Ventilation fan management device and learning device

Abstract

A ventilation fan management device and a learning device are provided that are advantageous in easily identifying the model names of existing ventilation fans in order to improve the efficiency of inspection work. [Solution] The ventilation fan management device 1 includes a data acquisition unit 21 that acquires image data of the exterior of the ventilation fan 4 and sound data indicating the sound emitted by the ventilation fan 4 when in operation, and an inference unit 10 that infers the model name of the ventilation fan 4 from the image data and sound data acquired by the data acquisition unit 21. The learning device 2 also includes the data acquisition unit 21 that acquires learning data including image data of the exterior of the ventilation fan 4, sound data indicating the sound emitted by the ventilation fan 4 when in operation, and model name information for the ventilation fan 4, and a model generation unit 22 that uses the learning data to generate a trained model for inferring the model name of the ventilation fan 4 from the image data and sound data of the ventilation fan 4.

Description

[Technical Field]

[0001] The present disclosure relates to a ventilation fan management system for inspecting ventilation fans. [Background technology]

[0002] In recent years, there has been a technology for improving the efficiency of inspections of equipment installed in buildings, such as ventilation fans and air conditioners, in which an operator photographs the equipment using imaging equipment that can communicate with an inspection system, and the inspection system analyzes the images of the equipment to obtain information about the equipment. For example, Patent Document 1 discloses an equipment management system in which an operator photographs an image including an identification mark of the equipment using an information terminal that has a camera and communicates with a management server, and the management server obtains equipment information from the photographed image by transmitting equipment data such as the model name and serial number to the information terminal, thereby improving the efficiency of inspection work. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-91530 Summary of the Invention [Problem to be solved by the invention]

[0004] When the equipment to be inspected is a ventilation fan, it is often installed high on the ceiling or wall of a building. In particular, in buildings such as factories and gymnasiums, the ceiling height is high and the ventilation fan is installed high. This makes it difficult for workers to see the nameplate or other location on the equipment where the model name can be confirmed. Furthermore, when trying to obtain the model name by analyzing an image of the ventilation fan's exterior, it is difficult to identify the model name from an external image because there are ventilation fans that have the same appearance but different performance.

[0005] In order to solve the above-mentioned problems, the present disclosure aims to provide a ventilation fan management device and a learning device that are advantageous in easily identifying the model names of existing ventilation fans in order to improve the efficiency of inspection work. [Means for solving the problem]

[0006] The ventilation fan management device of the present disclosure includes a data acquisition unit that acquires image data of the exterior of the ventilation fan and sound data indicating the sound emitted by the ventilation fan when in operation, and an inference unit that infers the model name of the ventilation fan from the image data and sound data acquired by the data acquisition unit.

[0007] In addition, the learning device according to the present disclosure includes a data acquisition unit that acquires learning data including image data of the exterior of the ventilation fan, sound data indicating the sound emitted by the ventilation fan when in operation, and model name information of the ventilation fan, and a model generation unit that uses the learning data to generate a trained model for inferring the model name of the ventilation fan from the image data and sound data of the ventilation fan. [Effects of the Invention]

[0008] According to the present disclosure, it is possible to provide a ventilation fan management device and a learning device that are advantageous in easily identifying the model name of an existing ventilation fan in order to improve the efficiency of inspection work. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a configuration diagram showing an example of a ventilation fan management system according to a first embodiment. [Figure 2] 10A and 10B are diagrams showing an example of the appearance of a ventilation fan 4. FIG. [Figure 3] FIG. 2 is a diagram illustrating an example of a neural network in the learning device according to the first embodiment. [Figure 4] 4 is a flowchart of a learning process in the learning device 2 according to the first embodiment. [Figure 5] FIG. 10 is a configuration diagram showing a modified example of the ventilation fan management system 100 according to the first embodiment. [Figure 6]6 is a flowchart showing a process for inferring model name information of a ventilation fan in the ventilation fan management device according to the first embodiment. [Figure 7] FIG. 10 is a configuration diagram showing an example of a ventilation fan management system according to a second embodiment. [Figure 8] 10 is a flowchart showing the process of inferring model name information of a ventilation fan and detecting an abnormality in the ventilation fan management device according to the second embodiment. [Figure 9] 10 is sound data showing the sound pressure level for each frequency emitted when the ventilation fan 4 is in operation. [Figure 10] 11 is a flowchart of a learning process in the learning device according to the third embodiment. [Figure 11] 11 is a flowchart showing a process for inferring the model name and abnormality details of a ventilation fan in the ventilation fan management device according to the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] The following description will discuss embodiments of the present disclosure with reference to the accompanying drawings. In each drawing, identical or corresponding parts are designated by the same reference numerals, and redundant descriptions are appropriately simplified or omitted. The present disclosure is not limited to the following embodiments, and any of the components of the embodiments may be modified, combined, or omitted without departing from the spirit of the present disclosure.

[0011] Embodiment 1 FIG. 1 is a configuration diagram showing an example of a ventilation fan management system 100 according to the present embodiment. The ventilation fan management system 100 will be described with reference to FIG. 1. The ventilation fan management system 100 has a ventilation fan management device 1 and an information terminal 3. The ventilation fan management device 1 and the information terminal 3 are communicatively connected via a network 5. The network 5 is a wide-area network such as the Internet. Furthermore, if the ventilation fan management device 1 and the information terminal 3 are located in a short distance, the network 5 may be a short-range communication path such as Bluetooth (registered trademark) or NFC (Near Field Communication). Furthermore, if the ventilation fan management device 1 and the information terminal 3 are located in a short distance, the network 5 may be a wired communication line connecting the ventilation fan management device 1 and the information terminal 3.

[0012] The information terminal 3 includes a camera unit 31 that captures still or video images, a sound recording unit 32 that records ambient sounds, and a communication unit 33 that communicates with the ventilation fan management device 1. The information terminal 3 is, for example, a computer such as a smartphone, a tablet, or a PC. Alternatively, the information terminal 3 may be a digital camera equipped with the sound recording unit 32. A manager who manages the ventilation fans 4 installed in a building uses the information terminal 3 to capture images of the exterior of the ventilation fans 4 and obtain image data. Hereinafter, image data with the exterior of the ventilation fans 4 as the subject will be referred to simply as image data. The dotted line connecting the ventilation fans 4 and the camera unit 31 in FIG. 1 indicates that the camera unit 31 captures an image of the ventilation fans 4. The manager also uses the information terminal 3 to record the sound emitted by the ventilation fans 4 when they are operating and obtain sound data. The sound data is data that indicates physical quantities related to sound, such as sound waveform data, data indicating sound pressure at each frequency, and the magnitude of the sound pressure. Hereinafter, data indicating the sound emitted by the ventilation fans 4 when they are operating will be referred to simply as sound data. 1, the dotted line connecting the ventilation fan 4 and the sound recording unit 32 indicates that the sound recording unit 32 records the sound emitted by the ventilation fan 4. The image data and sound data are transmitted to the ventilation fan management device 1 via the communication unit 33.

