor analysis, and the countermeasure proposal unit 152 determines the proposed countermeasures. As described above, the factor analysis unit 151 and the countermeasure proposal unit 152 may be integrated, and the factor analysis and the determination of proposed countermeasures may be performed simultaneously.

[0109] The mold growth estimation device 1c outputs countermeasure support information indicating the cause and countermeasures (step S23). Specifically, the countermeasure proposal unit 152 generates countermeasure support information indicating the cause and proposed countermeasures, outputs the generated countermeasure support information to the output unit 14, and the output unit 14 outputs the countermeasure support information. As described in Embodiment 1, the output from the output unit 14 may be displayed or transmitted to the terminal device 2 or display device. As mentioned above, the countermeasure support information may be generated by the generation AI.

[0110] After step S23, the mold growth estimation device 1c determines whether or not to terminate the countermeasure support process (step S24). For example, the mold growth estimation device 1 determines to terminate the countermeasure support process if the user instructs it to terminate. If the countermeasure support process is to be terminated (step S24 Yes), the mold growth estimation device 1 terminates the countermeasure support process. If the countermeasure support process is not to be terminated (step S24 No), the process from step S21 is repeated.

[0111] Figure 19 shows an example of a display screen for countermeasure support information in this embodiment. In the example shown in Figure 19, when mold growth risk information is displayed and a part with a high risk of mold growth is selected, the countermeasure display screen 401 is displayed. The countermeasure display screen 401 displays information indicating the cause (factor) and proposed countermeasures for the area estimated to have a high risk of mold growth as specified by the user. The display screen exemplified in Figure 19 is displayed by the output unit 14, terminal device 2, or display device. In the example shown in Figure 19, three countermeasures, Option 1 to Option 3, are displayed. In the example shown in Figure 19, regarding Option 3, strengthening the air conditioning equipment, selecting the button 402 labeled "Display recommended placement location" displays the recommended placement location for the air conditioning equipment. Note that Figure 19 shows an example where the countermeasure display screen 401 is displayed when mold growth risk information is displayed, but the countermeasure display screen 401 may be displayed separately from the mold growth risk information. For example, a countermeasure display button may be displayed on the mold growth risk information display screen, and when the countermeasure display button is pressed, the countermeasure display screen 401 may be displayed. Figure 19 is an example, and the display method and content are not limited to the example shown in Figure 19.

[0112] In the examples described above, both the estimated cause and the proposed solution are presented to the user. However, either the estimated cause or the proposed solution may be presented to the user. In other words, the solution support information may include at least one of the cause or the solution. For example, the solution support unit 15 may perform only factor analysis and generate solution support information showing the estimated cause. Alternatively, the solution support unit 15 may determine the proposed solution and generate solution support information showing the proposed solution. In this case as well, the results of the factor analysis, i.e., the estimated cause, may be used to determine the proposed solution, but the solution support unit 15 does not include the cause in the solution support information it generates.

[0113] In Figure 17, an example is shown in which the mold growth estimation device 1c is equipped with a countermeasure support unit 15. However, a separate mold countermeasure support device may be provided in addition to the mold growth estimation device that estimates the risk of mold growth.

[0114] Figure 20 shows an example configuration of the mold suppression support system 5 according to this embodiment. The mold suppression support system 5 shown in Figure 20 comprises the mold occurrence estimation device 1 described in Embodiment 1 and the mold countermeasure support device 4. The mold countermeasure support device 4 comprises a countermeasure support unit 15, an acquisition unit 41, and an output unit 42, similar to the example shown in Figure 17. The acquisition unit 41 acquires mold-related information. The acquisition unit 41 acquires mold occurrence risk information and input data from the mold occurrence estimation device 1. The output unit 42 outputs countermeasure support information. The output from the output unit 42 may be displayed, similar to the output unit 14 of the mold occurrence estimation device 1, or transmitted to a terminal device 2 or a display device. In addition, although Figure 20 shows the mold occurrence estimation device 1 of Embodiment 1, the mold occurrence estimation device 1a of Embodiment 2 or the mold occurrence estimation device 1b of Embodiment 3 may be used instead of the mold occurrence estimation device 1. Alternatively, instead of the mold growth estimation device 1, a mold growth estimation device that generates mold growth risk information in a manner different from that of Embodiments 1 to 3 may be used.

