Monitoring system, learning device, monitoring method, learning method and program

Abstract

To provide a monitoring system, a learning device, a monitoring method, a learning method and a program which can monitor the inner state of a solid-liquid separation tank for separating wastewater into solid and liquid.SOLUTION: A monitoring system comprises: a determination unit which determines the state inside a solid-liquid separation tank from a monitoring image showing the inside of the solid-liquid separation tank which is an object of diagnosis, by using a first learning model which has learned the relation between the monitoring image and the state inside the solid-liquid separation tank, on the basis of the monitoring image which is an image showing the inside of the solid-liquid separation tank for separating wastewater into solid and liquid and a diagnosis result based on the monitoring image inside the solid-liquid separation tank; and an output unit which outputs information identifying the state inside the solid-liquid separation tank determined by the determination unit by using the monitoring image of the solid-liquid separation tank which is the object of diagnosis and the first learning model.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a monitoring system, a learning device, a monitoring method, a learning method, and a program. [Background technology]

[0002] In wastewater treatment, aerobic biological treatment (activated sludge) with an aeration tank and a settling tank is a mature treatment form, and the equipment is almost complete in terms of engineering. However, wastewater changes due to changes in products, production processes, and production volume, and inflow conditions (substrate, concentration, flow rate) often differ from when the equipment was designed based on the assumed water quality and water volume. In addition, with the shift to multi-product manufacturing, there are more and more cases of large fluctuations, making plant maintenance and management more complex and difficult. In addition, while there is a demand for more stable treatment than ever before in response to the review of water quality environmental standards, there is also a demand for reducing treatment costs, that is, reducing electricity costs, waste costs, and chemical usage. For this reason, advanced management is required to accurately and quickly grasp the treatment status and make appropriate adjustments according to the status, but such measurement and monitoring is not easy.

[0003] Regarding technology for monitoring the state of processing, a technology is known in which an ultrasonic pulse is emitted vertically from the liquid surface, and based on the result of receiving the reflected pulse, the position (interface depth) of the interface formed by suspended turbidity in the liquid is measured (see, for example, Patent Document 1). There is also known a technique for monitoring the state inside a tank such as a solid-liquid separation tank (see, for example, Patent Document 2). This technique comprises an A / D converter that transmits ultrasonic waves or light, receives the ultrasonic waves or light propagated through water containing a suspended matter sediment layer, and converts the received signal into a digital signal by a sensor, a calculation unit that calculates the position of the interface inside the tank based on the digital signal, a graphic conversion unit that converts the digital signal into pixel data, a memory that stores the pixel data and interface position data, and a display unit that displays the pixel data stored in the memory. Also, an interface level meter capable of switching between and displaying image information of multiple time widths quickly and stably is known (see, for example, Patent Document 3). This interface level meter includes an ultrasonic sensor, an A / D converter that converts a signal received by the ultrasonic sensor into a digital signal, a calculation unit that calculates the position of the interface between the suspended matter sediment layer and supernatant water based on the digital signal, a graphic conversion unit that converts the digital signal into pixel data corresponding to a predetermined color gradation, a memory having a storage area for acquiring and storing pixel string data including multiple pixel data at different time intervals, and a display unit having a display area that displays the multiple pixel string data stored in one of the storage areas based on a color gradation, and a display area that displays the position of the interface calculated by the calculation unit.

[0004] There is also known a technique for measuring the sedimentation state (see, for example, Patent Document 4). This technique includes an ultrasonic transmission / reception means for transmitting and receiving ultrasonic waves vertically downward from the water surface of the sedimentation tank, and a waveform processing means for processing the reflected and received waves obtained by the ultrasonic transmission / reception means. The waveform processing means measures the amount of material suspended below the water surface and / or the concentration distribution of sediments based on the change in intensity of the reflected and received waves. There is also known a technique for detecting the interface between layers in a sludge sedimentary layer (see, for example, Patent Document 5). This technique uses a sensor that transmits ultrasonic waves or light in the liquid in a solid-liquid separation tank and receives the ultrasonic waves or light propagated through the water containing the sludge sedimentary layer, and detects the position of the interface between the sludge sedimentary layer and the supernatant water based on the signal from the sensor, as well as the interface between the free settling layer, which occupies the top layer in the sludge sedimentary layer, and the flocculating settling layer below it. In this technique, the band in the top of the sludge sedimentary layer where the distribution of the received signal strength of the sensor in the depth direction of the tank is constant is defined as the free settling layer, and the position where the distribution of the received signal strength begins to become greater than the received signal strength of the free settling layer is defined as the interface between the free settling layer and the flocculating settling layer. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 3-274484 [Patent Document 2] JP 2011-047761 A [Patent Document 3] JP 2011-13084 A [Patent Document 4] Japanese Patent Application Publication No. 4-264235 [Patent Document 5] JP 2011-47760 A Summary of the Invention [Problem to be solved by the invention]