[0013] FIG. 2 is a diagram showing an example of the appearance of ventilation fan 4. In this embodiment, image data refers to image data obtained by capturing an image of the appearance of ventilation fan 4 shown in FIG. 2. When the blades 41 and motor 42 of the blower provided in ventilation fan 4 are exposed, as in FIG. 2(a), the model name can be inferred by extracting features such as the shape of blades 41, the number of blades 41, the shape of motor 42, the size of motor 42, and the structure of motor 42 from the image data of ventilation fan 4. FIG. 2(b) shows an example of ventilation fan 4 installed on a ceiling. When the structure of ventilation fan 4 is not exposed due to decorative grill 43 as in FIG. 2(b), the model name can be inferred by extracting features such as the shape of decorative grill 43 from the image data of ventilation fan 4.

[0014] Some ventilation fans 4 have very similar external shapes even if they have different model names. Furthermore, even if the appearance is exactly the same, ventilation fans 4 have different model names due to differences in the rotation speed of the blower equipped in the ventilation fan 4 or the structure of the air passage inside the ventilation fan 4. Therefore, it may be difficult to infer the model name from features that can be extracted from image data. The volume and frequency of the sound emitted by the ventilation fan 4 during operation change depending on the rotation speed of the blower, the shape of the blades 41, the structure of the air passage, etc. Therefore, the accuracy of inferring the model name of the ventilation fan 4 can be improved by acquiring sound data of the ventilation fan 4 and comparing it with sound data of ventilation fans 4 with known model names.

[0015] The ventilation fan management device 1 is composed of at least one processor and at least one memory. The processor is also called a CPU (Central Processing Unit), processing device, arithmetic device, microprocessor, microcomputer, or DSP. The memory is composed of, for example, non-volatile or volatile semiconductor memory such as RAM, ROM, flash memory, EPROM, or EEPROM. The ventilation fan management device 1 includes an inference unit 10, a ventilation fan database storage unit 11, a data acquisition unit 21, a model generation unit 22, and a trained model storage unit 23. The data acquisition unit 21, the model generation unit 22, and the trained model storage unit 23 function as a learning device 2.

[0016] The ventilation fan database storage unit 11 stores a ventilation fan database that associates model name information indicating the model names of multiple types of ventilation fans 4 with image data of the ventilation fans 4 and sound data when the ventilation fans 4 are operating. Model name information is information that uniquely indicates the model name of a ventilation fan, such as a character string indicating the model name or a number corresponding to the model name. The ventilation fan database has multiple image data and multiple sound data for each model name information of a ventilation fan 4. The ventilation fan database is created in advance by the administrator of the ventilation fan management device 1 for each ventilation fan 4 to be managed. Alternatively, the ventilation fan database may be created by having the data acquisition unit 21 (described later) acquire image data and sound data photographed and recorded using the information terminal 3 by the installer, administrator, or user of the ventilation fan 4, as well as model name information entered into the information terminal 3, and store the acquired data in the ventilation fan database storage unit 11.

[0017] Next, we will explain the configuration of the learning device 2 that learns the model name information of the ventilation fan 4. The data acquisition unit 21 acquires image data, sound data, and model name information of the ventilation fan 4 as learning data from the ventilation fan database storage unit 11. Alternatively, the data acquisition unit 21 may acquire the image data, sound data, and model name information of the ventilation fan 4 from the information terminal 3 via the network 5.

[0018] The model generation unit 22 learns the model name of the ventilation fan 4 based on learning data created based on a combination of image data of the ventilation fan 4 output from the data acquisition unit 21, sound data indicating the sound emitted by the ventilation fan 4 when in operation, and model name information of the ventilation fan 4. In other words, a trained model is generated that infers optimal model name information of the ventilation fan 4 from the image data, sound data, and model name information of the ventilation fan 4. Here, the learning data is data in which the image data, sound data, and model name information of the ventilation fan 4 are associated with each other.

[0019] The learning device 2 is used to learn the model name information of the ventilation fan 4, but may be, for example, connected to the ventilation fan management device 1 via a network and be a separate device from the ventilation fan management device 1. Furthermore, the learning device 2 may be located away from the ventilation fan management device 1 as long as it is connected to the ventilation fan management device 1 via a network 5.

[0020] The learning algorithm used by the model generation unit 22 may be a known algorithm such as supervised learning, unsupervised learning, reinforcement learning, etc. As an example, a case where a neural network is applied will be described.

[0021] The model generation unit 22 learns the model name information of the ventilation fan 4, for example, by so-called supervised learning in accordance with a neural network model. Here, supervised learning refers to a method in which a learning device is provided with pairs of input and result (label) data, and the device learns the features of the learning data and infers the result from the input.

[0022] A neural network consists of an input layer consisting of multiple neurons, an intermediate layer (hidden layer) consisting of multiple neurons, and an output layer consisting of multiple neurons. The intermediate layer may be one layer, or two or more layers.

[0023] Figure 3 is a diagram showing an example of a neural network in the learning device according to this embodiment. For example, in a three-layer neural network as shown in Figure 3, when multiple inputs are input to the input layer (X1-X3), the values ​​are multiplied by weight W1 (w11-w16) and input to the intermediate layer (Y1-Y2), and the result is further multiplied by weight W2 (w21-w26) and output from the output layer (Z1-Z3). This output result varies depending on the values ​​of weights W1 and W2.

[0024] In the present disclosure, the neural network learns the model name information of the ventilation fan 4 through so-called supervised learning in accordance with learning data created based on a combination of image data, sound data, and model name information of the ventilation fan 4 acquired by the data acquisition unit 21.

[0025] That is, the neural network learns by inputting image data and sound data of the ventilation fan 4 into the input layer and adjusting the weights W1 and W2 so that the results output from the output layer approach the model name information of the ventilation fan 4.