[0115] The mold growth estimation device 1c of this embodiment is implemented, similar to the mold growth estimation device 1 of Embodiment 1, by, for example, the computer system illustrated in Figure 13. The countermeasure support unit 15 shown in Figure 17 is implemented by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the countermeasure support unit 15. The mold growth estimation device 1c may be implemented by multiple computer systems. Furthermore, for example, the mold growth estimation device 1c may be implemented by a cloud computing system.

[0116] The mold prevention support device 4 shown in Figure 20 can also be implemented, for example, by the computer system illustrated in Figure 13. The countermeasure support unit 15 shown in Figure 20 is implemented by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the countermeasure support unit 15. The acquisition unit 41 shown in Figure 20 is implemented by the communication unit 105 and input unit 102 shown in Figure 13. The output unit 42 shown in Figure 20 is implemented by at least one of the display unit 104 and communication unit 105 shown in Figure 13. The mold prevention support device 4 may be implemented by multiple computer systems. Also, for example, the mold prevention support device 4 may be implemented by a cloud computer system. Furthermore, the generation device having the function of a generation AI as described above can also be implemented, for example, by the computer system illustrated in Figure 13. The generation device may be implemented by multiple computer systems. Also, for example, the generation device may be implemented by a cloud computer system.

[0117] As described above, in this embodiment, the mold occurrence estimation device 1c or the mold control support device 4 outputs countermeasure support information that indicates at least one of the causes of mold occurrence and proposed countermeasures to suppress mold occurrence in the target area, using mold occurrence risk information that shows the distribution of mold occurrence risk in the target area. This makes it possible to provide the user with useful information for suppressing mold occurrence and to support the efficient implementation of countermeasures to suppress mold occurrence. Furthermore, when using the long-term mold occurrence risk estimation results described in Embodiment 2, the user can consider the necessary countermeasures with ample time, and can arrange to implement countermeasures while avoiding peak seasons.

[0118] Embodiment 5. Figure 21 shows an example of the configuration of the mold suppression support system 5a according to Embodiment 5. The mold suppression support system 5a comprises a mold occurrence estimation device 1d and a countermeasure execution device 6. The mold occurrence estimation device 1d is the same as the mold occurrence estimation device 1 of Embodiment 1, except that a device control unit 16 is added. Components having the same functions as in Embodiment 1 are denoted by the same reference numerals as in Embodiment 1, and redundant descriptions are omitted. The following mainly describes the differences from Embodiment 1.

[0119] The countermeasure execution device 6 includes, but is not limited to, at least one of the following: an air conditioning system, a ventilation system, a blower, and a dehumidifier. Thus, the countermeasure execution device 6 is a device capable of controlling the airflow, temperature, humidity, etc., in areas where mold is likely to grow, i.e., areas where the risk of mold growth is estimated to be high. The device control unit 16 uses the mold growth risk information to control the countermeasure execution device 6 to suppress mold growth in areas where the risk of mold growth is estimated to be high. Specifically, the device control unit 16 determines the control content to suppress mold growth in areas where the risk of mold growth is estimated to be high, generates a control signal to control the countermeasure execution device 6 based on the determined control content, and outputs the generated control signal to the countermeasure execution device 6. The information necessary to determine the control content will hereafter be referred to as control information. The control content of the countermeasure execution device 6 indicates what kind of control the countermeasure execution device 6 will perform in order to suppress mold growth in areas where mold is estimated to be likely to grow.

[0120] The control information includes mold growth risk information. The control information may include at least a portion of the input data used to estimate the mold growth risk. The control information may also include information other than the mold growth risk information and input data. For example, the control information may include at least one of the following: a layout diagram showing the layout of each piece of equipment, including the countermeasure execution device 6, in the target area; images of the target area; and the materials of items such as ceilings and walls. The control information may be the same as or different from the mold-related information described in Embodiment 4. Of the control information, the information other than the mold growth risk information and input data may be acquired by the acquisition unit 11 or stored in the information storage unit 12.