[0006] In the above-mentioned technology, images obtained by monitoring the processing status (hereinafter referred to as "monitoring images") were interpreted by people. People would diagnose the current processing status by interpreting the monitoring images. People would also diagnose the cause of the results of the diagnosed processing status based on their experience. Furthermore, people would also decide on how to deal with the problem. Experience is required to interpret monitoring images, and the interpretation results may differ depending on the person interpreting them. The present invention has been made in consideration of the above-mentioned circumstances, and has an object to provide a monitoring system, learning device, monitoring method, learning method, and program capable of monitoring the internal condition of a solid-liquid separation tank for separating wastewater into solid and liquid. [Means for solving the problem]

[0007] (1) One aspect of the present invention is a method for solid-liquid separation of wastewater. Multiple Solid-liquid separation tank Each of The surveillance image is an image showing the inside of the Multiple The solid-liquid separation tank Each of Based on the diagnosis result based on the monitoring image of the inside of the solid-liquid separation tank, a first learning model that has learned the relationship between the monitoring image and the internal state of the solid-liquid separation tank is used to determine the internal state of the solid-liquid separation tank. Multiple Solid-liquid separation tank Each of From surveillance images showing the inside of Multiple Solid-liquid separation tank Each of A determination unit that determines the internal state of the Multiple The solid-liquid separation tank Each of The determination unit determines using the surveillance image and the first learning model. Multiple The solid-liquid separation tank Each of and an output unit that outputs information specifying the internal state of the monitoring system. (2) One aspect of the present invention is a monitoring system according to the above (1), Multiple The surveillance image; Multiple The monitoring image Each of The solid-liquid separation tank further includes a cause determination unit that determines information identifying the cause of the state inside the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, using a second learning model that has learned the relationship between the monitoring image and the information identifying the cause of the diagnosis result inside the solid-liquid separation tank based on information identifying the cause of the diagnosis result based on the monitoring image of the solid-liquid separation tank that is the target of diagnosis, and the output unit further outputs information identifying the cause of the state inside the solid-liquid separation tank determined by the cause determination unit using the monitoring image of the solid-liquid separation tank that is the target of diagnosis and the second learning model. (3) One aspect of the present invention is a monitoring system according to the above (1) or (2), Multiple The surveillance image; Multiple The monitoring image Each of The solid-liquid separation tank further includes a countermeasure determination unit that determines information specifying how to deal with the condition inside the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, using a third learning model that has learned the relationship between the monitoring image and information specifying how to deal with the diagnosis result inside the solid-liquid separation tank based on the monitoring image of the solid-liquid separation tank that is the target of diagnosis, and the output unit further outputs information specifying how to deal with the condition inside the solid-liquid separation tank determined by the countermeasure determination unit using the monitoring image of the solid-liquid separation tank that is the target of diagnosis and the third learning model. (4) One aspect of the present invention is a monitoring system according to any one of (1) to (3), Multiple The surveillance image; Multiple The monitoring image Each ofThe system further includes a change prediction deriving unit that detects signs of a change in the state inside the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, using a fourth learning model that has learned the relationship between the monitoring image and the information that specifies the change in the state inside the solid-liquid separation tank based on information that specifies the change in the state inside the solid-liquid separation tank after the monitoring image has been obtained, and the output unit further outputs information that specifies the sign of a change in the state inside the solid-liquid separation tank detected by the change prediction deriving unit using the monitoring image of the solid-liquid separation tank that is the target of diagnosis and the fourth learning model. (5) In one aspect of the present invention, in a monitoring system described in any one of (1) to (4) above, the diagnostic result is generated based on either or both of the accumulation state of solid matter and the floating state of solid matter contained in the monitoring image. (6) One aspect of the present invention is a monitoring system described in any one of (1) to (5) above, wherein the judgment unit judges whether the internal condition of the solid-liquid separation tank, which is the subject of diagnosis, is normal, malfunction, or abnormal from the monitoring image showing the interior of the solid-liquid separation tank. (7) In one aspect of the present invention, in the monitoring system described in any one of (1) to (6) above, the monitoring system further includes a notification unit that notifies that the condition inside the solid-liquid separation tank is either malfunction or abnormal when the judgment unit judges that the condition inside the solid-liquid separation tank is either malfunction or abnormal.