[0026] The model generation unit 22 generates and outputs a trained model by executing the above-described learning.

[0027] The trained model storage unit 23 stores the trained model output from the model generation unit 22.

[0028] Next, the learning process performed by the learning device 2 will be described with reference to Fig. 4. Fig. 4 is a flowchart of the learning process performed by the learning device 2 according to this embodiment. The timing of the learning process can be arbitrary, but it is desirable to perform the learning process when new image data, sound data, or model name information of the ventilation fan 4 is added to the ventilation fan database storage unit 11.

[0029] First, in step S101, the data acquisition unit 21 acquires image data, sound data, and model name information of the ventilation fan 4 from the ventilation fan database storage unit 11. Note that, although the image data, sound data, and model name information of the ventilation fan 4 are acquired simultaneously, it is sufficient that the image data, sound data, and model name information of the ventilation fan 4 are input in association with each other, and the image data, sound data, and model name information of the ventilation fan 4 may be acquired at different times. Next, the process proceeds to step S102.

[0030] In step S102, the model generation unit 22 learns the model name information of the ventilation fan 4 by so-called supervised learning in accordance with the learning data created based on a combination of the image data, sound data, and model name information of the ventilation fan 4 acquired by the data acquisition unit 21, and generates a learned model. Next, the process proceeds to step S103.

[0031] In step S103, the trained model storage unit 23 stores the trained model generated by the model generation unit 22. This completes the learning process of the model name of the ventilation fan 4 by the learning device 2.

[0032] Next, the inference unit 10 that infers the model name information of the ventilation fan 4 will be described. The inference unit 10 infers the model name information of the ventilation fan 4 from the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21. That is, the inference unit 10 can output the model name information of the ventilation fan 4 inferred from the image data and sound data of the ventilation fan 4 by inputting the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21 into a trained model. Note that if the model name information of the ventilation fan 4 inferred from the image data and sound data cannot be narrowed down to one, the inference unit 10 may output multiple model name information.

[0033] In the present embodiment, the inference unit 10 has been described as outputting the model name information of the ventilation fan 4 using a trained model trained by the model generation unit 22 of the ventilation fan management device 1. However, the inference unit 10 may acquire a trained model from an external source, such as another ventilation fan management device 1, and output the model name information of the ventilation fan 4 based on this trained model. FIG. 5 is a configuration diagram showing a modified example of the ventilation fan management system 100 according to the present embodiment. As shown in FIG. 5, the inference unit 10 may be installed in an information terminal 3. In this case, the inference unit 10 of the information terminal 3 acquires the trained model from the trained model storage unit 23 via the network 5 and the communication unit 33. The inference unit 10 of the information terminal 3 inputs image data of the ventilation fan 4 photographed by the photographing unit 31 and sound data of the ventilation fan 4 recorded by the recording unit 32 into the trained model acquired from the trained model storage unit 23, thereby outputting the model name information of the ventilation fan 4.

[0034] 6 is a flowchart showing the process of inferring the model name information of the ventilation fan 4 in the ventilation fan management device 1 according to this embodiment. The process of inferring the model name information of the ventilation fan 4 by the inference unit 10 will be described with reference to FIG.

[0035] First, in step S201, the manager of the ventilation fan 4 takes an image of the exterior of the ventilation fan 4 using the photographing unit 31 of the information terminal 3. Next, the process proceeds to step S202.

[0036] In step S202, the manager of the ventilation fan 4 uses the recording unit 32 of the information terminal 3 to record the sound emitted by the ventilation fan 4 when it is operating, and obtains sound data. Note that the processes of steps S201 and S202 may be performed simultaneously by the manager using the information terminal 3 to take a video of the ventilation fan 4 while the ventilation fan 4 is operating. Next, the process proceeds to step S203.

[0037] In step S203, the data acquisition unit 21 acquires image data and sound data of the ventilation fan 4 from the information terminal 3 via the network 5 and the communication unit 33. Next, the process proceeds to step S204.

[0038] In step S204, the inference unit 10 acquires the trained model from the trained model storage unit 23, and inputs the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21 into the trained model. Next, the process proceeds to step S205.

[0039] In step S205, the inference unit 10 obtains the model name information of the ventilation fan 4 output from the trained model. Next, the process proceeds to step S206.

[0040] In step S206, the inference unit 10 transmits the model name information of the ventilation fan 4 to the information terminal 3. This completes the process of inferring the model name information of the ventilation fan 4 by the inference unit 10.

[0041] The information terminal 3 that has received the model name information of the ventilation fan 4 can notify the manager of the model name information of the ventilation fan 4, for example, by displaying the model name information of the ventilation fan 4 on a display unit (not shown). By knowing the model name information of the ventilation fan 4, the manager can, for example, refer to the instruction manual for the corresponding model or smoothly inquire with a repair company if there is a problem with the ventilation fan 4. Note that the inference unit 10 may use the inferred model name information of the ventilation fan 4 for managing the ventilation fan 4 in the ventilation fan management device 1 without transmitting it to the information terminal 3. Managing the ventilation fan 4 refers to transmitting information necessary for the proper operation of the ventilation fan 4, such as storing the model name of the ventilation fan 4 and the manager and using it to notify the manager of inspection times and information about the successor model of the ventilation fan 4. Knowing the model name of the ventilation fan 4 in this way allows the manager to properly manage the ventilation fan 4.

[0042] When determining the model name information of a ventilation fan 4, existing ventilation fans 4 are often installed high on the ceiling or wall. In particular, in buildings such as factories and gymnasiums, the ceiling height is high and the ventilation fans 4 are installed high. This makes it difficult for workers to check the nameplate displaying the model name. Therefore, it was necessary to infer the model name from image data capturing the exterior of the ventilation fan 4. However, there are models of ventilation fans 4 that have very similar appearances, and different models that have the exact same appearance but different internal structures and fan performance. Therefore, it was not possible to infer the model name of a ventilation fan 4 from image data. In this embodiment, by learning the model name information of a ventilation fan 4 using image data and sound data, it is possible to obtain a trained model that can infer the model name information of a ventilation fan 4 from image data and sound data.

[0043] In the above explanation, the inference unit 10 infers the model name information of the ventilation fan 4 by generating a trained model, but the corresponding model name information can also be obtained from the ventilation fan database by comparing the image data and sound data with the ventilation fan database. Specifically, this is done in the following manner. The inference unit 10 extracts feature amounts from the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21. It compares the extracted feature amounts with the feature amounts of the image data and sound data contained in the ventilation fan database and outputs the model name information of the ventilation fan 4 that is most similar.