[0121] The control performed by the countermeasure execution device 6 to suppress mold growth in areas estimated to be prone to mold growth may, for example, involve controlling the air conditioning system to keep the temperature of areas prone to mold growth below a certain value, or lowering the humidity through dehumidification. Alternatively, the control to suppress mold growth in areas estimated to be prone to mold growth may involve controlling the airflow direction of the air conditioning system, or controlling the operation of a ventilation device to promote ventilation. It may also involve adjusting the angle of the air outlets or ducts of the countermeasure execution device 6 to promote ventilation in the relevant areas. These control contents may be predetermined as correspondence information based on the location and material of areas estimated to be prone to mold growth and corresponding measurement data, or the control contents may be determined by a trained model generated by generative AI or supervised learning.

[0122] When a generation AI is used to determine the control content, it is desirable to pre-train the generation AI model on what kind of control is appropriate under what conditions, or to include this in the prompt. When a generation AI is used to determine the control content, the device control unit 16 may have the function of a generation AI, a generation unit having the function of a generation AI may be provided within the mold occurrence estimation device 1d, or a generation device having the function of a generation AI outside the mold occurrence estimation device 1d may be used. For example, the prompt may include control information and an instruction to generate a control signal to control the countermeasure execution device 6 so as to suppress mold in areas where the risk of mold occurrence is estimated to be high, and the device control unit 16 inputs this prompt to the generation AI. The part of the prompt other than the control information is, for example, created in advance and stored in the information storage unit 12. The device control unit 16 controls the countermeasure execution device 6 by outputting the control signal generated by the generation AI to the countermeasure execution device 6.

[0123] Furthermore, if the corresponding information is predetermined, the device control unit 16 determines the control content based on the corresponding information, the location and material of the area estimated to be prone to mold growth, and the corresponding measurement data for that area, generates a control signal based on the control content, and outputs the control signal to the device control unit 16. If the control content is determined by supervised learning, a trained model for determining the control content is generated in advance from the control information, and the device control unit 16 determines the control content by inputting the control information into the trained model. Based on the determined control content, the device control unit 16 generates a control signal to control the countermeasure execution device 6 and outputs the determined control signal to the countermeasure execution device 6. Note that the method for determining the control content of the countermeasure execution device 6 is not limited to the example described above.

[0124] The mold growth estimation device 1d of this embodiment is implemented, similar to the mold growth estimation device 1 of Embodiment 1, by, for example, the computer system illustrated in Figure 13. The device control unit 16 shown in Figure 21 is implemented by the control unit 101 shown in Figure 13 executing a computer program stored in the storage unit 103 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the device control unit 16. The mold growth estimation device 1d may be implemented by multiple computer systems. Furthermore, for example, the mold growth estimation device 1d may be implemented by a cloud computing system.

[0125] In addition, Figure 21 shows an example in which a device control unit 16 is added to the configuration example shown in Embodiment 1, but it is not limited to this, and the estimation units 13a and 13b described in Embodiments 2 and 3 may be used instead of the estimation unit 13. Furthermore, instead of the estimation unit 13, an estimation unit that generates mold growth risk information in a different way than that of Embodiments 1 to 3 may be used.

[0126] For example, if short-term mold growth risk information is generated, the equipment control unit 16 may perform the above-described control if there are areas with a high risk of mold growth based on the mold growth risk information, such as areas where the mold growth risk exceeds a threshold. Furthermore, if long-term mold growth risk information is generated, the control content may be defined to change according to the season.

[0127] Furthermore, an equipment control unit 16 may be added to the configuration example shown in Figure 17 of Embodiment 4. In this case, for example, control by a countermeasure execution device 6 may be included as one of the countermeasures. For example, the countermeasure support unit 15 determines the countermeasures to be proposed as described in Embodiment 4, and if the determined countermeasures include control of a countermeasure execution device 6, it instructs the equipment control unit 16 to control the countermeasure execution device 6. When the equipment control unit 16 receives the above instruction from the countermeasure support unit 15, it determines the control content of the countermeasure execution device 6 based on the control information, generates a control signal to control the countermeasure execution device 6 based on the determined control content, and outputs the generated control signal to the countermeasure execution device 6. In this case, the countermeasure support unit 15 may determine the control content of the countermeasure execution device 6 to suppress mold growth in areas that are estimated to be prone to mold growth, and output the determined control content to the equipment control unit 16. Also, in this case, AI generation may be used for determining the control content of the equipment control unit 16 and for the countermeasure support unit 15. Furthermore, when the countermeasure support unit 15 instructs the generation AI model to propose countermeasures, it may also include an instruction to generate a control signal if the control of the countermeasure execution device 6 is effective as a countermeasure, so that the generation AI outputs a control signal along with the proposed countermeasure. In this case, the device control unit 16 only needs to output the control signal output from the generation AI to the countermeasure execution device 6.