[0008] (8) One aspect of the present invention is A learning device for generating a first learning model for determining an internal state of each of a plurality of solid-liquid separation tanks to be diagnosed from a monitored image showing the interior of each of the plurality of solid-liquid separation tanks, comprising: For solid-liquid separation of wastewater Multiple Solid-liquid separation tank Each of The surveillance image is an image showing the inside of the Multiple The solid-liquid separation tank Each of The learning device has a learning unit that generates by learning a first learning model that represents the relationship between the monitoring image and the internal state of the solid-liquid separation tank based on the diagnosis results based on the monitoring image of the internal state of the solid-liquid separation tank. (9) In one aspect of the present invention, in the learning device according to (8) above, the learning unit Multiple The surveillance image; Multiple The monitoring image Each ofBased on the information identifying the cause of the diagnostic result based on the above, a second learning model is generated by learning, which represents the relationship between the monitoring image and the information identifying the cause of the diagnostic result inside the solid-liquid separation tank. (10) In one aspect of the present invention, in the learning device according to (8) or (9), the learning unit: Multiple The surveillance image; Multiple The monitoring image Each of Based on the information specifying how to deal with the diagnostic results based on the monitoring image, a third learning model is generated by learning, which represents the relationship between the monitoring image and the information specifying how to deal with the diagnostic results of the inside of the solid-liquid separation tank. (11) In one aspect of the present invention, in the learning device according to any one of (8) to (10), the learning unit: Multiple The surveillance image; Multiple The monitoring image Each of Based on the information specifying the change in the internal state of the solid-liquid separation tank after the monitoring image has been obtained, a fourth learning model is generated that represents the relationship between the monitoring image and the information specifying the change in the internal state of the solid-liquid separation tank. (12) In one aspect of the present invention, in a learning device described in any one of (8) to (11) above, the diagnosis result is generated based on either or both of the accumulation state of solid matter and the floating state of solid matter contained in a monitoring image.

[0009] (13) One aspect of the present invention is a method for solid-liquid separation of wastewater. Multiple Solid-liquid separation tank Each of The surveillance image is an image showing the inside of the Multiple The solid-liquid separation tank Each of Based on the diagnosis result based on the monitoring image of the inside of the solid-liquid separation tank, a first learning model that has learned the relationship between the monitoring image and the internal state of the solid-liquid separation tank is used to determine the internal state of the solid-liquid separation tank. Multiple Solid-liquid separation tank Each of From surveillance images showing the inside of Multiple Solid-liquid separation tank Each of determining the internal state of the Multiple The solid-liquid separation tank Each of The step of determining using the surveillance image and the first learning model Multiple The solid-liquid separation tank Each of and outputting information specifying an internal state of the monitoring system.

[0010] (14) One aspect of the present invention is a learning method executed by a learning device that generates a first learning model for determining an internal state of each of a plurality of solid-liquid separation tanks to be diagnosed from a monitoring image showing the interior of each of the plurality of solid-liquid separation tanks to be diagnosed, the method comprising: Multiple Solid-liquid separation tank Each of The surveillance image is an image showing the inside of the Multiple The solid-liquid separation tank Each of The learning method is executed by a learning device, and includes a step of generating, by learning, a first learning model that represents the relationship between the monitoring image and the internal state of the solid-liquid separation tank based on a diagnosis result based on the monitoring image of the interior of the solid-liquid separation tank.