[0044] As described above, the ventilation fan management device 1 according to this embodiment includes a data acquisition unit 21 that acquires image data of the exterior of the ventilation fan 4 and sound data indicating the sound emitted by the ventilation fan 4 when it is operating, and an inference unit 10 that infers the model name of the ventilation fan 4 from the image data and sound data acquired by the data acquisition unit 21. In this way, by using the image data and sound data of the ventilation fan 4 to infer model name information, it is possible to determine model name information with greater accuracy than when only image data is used.

[0045] The inference unit 10 is also characterized in that it outputs the model name of the ventilation fan 4 by inputting the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21 into a trained model that infers the model name of the ventilation fan 4 from the image data and sound data of the ventilation fan 4. In this way, the inference unit 10 can infer the model name information with high accuracy by inferring the model name information using a trained model generated by an existing learning algorithm.

[0046] Furthermore, the learning device 2 according to this embodiment includes a data acquisition unit 21 that acquires learning data including image data of the exterior of the ventilation fan 4, sound data indicating sounds emitted by the ventilation fan 4 when in operation, and model name information of the ventilation fan 4, and a model generation unit 22 that uses the learning data to generate a trained model for inferring the model name of the ventilation fan 4 from the image data and sound data of the ventilation fan 4. In this way, by learning the model name information using the image data and sound data, it is possible to generate a trained model that infers the model name information from the image data and sound data.

[0047] The ventilation fan management device 1 may further include a ventilation fan database storage unit 11 that stores a ventilation fan database that associates model name information of multiple types of ventilation fans 4 with image data and sound data of the ventilation fans 4, and the inference unit 10 may output the model name of the ventilation fan from the ventilation fan database by comparing the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21 with the ventilation fan database. This makes it possible to infer model name information from the image data and sound data without having to learn the model name information using the image data and sound data.

[0048] Embodiment 2 Next, a second embodiment will be described. FIG. 7 is a configuration diagram showing an example of a ventilation fan management system 100 according to the second embodiment. In the configuration of the ventilation fan management system 100 of this embodiment shown in FIG. 7, components similar to those of the first embodiment are assigned the same reference numerals and will not be described again. In the second embodiment, the ventilation fan management device 1 processes the sound data acquired by the data acquisition unit 21 into adjustment sound data corresponding to sound data at a predetermined measurement reference distance. Furthermore, the ventilation fan management device 1 compares the adjustment sound data with reference sound data corresponding to the model name information of the ventilation fan 4 inferred by the inference unit 10, and detects an abnormality. The reference sound data is sound data recorded at a location a predetermined measurement reference distance away when the ventilation fan 4 is operating normally. Hereinafter, when there is no need to distinguish between the adjustment sound data and the reference sound data, they will simply be referred to as sound data.

[0049] 7, the information terminal 3 according to the present embodiment further includes a distance acquisition unit 34 and a notification unit 35 in addition to the configuration of the information terminal 3 according to the first embodiment. Also, the ventilation fan management device 1 according to the present embodiment includes a sound processing unit 12, an abnormality detection unit 13, and a reference sound data storage unit 14 in addition to the configuration of the ventilation fan management device 1 according to the first embodiment.

[0050] The distance acquisition unit 34 acquires a recording distance, which is information indicating the distance between the ventilation fan 4 and the information terminal 3 when sound data from the ventilation fan 4 is recorded using the information terminal 3. The distance acquisition unit 34 detects surrounding objects using, for example, millimeter-wave radar using radio waves or LiDAR (Light Detection and Ranging) using light, and calculates the distance to the object. Note that the distance acquisition unit 34 is not limited to the above as long as it can acquire the distance between the information terminal 3 and the ventilation fan 4, which is the object. For example, the information terminal 3 may be equipped with multiple image capture units 31, and the distance acquisition unit 34 may calculate the distance to the object by comparing images captured from different viewpoints. Alternatively, the administrator of the ventilation fan 4 may input the distance into an operation unit (not shown) of the information terminal 3 to acquire the recording distance. The dotted line connecting the ventilation fan 4 and the distance acquisition unit 34 in FIG. 7 indicates that the distance acquisition unit 34 acquires information indicating the distance from the ventilation fan 4. The information terminal 3 transmits the recording distance acquired by the distance acquisition unit 34 together with the image data and sound data to the data acquisition unit 21 of the ventilation fan management device 1 via the communication unit 33.

[0051] When the anomaly detection unit 13, which will be described later, notifies the administrator operating the information terminal 3 that there is an abnormality in the ventilation fan 4, the anomaly notification unit 35 notifies the administrator that there is an abnormality in the ventilation fan 4. The notification is made by voice or by displaying on a screen, for example. This allows the administrator to know that there is an abnormality in the ventilation fan 4.

[0052] The sound processing unit 12 uses the sound data acquired by the data acquisition unit 21 and the recording distance to process the sound data into adjusted sound data equivalent to the sound data when measured at a predetermined measurement reference distance. Note that the measurement reference distance is assumed to be stored in the memory (not shown) of the ventilation fan management device 1. The sound emitted by the ventilation fan 4 when it is operating can be considered as a point sound source, and it is known that the sound attenuates by approximately 6 dB when the distance is doubled. In other words, if the adjusted sound data is SPL2, the sound data is SPL1, the recording distance is r1, and the measurement reference distance is r2, the adjusted sound data SPL2 can be calculated using equation (1).

[0053]

number

[0054] However, in reality, sound does not attenuate as much with distance as in equation (1) because it is reflected by walls and floors. Indoors, the amount of sound attenuation with distance can be considered to be about half of that in equation (1). Therefore, the adjustment sound data SPL2 may be calculated using equation (2). Note that the method of calculating the adjustment sound data from sound data is not limited to equation (1) or equation (2), and it may also be calculated taking into account the influence of sound reflection or sounds other than those emitted by the ventilation fan 4 depending on the environment in which the ventilation fan 4 is installed.

[0055]

number

[0056] In this embodiment, the inference unit 10 infers model name information of the ventilation fan 4 from the image data of the ventilation fan 4 acquired by the data acquisition unit 21 and the adjusted sound data processed by the sound processing unit 12. That is, the inference unit 10 inputs the image data of the ventilation fan 4 acquired by the data acquisition unit 21 and the adjusted sound data processed by the sound processing unit 12 to the trained model, and thereby can output the model name information of the ventilation fan 4 inferred from the image data of the ventilation fan 4 and the adjusted sound data.