[0128] Furthermore, if the countermeasure support unit 15 includes control of the countermeasure execution device 6 in the decided countermeasure, it may instruct the device control unit 16 to control the countermeasure execution device 6 regardless of instructions from the user. Alternatively, when countermeasure support information including control of the countermeasure execution device 6 is displayed as a countermeasure, a button or the like that accepts instructions to execute control of the countermeasure execution device 6 may be displayed, and the device control unit 16 may execute control of the countermeasure execution device 6 when the user presses the button. Note that the method of receiving instructions from the user to execute control of the countermeasure execution device 6 is not limited to the examples described above.

[0129] Furthermore, a countermeasure execution device 6 may be added to the mold suppression support system 5 shown in Figure 20 of Embodiment 4, and the mold countermeasure support device 4 may be equipped with a device control unit 16 instead of a countermeasure support unit 15, with the device control unit 16 performing the operations of this embodiment. Alternatively, a countermeasure execution device 6 may be added to the mold suppression support system 5 shown in Figure 20 of Embodiment 4, and a device control unit 16 may be added to the mold countermeasure support device 4 in the mold suppression support system 5. In these cases as well, the mold occurrence estimation devices 1a and 1b described in Embodiments 2 and 3 may be used instead of the mold occurrence estimation device 1, or a mold occurrence estimation device that generates mold occurrence risk information in a manner different from the method described in Embodiments 1 to 3 may be used.

[0130] In this embodiment, the equipment control unit 16 controls the countermeasure execution device 6 to suppress mold growth in areas where mold is estimated to be likely to occur. This reduces the user's effort and suppresses mold growth. Furthermore, when long-term mold growth risk information is used, the control of the countermeasure execution device 6 may be performed in conjunction with the enhancement of the countermeasure execution device 6, such as air conditioning equipment. That is, the enhancement of the countermeasure execution device 6, such as air conditioning equipment, may be presented to the user as a countermeasure, and if the user selects this countermeasure, the enhanced countermeasure execution device 6 may be controlled to suppress mold growth in areas where mold is estimated to be likely to occur.

[0131] Embodiment 6. Figure 22 shows an example of the configuration of the mold growth estimation device 1e according to Embodiment 6. The mold growth estimation device 1e is the same as the mold growth estimation device 1c shown in Figure 17 of Embodiment 4, except that it is equipped with a countermeasure support unit 15a instead of a countermeasure support unit 15. Components that have the same functions as Embodiments 1 and 4 are denoted by the same reference numerals as in Embodiments 1 and 4, and redundant explanations are omitted. The following mainly describes the differences from Embodiments 1 and 4.

[0132] The countermeasure support unit 15a is the same as the countermeasure support unit 15 of Embodiment 4, except that a cost calculation unit 153 is added. The cost calculation unit 153 calculates the cost required for the countermeasures proposed by the countermeasure proposal unit 152 and outputs cost information showing the cost for each countermeasure to the output unit 14. The output unit 14 outputs the cost information. The information necessary for cost calculation is stored in advance in the information storage unit 12. The information necessary for cost calculation may be updated as appropriate. The information necessary for cost calculation includes the cost of introducing air conditioning equipment, vents, ventilation devices, etc., the material cost of materials used for countermeasures such as coating materials, the construction cost for coating application, etc., and the running cost of operating air conditioning equipment, vents, ventilation devices, etc.

[0133] The information required for cost calculation may include multiple types of information for a single item, such as multiple manufacturers, multiple contractors, material manufacturers, and material product names. For example, for coating materials, it may include information on multiple product names or product numbers of coating materials. If the proposed measures specify which of several types to propose, the cost calculation unit 153 calculates the cost based on those measures. If the proposed measures do not specify which of several types to propose, the cost may be calculated for each of the several types, or the cost may be calculated by selecting the cheapest option.