[0011] (15) In one aspect of the present invention, a computer of the monitoring system is provided with a device for performing solid-liquid separation of wastewater. Multiple Solid-liquid separation tank Each of The surveillance image is an image showing the inside of the Multiple The solid-liquid separation tank Each of Based on the diagnosis result based on the monitoring image of the inside of the solid-liquid separation tank, a first learning model that has learned the relationship between the monitoring image and the internal state of the solid-liquid separation tank is used to determine the internal state of the solid-liquid separation tank. Multiple Solid-liquid separation tank Each of From surveillance images showing the inside of Multiple Solid-liquid separation tank Each of determining the internal state of the Multiple The solid-liquid separation tank Each of The step of determining using the surveillance image and the first learning model Multiple The solid-liquid separation tank Each of and outputting information that specifies the internal state of the program.

[0012] (16) One aspect of the present invention is A first learning model is generated to determine the internal state of each of the plurality of solid-liquid separation tanks from a monitoring image showing the interior of each of the plurality of solid-liquid separation tanks to be diagnosed. The learning device's computer is equipped with a Multiple Solid-liquid separation tank Each of The surveillance image is an image showing the inside of the MultipleThe solid-liquid separation tank Each of The program executes a step of generating, by learning, a first learning model that represents a relationship between the monitoring image and the internal state of the solid-liquid separation tank, based on a diagnosis result based on the monitoring image of the interior of the solid-liquid separation tank. Effect of the Invention

[0013] According to the present invention, it is possible to provide a monitoring system, a learning device, a monitoring method, a learning method and a program capable of monitoring the internal condition of a solid-liquid separation tank for separating wastewater into solid and liquid. [Brief description of the drawings]

[0014] [Figure 1] 1 is a diagram illustrating an example of the configuration of a monitoring system according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram illustrating an example of an ultrasonic sensor. [Diagram 3] FIG. 2 is a diagram illustrating an example of a data processing device of the monitoring system according to the present embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of an operation of the monitoring system according to the present embodiment. [Diagram 5] FIG. 13 is a diagram showing an example of a monitoring image. [Figure 6] FIG. 11 is a diagram illustrating an example of teacher data. [Figure 7] FIG. 4 is a diagram illustrating a first example of the operation of the monitoring system according to the present embodiment. [Figure 8] FIG. 11 is a diagram illustrating a second example of the operation of the monitoring system according to the present embodiment. [Figure 9] FIG. 11 is a diagram illustrating a third example of the operation of the monitoring system according to the present embodiment. [Figure 10] FIG. 11 is a diagram illustrating another example of the data processing device according to the present embodiment. [Figure 11] FIG. 2 is a diagram illustrating an example of a configuration of a monitoring system according to a first modified example of an embodiment of the present invention. [Figure 12] FIG. 11 is a diagram illustrating an example of teacher data. [Figure 13] FIG. 11 is a diagram illustrating a first example of an operation of a monitoring system according to a first modified example of an embodiment. [Figure 14] FIG. 11 is a diagram illustrating a second example of the operation of the monitoring system according to the first modified example of the embodiment. [Figure 15] FIG. 11 is a diagram illustrating an example of the configuration of a monitoring system according to a second modified example of the embodiment of the present invention. [Figure 16] FIG. 11 is a diagram illustrating an example of teacher data. [Figure 17] FIG. 11 is a diagram illustrating a first example of an operation of a monitoring system according to a second modification of an embodiment. [Figure 18] FIG. 11 is a diagram illustrating a second example of the operation of the monitoring system according to the second modification of the embodiment. [Figure 19] FIG. 11 is a diagram illustrating an example of the configuration of a monitoring system according to a third modified example of an embodiment of the present invention. [Figure 20] FIG. 11 is a diagram illustrating a first example of an operation of a monitoring system according to a third modified example of an embodiment. [Figure 21] FIG. 11 is a diagram illustrating a second example of the operation of the monitoring system according to the third modification of the embodiment. [Figure 22] FIG. 11 is a diagram illustrating an example of the configuration of a monitoring system according to a fourth modified example of an embodiment of the present invention. [Figure 23] FIG. 11 is a diagram illustrating an example of a monitoring device of a monitoring system according to a fourth modified example of the present embodiment. [Figure 24] FIG. 13 is a diagram illustrating a first example of an operation of a monitoring system according to a fourth modified example of an embodiment. [Diagram 25] FIG. 13 is a diagram illustrating a second example of the operation of the monitoring system according to the fourth modified example of the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] A monitoring system, a monitoring method, and a program according to the present embodiment will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment. In addition, in all the drawings for explaining the embodiments, the same reference numerals are used for the parts having the same functions, and the repeated explanation is omitted. In addition, "based on XX" in this application means "based on at least XX" and includes cases where it is based on other elements in addition to XX. Furthermore, "based on XX" is not limited to cases where XX is used directly, but also includes cases where it is based on XX that has been calculated or processed. "XX" is any element (for example, any information).