[0057] Note that the generation of the trained model by the model generation unit 22 is performed in the same manner as in embodiment 1, and therefore a description thereof will be omitted. However, in the combination of image data of the ventilation fan 4, which is the training data used by the model generation unit 22, sound data indicating the sound emitted by the ventilation fan 4 when in operation, and model name information of the ventilation fan 4, it is desirable to use sound data recorded at the measurement reference distance or adjusted sound data equivalent to the sound data measured at the measurement reference distance for the sound data. This results in consistent training data, and a highly accurate trained model can be generated even with a small amount of training data. Note that when sound data recorded at the measurement reference distance is used as training data, it is desirable to use sound data recorded at the measurement reference distance or adjusted sound data as sound data input to the trained model in inferring model name information.

[0058] The reference sound data storage unit 14 stores reference sound data, which is sound data obtained by recording sounds emitted during operation of multiple types of ventilation fans 4 at a measurement reference distance. The reference sound data is stored in association with model name information of the ventilation fans 4.

[0059] The abnormality detection unit 13 compares the adjustment sound data processed by the sound processing unit 12 with the reference sound data corresponding to the model name information of the ventilation fan 4 inferred by the inference unit 10 to detect an abnormality. The ventilation fan 4 emits abnormal sounds when it is operated in an abnormal state, such as a lack of grease used to lubricate the bearings, wear or damage to parts due to aging, or application of harmonics to the power supply. Abnormal sounds have higher sound pressure at specific frequencies compared to normal sounds. Therefore, the abnormality detection unit 13 can detect the presence or absence of an abnormality in the ventilation fan 4 by comparing the adjustment sound data with the reference sound data. The method for comparing the adjustment sound data with the reference sound data varies depending on the type of sound data. For example, if the sound data is a value indicating the magnitude of sound pressure, the difference between the adjustment sound data and the reference sound data is calculated, and if the difference is greater than or equal to a threshold, an abnormality is determined. Furthermore, if the sound data is frequency data, the difference between the adjustment sound data and the reference sound data is calculated for each frequency, and if there is a frequency where the difference is greater than or equal to a threshold, an abnormality is determined. Furthermore, if the sound data is sound waveform data, the difference in amplitude between the adjustment sound data and the reference sound data is calculated, and if the difference is equal to or greater than a threshold, it is determined that an abnormality has occurred. When the abnormality detection unit 13 detects an abnormality, it notifies the notification unit 35 of the information terminal 3 of the abnormality via the network 5 and the communication unit 33.

[0060] Fig. 8 is a flowchart showing the process of inferring model name information of a ventilation fan 4 and detecting an abnormality in the ventilation fan management device 1 according to this embodiment. The flow of inferring model name information of a ventilation fan 4 and detecting an abnormality in the ventilation fan management device 1 will be described using Fig. 8. Note that in Fig. 8, the same processes as those inferring model name information of a ventilation fan 4 in the ventilation fan management device 1 according to embodiment 1 are assigned the same reference numerals as in Fig. 6, and the description will be simplified.

[0061] First, in step S201, the manager of the ventilation fan 4 takes an image of the exterior of the ventilation fan 4 using the photographing unit 31 of the information terminal 3. Next, the process proceeds to step S301.

[0062] In step S301, the manager of the ventilation fan 4 uses the recording unit 32 of the information terminal 3 to record the sound emitted when the ventilation fan 4 is operating, and obtains sound data. At this time, the distance acquisition unit 34 acquires a recording distance, which is information indicating the distance between the information terminal 3 and the ventilation fan 4. Note that the processes of steps S201 and S301 may be performed simultaneously by the manager using the information terminal 3 to take a video of the ventilation fan 4 while the ventilation fan 4 is operating. Next, proceed to step S302.

[0063] In step S302, the data acquisition unit 21 acquires image data, sound data, and recording distance of the ventilation fan 4 from the information terminal 3 via the network 5 and the communication unit 33. Next, the process proceeds to step S303.

[0064] In step S303, the sound processing unit 12 processes the sound data of the ventilation fan 4 acquired by the data acquisition unit 21 according to the recording distance and the measurement reference distance to obtain adjusted sound data. The adjusted sound data is obtained, for example, by applying Equation (1) or Equation (2) to the sound data using the recording distance and the measurement reference distance. Next, the process proceeds to step S304.

[0065] In step S304, the inference unit 10 acquires the trained model from the trained model storage unit 23, and inputs into the trained model the image data of the ventilation fan 4 acquired by the data acquisition unit 21 and the adjusted sound data processed by the sound processing unit 12. Next, the process proceeds to step S205.

[0066] In step S205, the inference unit 10 obtains the model name information of the ventilation fan 4 output from the trained model. Next, the process proceeds to step S305.

[0067] In step S305, the abnormality detection unit 13 reads out the reference sound data corresponding to the model name information of the ventilation fan 4 output by the inference unit 10 from the reference sound data storage unit 14. Next, the process proceeds to step S306.

[0068] In step S306, the abnormality detection unit 13 compares the adjustment sound data with the reference sound data, and then the process proceeds to step S307.

[0069] In step S307, if the abnormality detection unit 13 determines that the difference between the adjusted sound data and the reference sound data is equal to or greater than the threshold (step S307: Yes), the process proceeds to step S308. If the abnormality detection unit 13 determines that the difference between the adjusted sound data and the reference sound data is not equal to or greater than the threshold (step S307: No), the process of inferring the model name information of the ventilation fan 4 and detecting an abnormality in the ventilation fan management device 1 ends.

[0070] In step S307, the abnormality detection unit 13 notifies the notification unit 35 of the information terminal 3 of the abnormality. The notification unit 35 of the information terminal 3 notifies the administrator of the abnormality by voice or by displaying on a display unit (not shown). This completes the processing of the flow of inferring the model name information of the ventilation fan 4 and detecting an abnormality in the ventilation fan management device 1.

[0071] In this embodiment, anomaly detection is performed using the inferred model name information. If an anomaly is detected, the information terminal 3 is notified in step S307, so the process of notifying the information terminal 3 of the model name information in step S205 may be omitted.