[0134] Figure 23 shows an example of a cost information display screen in this embodiment. In the example shown in Figure 23, the approximate costs for each of the proposed countermeasures, Option 1 to Option 3, are displayed. In Figure 23, a range of values ​​is shown for the approximate costs, but if the cost is uniquely determined, a single value may be displayed. Furthermore, as mentioned above, if the coating material is not specified in the countermeasure, multiple costs may be displayed for a single option, along with conditions, such as displaying the cost for each type of coating agent. The conditions may include the material part number, equipment manufacturer, equipment model number, and contractor, and their respective characteristics may also be displayed.

[0135] Furthermore, in the example shown in Figure 23, order processing buttons 501 are displayed for both Option 2 and Option 3, and when the user presses the order processing button 501, an order processing is performed to execute the countermeasure corresponding to the order processing button 501. The order processing may be performed by the cost calculation unit 153 or by an order processing unit not shown. In the order processing, for example, an order form for the order to execute the countermeasure may be generated, or the order form may be created and then sent to the supplier, or a process may be performed to transition the screen to a web page that accepts orders. The order form may be generated by the generation AI, for example, by inputting the order form format into the generation AI. Alternatively, the cost calculation unit 153 may be equipped with a function to fill in the order details according to the countermeasure in the order form format, and the order form may be generated without using the generation AI.

[0136] Although the details of the countermeasures are not shown in Figure 23, the details of the countermeasures for each proposal may be displayed together. Furthermore, the costs corresponding to each proposal may also be displayed on the display screen shown in Figure 19. Figure 23 is just one example, and the method of displaying the costs of the countermeasures is not limited to these examples.

[0137] Furthermore, the countermeasure proposal unit 152 can propose countermeasures so that the total cost fits within the customer's budget. Specifically, for example, in the mold growth risk minimization problem described above, it finds a combination of input variables whose objective function is to minimize the mold growth risk estimated by the trained model while simultaneously ensuring that the cost calculated by the cost calculation unit 153 fits within the customer's budget. However, the statement that the cost fits within the customer's budget is just one example of an objective function related to cost; the objective function may also be set to prioritize reducing the mold growth risk even if it exceeds the cost, and various methods can be applied to set the objective function; it is not limited to this example.

[0138] The mold growth estimation device 1e of this embodiment is implemented, similar to the mold growth estimation device 1 of Embodiment 1, by, for example, the computer system illustrated in Figure 13. The countermeasure support unit 15a shown in Figure 22 is implemented by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the countermeasure support unit 15a. The mold growth estimation device 1e may be implemented by multiple computer systems. Furthermore, for example, the mold growth estimation device 1e may be implemented by a cloud computing system.

[0139] In Figure 22, an example is shown in which the mold occurrence estimation device 1e includes the estimation unit 13 described in Embodiment 1. However, the device is not limited to this, and the estimation units 13a and 13b described in Embodiments 2 and 3 may be used instead of the estimation unit 13. Alternatively, an estimation unit that generates mold occurrence risk information in a manner different from that of Embodiments 1 to 3 may be used instead of the estimation unit 13. Furthermore, the countermeasure support unit 15a of this embodiment may be added to the mold occurrence estimation device 1d of the mold suppression support system 5a described in Embodiment 5.

[0140] Furthermore, the mold control support device 4 of the mold suppression support system 5 shown in Figure 20 of Embodiment 4 may include a countermeasure support unit 15a instead of a countermeasure support unit 15. Alternatively, the mold control support device 4 may include a countermeasure support unit 15a instead of a countermeasure support unit 15, and also include the equipment control unit 16 of Embodiment 5. In these mold suppression support systems 5 as well, the mold occurrence estimation devices 1a and 1b described in Embodiments 2 and 3 may be used instead of the mold occurrence estimation device 1, or a mold occurrence estimation device that generates mold occurrence risk information in a manner different from the method described in Embodiments 1 to 3 may be used.

[0141] As described above, in this embodiment, the mold occurrence estimation device 1e outputs cost information indicating the cost for each countermeasure. Therefore, the user can select the optimal mold prevention measure while considering the cost.

[0142] Embodiment 7. Figure 24 shows an example of the configuration of the mold growth estimation device 1f according to Embodiment 7. The mold growth estimation device 1f is the same as the mold growth estimation device 1c shown in Figure 17 of Embodiment 4, except that a countermeasure evaluation unit 17 is added. Components having the same functions as Embodiments 1 and 4 are denoted by the same reference numerals as in Embodiments 1 and 4, and redundant explanations are omitted. The following mainly describes the differences from Embodiments 1 and 4.