[0016] [Embodiment] (Surveillance System) 1 is a diagram showing an example of the configuration of a monitoring system according to an embodiment of the present invention. A monitoring system 100 according to this embodiment diagnoses the sludge accumulation state in a solid-liquid separation tank such as a settling tank or a thickening tank. In this embodiment, a sewage treatment facility 10 will be described as an example of a facility equipped with a solid-liquid separation tank. (Sewage treatment facility 10) An example of a sewage treatment facility 10 includes a pre-settling tank 11, a thickening tank 12, a storage tank 13, a dehydrator 14, a container 15, an aeration tank 16, a post-settling tank 17, a pump 18, and a facility control device 19. The pre-settling tank 11 is connected to the aeration tank 16 by a flow path P1. Raw water is introduced into the pre-settling tank 11. The pre-settling tank 11 separates primary sludge (extracted sludge) from the introduced raw water by settling. The treated water after settling is introduced into the aeration tank 16 via the flow path P1. The aeration tank 16 is connected to the post-settling tank 17 by a flow path P2. The aeration tank 16 performs aerobic treatment on the water to be treated introduced from the pre-settling tank 11 by aeration with air from an aeration pipe. The water to be treated that has been aerobically treated in the aeration tank 16 is introduced into the post-settling tank 17 via the flow path P2.

[0017] The post-settling tank 17 is connected to a pump 18 via a flow path P3. The pump 18 is connected to a flow path P4. The flow path P4 is branched into a flow path P5 and a flow path P6. The flow path P5 is connected to the thickening tank 12, and the flow path P6 is connected to the aeration tank 16. The post-settling tank 17 separates the water to be treated introduced from the aeration tank 16 into settled sludge (extracted sludge) and supernatant water. The supernatant water of the post-settling tank 17 is discharged outside the sewage treatment facility 10 as effluent. A part of the sludge that has settled in the post-settling tank 17 is introduced as excess sludge into the thickening tank 12 via the pump 18, the flow path P4, and the flow path P5. The remaining sludge that has settled in the post-settling tank 17 is returned as returned sludge to the aeration tank 16 via the flow path P4 and the piping P6. The pump 18 is controlled by the equipment control device 19, so that a predetermined amount of sludge that has settled in the post-settling tank 17 is introduced into the flow path P4. In addition, the pre-settling tank 11 is connected to the thickening tank 12 by a flow path P7. Extracted sludge is introduced from the pre-settling tank 11 to the thickening tank 12 via the flow path P7. The thickening tank 12 is connected to the pre-settling tank 11 by a flow path P8, and is connected to the storage tank 13 by a flow path P9.

[0018] In the thickening tank 12, the sludge introduced is separated into supernatant water and thickened sludge by gravity. The supernatant water is returned to the pre-settling tank 11 via flow path P8. The thickened sludge is extracted from the bottom of the thickening tank 12 and introduced into the storage tank 13 via flow path P9. The storage tank 13 is connected to the dehydrator 14 by a flow path P10. The storage tank 13 temporarily stores the concentrated sludge introduced from the concentration tank 12. The concentrated sludge stored in the concentration tank 12 is introduced to the dehydrator 14. The dehydrator 14 is connected to a container 15 by a conveyor P11. The dehydrator 14 dehydrates the concentrated sludge introduced from the storage tank 13. The dehydrated cake generated by the dehydration is introduced into the container 15 via the conveyor P11. The container 15 stores the dehydrated cake introduced by the dehydrator 14 and carries out the stored dehydrated cake.