[0072] Furthermore, in step S307, if it is determined that the difference between the adjustment sound data and the reference sound data is not equal to or greater than the threshold value (step S307: No), the abnormality detection unit 13 may notify the notification unit 35 of the information terminal 3 that there is no abnormality in the ventilation fan 4. The notification unit 35 of the information terminal 3 notifies the administrator that there is no abnormality in the ventilation fan 4 by audio notification or by displaying on a display unit (not shown). This allows the administrator to confirm that the ventilation fan 4 is normal.

[0073] 8, the flow of inferring the model name information of the ventilation fan 4 and detecting an anomaly is shown. The sound processing unit 12 processes the sound data acquired by the data acquisition unit 21 according to the recording distance and the measurement reference distance to obtain adjusted sound data. The inference unit 10 then infers the model name of the ventilation fan 4 using the adjusted sound data. When the model generation unit 22 generates a trained model by learning using sound data recorded at the measurement reference distance as learning data, using image data and adjusted sound data as input to the trained model in this way enables more accurate inference of the model name information of the ventilation fan 4 than when the recorded sound data is input without processing. However, the inference unit 10 may input the image data and the recorded sound data to the trained model to infer the model name information of the ventilation fan 4.

[0074] Furthermore, inference unit 10 according to this embodiment may infer the corresponding model name information from the ventilation fan database by comparing the image data and the adjustment sound data with the ventilation fan database when inferring the model name information of ventilation fan 4. In this case, by using the sound data included in the ventilation fan database as sound data recorded at the measurement reference distance, it is possible to infer the model name information of ventilation fan 4 with higher accuracy than when the recorded sound data is compared with the ventilation fan database without processing it.

[0075] As described above, the data acquisition unit 21 of the ventilation fan management device 1 according to this embodiment is characterized by further comprising a sound processing unit 12 that acquires a recording distance, which is information indicating the distance between the ventilation fan 4 and the location where the sound data was recorded, processes the sound data of the ventilation fan 4 according to the recording distance, and obtains adjusted sound data equivalent to the sound data when measured at a predetermined measurement reference distance. With this configuration, it is possible to infer model name information of the ventilation fan 4 with higher accuracy than when the recorded sound data is input without processing.

[0076] Furthermore, the ventilation fan management device 1 according to this embodiment further includes a reference sound data storage unit 14 that stores reference sound data, which is sound data obtained by measuring sounds emitted by multiple ventilation fans 4 during operation at a measurement reference distance, in association with the model names of the corresponding ventilation fans 4, and an abnormality detection unit 13 that determines that there is an abnormality in the ventilation fan if the difference between the reference sound data associated with the model name of the ventilation fan 4 output by the inference unit 10 and the adjustment sound data is equal to or greater than a threshold. With this configuration, the manager can know whether or not there is an abnormality in the ventilation fan 4 by measuring the image data, sound data, and recording distance.

[0077] Embodiment 3 Next, a third embodiment will be described. The configuration of the ventilation fan management system 100 according to the third embodiment is the same as that of the first embodiment, and therefore a description thereof will be omitted. In the third embodiment, the learning device 2 of the ventilation fan management device 1 learns the model name information of the ventilation fan 4 and the details of an abnormality in the ventilation fan 4. Furthermore, the inference unit 10 infers the model name information of the ventilation fan 4 and the details of an abnormality in the ventilation fan 4 using the image data and sound data acquired by the data acquisition unit 21.

[0078] In this embodiment, the ventilation fan database storage unit 11 stores a ventilation fan database that associates model name information indicating the model names of multiple types of ventilation fans 4, image data of the ventilation fans 4, sound data when the ventilation fans 4 are operating, and abnormality details indicating abnormalities occurring in the ventilation fans 4 when the image data and sound data were measured. Examples of abnormality details include a lack of grease used to lubricate the bearings, wear of parts due to aging, and application of harmonics to the power supply. Furthermore, if no abnormality has occurred, the abnormality details are set to a value indicating normality.

[0079] FIG. 9 shows sound data indicating the sound pressure level for each frequency emitted when the ventilation fan 4 is operating. In FIG. 9, (a) shows sound data during normal operation, and (b) shows sound data when an abnormality occurs in the ventilation fan 4. The sound pressure level at the frequencies indicated by the dashed circle in (b) is higher than that in (a). This is because in (b), the grease applied to the bearings of the blower of the ventilation fan 4 has run out, causing the motor to vibrate and resonate. In this way, the nature of the abnormality can be inferred from the sound data. Furthermore, there are cases where the nature of the abnormality can be determined from the appearance, such as damage to the fan blades 41. Therefore, the inference unit 10 in this embodiment can accurately infer the nature of the abnormality by using image data and sound data.

[0080] 10 is a flowchart of the learning process in the learning device 2 according to this embodiment. The flow of learning the model name information of the ventilation fan 4 and the details of an abnormality in the ventilation fan 4 in the learning device 2 will be described with reference to FIG.

[0081] First, in step S401, the data acquisition unit 21 acquires image data, sound data, model name information, and abnormality details of the ventilation fan 4. While the image data, sound data, model name information, and abnormality details of the ventilation fan 4 are acquired simultaneously, it is sufficient that the image data, sound data, model name information, and abnormality details of the ventilation fan 4 are input in association with each other, and the image data, sound data, model name information, and abnormality details of the ventilation fan 4 may be acquired at different times. The data acquisition unit 21 acquires the image data, sound data, model name information, and abnormality details of the ventilation fan 4 from the ventilation fan database storage unit 11. Alternatively, the data acquisition unit 21 may acquire the image data, sound data, model name information, and abnormality details of the ventilation fan 4 from the information terminal 3 via the network 5. Next, the process proceeds to step S402.

[0082] In step S402, the model generation unit 22 learns the model name information and the abnormality details of the ventilation fan 4 by so-called supervised learning in accordance with the learning data created based on a combination of the image data, sound data, model name information, and abnormality details of the ventilation fan 4 acquired by the data acquisition unit 21, and generates a learned model. Next, the process proceeds to step S403.

[0083] In step S403, the trained model storage unit 23 stores the trained model generated by the model generation unit 22. This completes the learning process of the learning device 2 regarding the model name of the ventilation fan 4 and the abnormality details of the ventilation fan 4.