[0143] In this embodiment, the countermeasure evaluation unit 17 outputs countermeasure effectiveness information indicating the degree of reduction in the mold growth risk from the pre-measure risk, i.e., the pre-measure risk, assuming that the proposed countermeasure has been implemented. For example, the countermeasure evaluation unit 17 modifies data that is estimated to change due to the countermeasure from the input data used to generate the mold growth risk information that formed the basis of the proposed countermeasure, inputs the modified input data to the estimation unit 13, and the estimation unit 13 uses the modified input data to generate mold growth risk information in the same manner as in Embodiment 1. For example, rules may be predetermined to determine how to modify the input data for each countermeasure (or for each combination of countermeasure and the material, arrangement, etc., of areas with a high risk of mold growth), or the modified input data corresponding to the countermeasure (or according to the combination of countermeasure and the material, arrangement, etc., of areas with a high risk of mold growth) may be determined by generational AI or supervised learning other than generational AI. The countermeasure evaluation unit 17 generates countermeasure effectiveness information using the mold growth risk information before the countermeasure and the mold growth risk information after the countermeasure, i.e., the mold growth risk information generated using the modified input data, and outputs the countermeasure effectiveness information to the output unit 14.

[0144] For example, the countermeasure evaluation unit 17 calculates the difference in mold growth risk before and after countermeasures for each location based on mold growth risk information before and after countermeasures, and generates countermeasure effectiveness information by associating this difference with the location. In this case, the countermeasure effectiveness information is information showing the distribution of the difference in mold growth risk before and after countermeasures. Alternatively, the countermeasure evaluation unit 17 may calculate the difference in mold growth risk before and after countermeasures for each location based on mold growth risk information before and after countermeasures, and use the difference for each location to calculate indicators such as the maximum value and average value of multiple differences at different locations, and output these indicators as countermeasure effectiveness information. Alternatively, instead of the difference in mold growth risk before and after countermeasures, information showing the ratio of mold growth risk before and after countermeasures may be used, or the ratio of the difference in mold growth risk before and after countermeasures to the mold growth risk before countermeasures may be used. Alternatively, the information on the effectiveness of countermeasures may include information on the risk of mold growth before and after the countermeasures. For example, by displaying the information on the risk of mold growth before and after the countermeasures side by side, or by switching between them, users can grasp the degree to which the risk of mold growth has been reduced. The information on the effectiveness of countermeasures does not need to include any information that indicates the degree to which the risk of mold growth has been reduced.

[0145] Furthermore, the degree of reduction in mold growth risk may be predetermined for each countermeasure (or for each combination of countermeasures and the materials, arrangement, etc., of areas with a high risk of mold growth), and the countermeasure evaluation unit 17 may calculate the degree of reduction in mold growth risk according to the proposed countermeasures. Alternatively, the degree of reduction in mold growth risk according to each countermeasure (or according to the combination of countermeasures and the materials, arrangement, etc., of areas with a high risk of mold growth) may be calculated by generative AI or other supervised learning methods.

[0146] The output unit 14 outputs information on the effectiveness of the countermeasures. This allows the user to confirm whether the proposed countermeasures have yielded the expected results. Furthermore, if the user feels that the countermeasures are insufficient, they can plan further countermeasures, such as changing the countermeasures or strengthening them by increasing the budget. In addition, by calculating and outputting information on the risk of mold growth after the countermeasures have been implemented at different time intervals, it is possible to understand how far into the future mold will not grow, for example, how far into the future the risk of mold growth will remain below a threshold, and to set a guaranteed period for suppressing mold growth.

[0147] The mold growth estimation device 1f of this embodiment is implemented, similar to the mold growth estimation device 1 of Embodiment 1, by, for example, the computer system illustrated in Figure 13. The countermeasure evaluation unit 17 shown in Figure 24 is implemented by executing a computer program stored in the storage unit 103 shown in Figure 13 by the control unit 101 shown in Figure 13. The storage unit 103 shown in Figure 13 is also used to implement the countermeasure evaluation unit 17. The mold growth estimation device 1f may be implemented by multiple computer systems. Furthermore, for example, the mold growth estimation device 1f may be implemented by a cloud computing system.