[0084] FIG. 11 is a flowchart showing a process for inferring the model name and abnormality details of the ventilation fan 4 in the ventilation fan management device 1 according to the present embodiment. The process for inferring the model name information and abnormality details of the ventilation fan 4 in the inference unit 10 will be described with reference to FIG. 11. The inference unit 10 according to the present embodiment infers the model name information and abnormality details of the ventilation fan 4 from the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21. That is, the inference unit 10 inputs the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21 into a trained model, thereby outputting the model name information and abnormality details of the ventilation fan 4 inferred from the image data and sound data of the ventilation fan 4. Note that in FIG. 11, processes similar to the process for inferring the model name of the ventilation fan in the first embodiment are assigned the same reference numerals as in FIG. 6, and the description thereof will be simplified.

[0085] First, in step S201, the manager of the ventilation fan 4 takes an image of the exterior of the ventilation fan 4 using the photographing unit 31 of the information terminal 3. Next, the process proceeds to step S202.

[0086] In step S202, the manager of the ventilation fan 4 uses the recording unit 32 of the information terminal 3 to record the sound emitted by the ventilation fan 4 when it is operating, and obtains sound data. Note that the processes of steps S201 and S202 may be performed simultaneously by the manager using the information terminal 3 to take a video of the ventilation fan 4 while the ventilation fan 4 is operating. Next, the process proceeds to step S203.

[0087] In step S203, the data acquisition unit 21 acquires image data and sound data of the ventilation fan 4 from the information terminal 3 via the network 5 and the communication unit 33. Next, the process proceeds to step S204.

[0088] In step S204, the inference unit 10 acquires the trained model from the trained model storage unit 23, and inputs into the trained model the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21. Next, the process proceeds to step S501.

[0089] In step S501, the inference unit 10 obtains the model name information and abnormality details of the ventilation fan 4 output from the trained model. Next, the process proceeds to step S502.

[0090] In step S206, the inference unit 10 transmits the model name information and the abnormality details of the ventilation fan 4 to the information terminal 3. This completes the process of the inference unit 10 for inferring the model name information and the abnormality details of the ventilation fan 4.

[0091] The information terminal 3 that has received the model name information and abnormality details of the ventilation fan 4 can notify the manager of the model name information and abnormality details of the ventilation fan 4, for example, by displaying the model name information and abnormality details on a display unit (not shown). By knowing the model name information and abnormality details of the ventilation fan 4, the manager can, for example, inform a repair company of the model name information and abnormality details when requesting repair of the ventilation fan 4, which has the effect of allowing the manager to smoothly make an inquiry to the repair company. Note that the inference unit 10 may not transmit the inferred model name information of the ventilation fan 4 to the information terminal 3, and the ventilation fan management device 1 may notify the repair company of the abnormality details of the ventilation fan 4.

[0092] In the process of inferring the model name and abnormality details of the ventilation fan 4 in FIG. 11 , image data and sound data of the ventilation fan 4 are input to the trained model. However, image data and adjusted sound data of the ventilation fan 4 may also be input to the trained model. In this case, as described in the second embodiment, the data acquisition unit 21 acquires the image data, sound data, and recording distance, and the sound processing unit 12 processes the sound data into adjusted sound data using the recording distance. Furthermore, when inputting image data and adjusted sound data of the ventilation fan 4 to the trained model, it is desirable to also use the adjusted sound data as training data for generating the trained model by the model generation unit 22. In this way, using the adjusted sound data to learn the model name information and abnormality details results in consistent training data, making it possible to generate a highly accurate trained model even with a small amount of training data. Furthermore, by inputting image data and adjusted sound data to the trained model by the inference unit 10, it is possible to infer the model name information and abnormality details of the ventilation fan 4 with higher accuracy than when inputting recorded sound data without processing it.

[0093] As described above, the inference unit 10 of the ventilation fan management device 1 according to this embodiment is characterized in that it outputs the model name and abnormality details of the ventilation fan 4 from the image data and sound data of the ventilation fan 4 acquired by the data acquisition unit 21. With this configuration, it is possible to infer the model name information and the abnormality details from the image data and sound data, and thus the ventilation fan 4 can be managed efficiently.

[0094] Furthermore, in the learning device 2 according to this embodiment, the data acquisition unit 21, when the sound data indicates the sound emitted during operation if there is an abnormality in the ventilation fan 4, further acquires the abnormality details of the ventilation fan 4 as learning data, and includes a model generation unit 22 that uses the learning data to generate a trained model for inferring the model name and abnormality details of the ventilation fan 4 from the image data and sound data of the ventilation fan. This configuration makes it possible to generate a trained model that infers the abnormality details as well as model name information from the image data and sound data, which is advantageous for managing the ventilation fan 4.

[0095] Furthermore, as a modified example of the ventilation fan management device 1 according to this embodiment, the inference unit 10 outputs the model name and abnormality details of the ventilation fan 4 from the image data and adjustment sound data of the ventilation fan 4 acquired by the data acquisition unit 21. With this configuration, it is possible to infer the model name information and abnormality details of the ventilation fan 4 with higher accuracy than when recorded sound data is input without processing.

[0096] Furthermore, as a variation of the learning device 2 according to this embodiment, the sound data acquired by the data acquisition unit 21 is characterized in that it is adjusted sound data equivalent to sound data measured at a predetermined measurement reference distance, obtained by processing sound data indicating sounds emitted during operation when there is an abnormality in the ventilation fan 4. In this way, by using the adjusted sound data for learning, consistent learning data is obtained, and even with a small amount of learning data, a highly accurate learned model can be generated.

[0097] The configurations shown in the above embodiments are examples of the content of the present invention, and may be combined with other known technologies, and some of the configurations may be omitted or modified within the scope of the gist of the present invention.

[0098] Examples of aspects that may be included in the present disclosure are set forth below as appendices. (Appendix 1) a data acquisition unit that acquires image data of an external appearance of the ventilation fan and sound data indicating a sound emitted by the ventilation fan during operation; an inference unit that infers a model name of the ventilation fan from the image data and the sound data acquired by the data acquisition unit; A ventilation fan management device comprising: (Appendix 2) The ventilation fan management device described in Appendix 1 is characterized in that the inference unit outputs the model name of the ventilation fan by inputting the image data and sound data of the ventilation fan acquired by the data acquisition unit into a trained model that infers the model name of the ventilation fan from the image data and sound data of the ventilation fan. (Appendix 3) a data acquisition unit that acquires learning data including image data of an external appearance of the ventilation fan, sound data indicating a sound emitted by the ventilation fan during operation, and model name information of the ventilation fan; a model generation unit that generates a trained model for inferring a model name of the ventilation fan from the image data and the sound data of the ventilation fan using the learning data; A learning device comprising: (Appendix 4) Further provided is a ventilation fan database storage unit that stores a ventilation fan database in which model name information of a plurality of types of the ventilation fans is associated with the image data and the sound data of the ventilation fans, The ventilation fan management device described in Appendix 1 is characterized in that the inference unit outputs the model name of the ventilation fan from the ventilation fan database by comparing the image data and sound data of the ventilation fan acquired by the data acquisition unit with the ventilation fan database. (Appendix 5) The data acquisition unit further acquires a recording distance, which is information indicating the distance between the ventilation fan and a location where the sound data was recorded, The ventilation fan management device described in any one of appendix 1, 2 or 4, further comprises a sound processing unit that processes the sound data of the ventilation fan according to the recording distance and obtains adjusted sound data equivalent to the sound data when measured at a predetermined measurement standard distance. (Appendix 6) a reference sound data storage unit that stores reference sound data, which is sound data obtained by measuring sounds emitted by a plurality of types of ventilation fans during operation at the measurement reference distance, in association with the model names of the corresponding ventilation fans; an abnormality detection unit that determines that there is an abnormality in the ventilation fan when a difference between the reference sound data associated with the model name of the ventilation fan output by the inference unit and the adjustment sound data is equal to or greater than a threshold; 6. The ventilation fan management device according to claim 5, further comprising: (Appendix 7) The ventilation fan management device described in Appendix 1 is characterized in that the inference unit outputs the model name and abnormality details of the ventilation fan from the image data and sound data of the ventilation fan acquired by the data acquisition unit. (Appendix 8) When the sound data is data indicating a sound emitted during operation when there is an abnormality in the ventilation fan, the data acquisition unit further acquires details of the abnormality in the ventilation fan as the learning data, A model generation unit that generates a trained model for inferring a model name and an abnormality content of the ventilation fan from the image data and the sound data of the ventilation fan using the learning data; 4. The learning device according to claim 3, comprising: (Appendix 9) The ventilation fan management device described in Appendix 5 is characterized in that the inference unit outputs the model name and abnormality details of the ventilation fan from the image data and adjustment sound data of the ventilation fan acquired by the data acquisition unit. (Appendix 10) The learning device described in Appendix 8 is characterized in that the sound data acquired by the data acquisition unit is adjusted sound data equivalent to the sound data when measured at a predetermined measurement reference distance, obtained by processing the sound data indicating the sound emitted during operation when there is an abnormality in the ventilation fan. [Explanation of symbols]

[0099] 1 Ventilation fan management device, 2 Learning device, 3 Information terminal, 4 Ventilation fan, 5 Network, 10 Inference unit, 11 Ventilation fan database storage unit, 12 Sound processing unit, 13 Anomaly detection unit, 14 Reference sound data storage unit, 21 Data acquisition unit, 22 Model generation unit, 23 Trained model storage unit, 31 Photography unit, 32 Sound recording unit, 33 Communication unit, 34 Distance acquisition unit, 35 Alarm unit, 41 Blade, 42 Motor, 43 Decorative grill, 100 Ventilation fan management system.

Claims

1. a data acquisition unit that acquires image data of an external appearance of the ventilation fan and sound data indicating a sound emitted by the ventilation fan during operation; an inference unit that infers a model name of the ventilation fan from the image data and the sound data acquired by the data acquisition unit; A ventilation fan management device comprising:

2. The ventilation fan management device described in claim 1, characterized in that the inference unit outputs the model name of the ventilation fan by inputting the image data and sound data of the ventilation fan acquired by the data acquisition unit into a learned model that infers the model name of the ventilation fan from the image data and sound data of the ventilation fan.

3. a data acquisition unit that acquires learning data including image data of an external appearance of the ventilation fan, sound data indicating a sound emitted by the ventilation fan during operation, and model name information of the ventilation fan; a model generation unit that generates a trained model for inferring a model name of the ventilation fan from the image data and the sound data of the ventilation fan using the learning data; A learning device comprising:

4. Further provided is a ventilation fan database storage unit that stores a ventilation fan database in which model name information of a plurality of types of the ventilation fans is associated with the image data and the sound data of the ventilation fans, The ventilation fan management device described in claim 1, characterized in that the inference unit outputs the model name of the ventilation fan from the ventilation fan database by comparing the image data and sound data of the ventilation fan acquired by the data acquisition unit with the ventilation fan database.

5. The data acquisition unit further acquires a recording distance, which is information indicating the distance between the ventilation fan and a location where the sound data was recorded, The ventilation fan management device described in any one of claims 1, 2 or 4, further characterized in that it is provided with a sound processing unit that processes the sound data of the ventilation fan according to the recording distance and obtains adjusted sound data equivalent to the sound data when measured at a predetermined measurement standard distance.

6. a reference sound data storage unit that stores reference sound data, which is sound data obtained by measuring sounds emitted by a plurality of types of ventilation fans during operation at the measurement reference distance, in association with the model names of the corresponding ventilation fans; an abnormality detection unit that determines that there is an abnormality in the ventilation fan when a difference between the reference sound data associated with the model name of the ventilation fan output by the inference unit and the adjustment sound data is equal to or greater than a threshold; The ventilation fan management device according to claim 5 , further comprising:

7. The ventilation fan management device described in claim 1, characterized in that the inference unit outputs the model name and abnormality details of the ventilation fan from the image data and sound data of the ventilation fan acquired by the data acquisition unit.

8. When the sound data is data indicating a sound emitted during operation when there is an abnormality in the ventilation fan, the data acquisition unit further acquires details of the abnormality in the ventilation fan as the learning data, A model generation unit that generates a trained model for inferring a model name and an abnormality content of the ventilation fan from the image data and the sound data of the ventilation fan using the learning data; The learning device according to claim 3 , comprising:

9. The ventilation fan management device described in claim 5, characterized in that the inference unit outputs the model name and abnormality details of the ventilation fan from the image data and adjustment sound data of the ventilation fan acquired by the data acquisition unit.

10. The learning device described in claim 8, characterized in that the sound data acquired by the data acquisition unit is adjusted sound data equivalent to the sound data when measured at a predetermined measurement reference distance, obtained by processing the sound data indicating the sound emitted during operation when there is an abnormality in the ventilation fan.

Images