[0148] In Figure 24, an example is shown in which the mold occurrence estimation device 1f includes the estimation unit 13 described in Embodiment 1. However, the device is not limited to this, and the estimation units 13a and 13b described in Embodiments 2 and 3 may be used instead of the estimation unit 13. Alternatively, an estimation unit that generates mold occurrence risk information in a manner different from that of Embodiments 1 to 3 may be used instead of the estimation unit 13. Furthermore, the mold occurrence estimation device 1d of the mold suppression support system 5a described in Embodiment 5 may have the countermeasure support unit 15 and the countermeasure evaluation unit 17 of this embodiment added to it. In addition, the mold occurrence estimation device 1f may also calculate the cost for each countermeasure by including a countermeasure support unit 15a instead of the countermeasure support unit 15. In this case as well, the estimation units 13a and 13b described in Embodiments 2 and 3 may be used instead of the estimation unit 13, or an estimation unit that generates mold occurrence risk information in a manner different from that of Embodiments 1 to 3 may be used.

[0149] Furthermore, a countermeasure evaluation unit 17 may be added to the mold countermeasure support device 4 of the mold suppression support system 5 shown in Figure 20 of Embodiment 4. Also, a countermeasure evaluation unit 17 and the equipment control unit 16 of Embodiment 5 may be added to the mold countermeasure support device 4. In these cases as well, a countermeasure support unit 15a may be used instead of a countermeasure support unit 15. In addition, in these mold suppression support systems 5, the mold occurrence estimation devices 1a and 1b described in Embodiments 2 and 3 may be used instead of the mold occurrence estimation device 1, or a mold occurrence estimation device that generates mold occurrence risk information in a manner different from the method described in Embodiments 1 to 3 may be used.

[0150] The configurations shown in the above embodiments are merely examples, and it is possible to combine them with other known technologies, combine different embodiments, and omit or modify parts of the configuration without departing from the gist of the invention.

[0151] 1, 1a, 1b, 1c, 1d, 1e, 1f Mold occurrence estimation device, 2 Terminal device, 3 Sensor, 4 Mold countermeasure support device, 5, 5a Mold suppression support system, 6 Countermeasure execution device, 11, 41 Acquisition unit, 12 Information storage unit, 13, 13a, 13b Estimation unit, 14, 42, 106 Output unit, 15, 15a Countermeasure support unit, 16 Equipment control unit, 17 Countermeasure evaluation unit, 101 Control unit, 102 Input unit, 103 Storage unit, 104 Display unit, 105 Communication unit, 107 System bus, 131, 131a Learning unit, 132, 132a Model storage unit, 133, 133a Inference unit, 134 Distribution generation unit, 135 Correction unit, 151 Factor analysis unit, 152 Countermeasure proposal unit, 153 Cost calculation unit.

Claims

1. A mold growth estimation device comprising: an acquisition unit that acquires input data including at least the surface temperature of items at multiple locations within a target area for which mold growth is to be estimated; an estimation unit that generates mold growth risk information from the input data by performing an estimation process using machine learning to estimate the risk of mold growth at multiple locations from the input data, thereby generating mold growth risk information in which the estimated results of the risk of mold growth at multiple locations are associated with locations; and an output unit that outputs the mold growth risk information.

2. The mold growth estimation device according to claim 1, characterized in that the input data includes the moisture content of the article.

3. The mold growth estimation device according to claim 2, characterized in that the input data includes at least one of the following: temperature within the target area, ambient temperature around the target area, presence or absence of insulation material around the article, presence or absence of mold prevention treatment, humidity within the target area, operating status of equipment in the target area, and arrangement of equipment.

4. The mold growth estimation device according to any one of claims 1 to 3, characterized in that the estimation unit corrects at least a portion of the input data using measurement data acquired from a sensor permanently installed in the target area.

5. The mold occurrence estimation device according to any one of claims 1 to 4, wherein the estimation result is an estimation result of the risk of mold occurrence a first period after the date and time corresponding to the first input data which is the input data, the acquisition unit acquires second input data different from the first input data, and the estimation unit performs an estimation process using machine learning to estimate the risk of mold occurrence at multiple locations from the second input data, thereby generating mold occurrence risk information a second period after the date and time corresponding to the second input data from the second input data acquired by the acquisition unit, and the second period is longer than the first period.

6. The mold growth estimation device according to claim 5, wherein the second input data includes the surface temperature of articles at multiple locations in the target area and the ambient temperature around the target area at the date and time corresponding to the second input data, and the second input data further includes the ambient temperature around the target area for a certain period prior to the date and time corresponding to the second input data.

7. The mold growth estimation device according to claim 5 or 6, characterized in that the estimation unit corrects at least a portion of the second input data using measurement data acquired from a sensor permanently installed in the target area.

8. The mold growth estimation device according to any one of claims 1 to 7, characterized in that the article includes the ceiling of the target area.

9. The mold occurrence estimation device according to any one of claims 1 to 8, characterized in that the layout diagram of the target area and the mold occurrence risk information are output in a linked manner.

10. The mold growth estimation device according to claim 9, characterized in that the layout diagram includes a diagram showing the layout of the ceiling.

11. A mold growth estimation device according to any one of claims 1 to 10, comprising: a countermeasure support unit that generates countermeasure support information including at least one of the causes of the estimated high risk of mold growth in areas estimated to have a high risk of mold growth, and countermeasures to suppress mold growth, wherein the output unit outputs the countermeasure support information.

12. A mold growth estimation device according to any one of claims 1 to 11, comprising: an equipment control unit that uses the mold growth risk information to control the countermeasure execution equipment to suppress the growth of mold in areas where the risk of mold growth is estimated to be high.

13. The mold growth estimation device according to 12, characterized in that the countermeasure implementation device includes at least one of an air conditioning system, a ventilation system, a blower, and a dehumidifier.

14. The mold growth estimation device according to claim 11, comprising: a cost calculation unit that calculates the cost required for the countermeasures and generates cost information indicating the cost, wherein the output unit outputs the cost information.

15. The mold occurrence estimation device according to claim 14, characterized in that the countermeasure support unit determines the countermeasure to be included in the countermeasure support information using the risk of mold occurrence and the cost information.

16. The mold occurrence estimation device according to claim 11, 14, or 15, comprising: a countermeasure evaluation unit that generates countermeasure effectiveness information indicating the degree of reduction in the risk of mold occurrence when the countermeasure is assumed to have been implemented compared to the risk of mold occurrence before the implementation of the countermeasure, wherein the output unit outputs the countermeasure effectiveness information.

17. The mold occurrence estimation device according to any one of claims 1 to 16, characterized in that the mold occurrence risk information is displayed as a contour map showing the distribution of mold occurrence risk.

18. A mold suppression support system comprising: a mold growth estimation device and a countermeasure execution device, wherein the mold growth estimation device includes: an acquisition unit that acquires input data including at least the surface temperature of multiple items at multiple locations within a target area for which mold growth is to be estimated; an estimation unit that generates mold growth risk information from the input data, which is information where the estimated results of the mold growth risk at multiple locations are associated with locations, by performing an estimation process using machine learning to estimate the risk of mold growth at multiple locations from the input data; an output unit that outputs the mold growth risk information; and a device control unit that uses the mold growth risk information to control the countermeasure execution device to suppress mold growth at locations estimated to have a high risk of mold growth.

19. The mold suppression support system according to claim 18, comprising a sensor, wherein the input data includes measurement data from the sensor within the target area.

20. A mold growth estimation method performed by a mold growth estimation device, comprising: a step of acquiring input data that includes at least the surface temperature of an article at multiple locations within a target area for which mold growth is to be estimated; a step of generating mold growth risk information from the input data by performing an estimation process using machine learning to estimate the risk of mold growth at multiple locations from the input data, wherein the estimated result of the risk of mold growth at multiple locations is information associated with location; and a step of outputting the mold growth risk information.

21. A program characterized by causing a computer system to perform the following steps: acquire input data including at least the surface temperature of items at multiple locations within a target area for which mold growth is to be estimated; generate mold growth risk information from the input data by performing an estimation process using machine learning to estimate the risk of mold growth at multiple locations from the input data, wherein the estimated results of the risk of mold growth at multiple locations are associated with their locations; and output the mold growth risk information.

Images