[0019] (Monitoring system 100) The monitoring system 100 includes an ultrasonic sensor 20, a data processing device 30, a gateway device 31, an information processing device 40, a terminal device 45, and a monitoring device 50. The gateway device 31, the information processing device 40, the terminal device 45, and the monitoring device 50 are connected via a network NW. The network NW is a wireless or wired communication network. This network NW includes the Internet and an intranet. Specifically, the network NW is an information communication network configured by a WAN (Wide Area Network), a LAN (Local Area Network), and the like. This WAN includes, for example, a mobile phone network, a PHS (Personal Handy-phone System) network, a PSTN (Public Switched Telephone Network), a dedicated communication line network, and a VPN (Virtual Private Network).

[0020] The ultrasonic sensor 20 transmits ultrasonic waves into the water by applying a pulse voltage from the ultrasonic transmission circuit to the ultrasonic transducer. In this embodiment, as an example, the ultrasonic sensor 20 is installed in the post-settling tank 17, and the explanation will continue with the case where ultrasonic waves are transmitted into the interior of the post-settling tank. Here, the voltage [V] is converted into sound pressure [dB]. The transducer vibrates slightly to generate ultrasonic waves. An example of the transducer is a ceramic element. When the transducer receives a reflected wave that is reflected off "something" in the water, it generates an electromotive force by being electromotive, and the ultrasonic receiving circuit detects the voltage. Fig. 2 is a diagram showing an example of an ultrasonic sensor. As shown in Fig. 2, the ultrasonic sensor 20 includes an oscillator (transmitter) 21 which is a transmitting oscillator 2, and a receiver 22 which is a receiving oscillator. Fig. 2 shows, as an example, a case where the ultrasonic sensor 20 includes two oscillators, the oscillator 21 and the receiver 22. However, the transmitting oscillator and the receiving oscillator may be realized by a single oscillator. The ultrasonic sensor 20 is attached by a mechanism (not shown) at a predetermined height 27 of the post-settling tank 17 (hereinafter also referred to as the treatment tank 25) that stores the suspended solids sediment layer 23 such as sludge and the supernatant water 24 of the suspended solids sediment layer 23. By installing the ultrasonic sensor 20 in the treatment tank 25 to be measured, the depth is fixed, so the depth is not changed. For example, when the tank is 5 m deep, an image database can be created by allocating 200 dots to 5 m. In this case, one pixel is 2.5 cm, and the display resolution is 2.5 cm. For example, there is a setting item for inputting the depth when setting up the measurement, and the depth direction data may be thinned out based on this setting value. In order to allow a certain portion to be enlarged, all data may be stored and thinned out and displayed according to an instruction, or partially enlarged and displayed (full display). The oscillator 21 provides an electrical signal generated by a signal generating circuit (not shown) to an ultrasonic transducer, which transmits the signal toward the bottom surface of the treatment tank 25 .

[0021] The ultrasonic waves transmitted by the oscillator 21 are reflected by the interface 26 between the suspended matter sediment layer 23 and the supernatant water 24, the suspended matter below the interface 26, the bottom of the treatment tank 25, and the like. The reflected waves return one after another with a time difference (arrival time) proportional to the position (distance, depth) of the object that reflected them. The strength of the reflected waves is related to the properties (≒ density) of the object, and the information is expressed in sound pressure (dB). The reflected waves are received by the receiver 22. In the receiver 22, the sound pressure vibrates the transducer, generating a voltage according to the strength. Here, the sound pressure [dB] is converted to a voltage [V]. The receiver 22 outputs the received signal to the data processor 30. The data processor 30 receives the received signal output by the receiver 22 and converts the received signal into image data.

[0022] Fig. 3 is a diagram showing an example of a data processing device of the monitoring system according to the present embodiment. In the example shown in Fig. 3, a case will be described in which a transmitting transducer and a receiving transducer are realized by a single transducer. The data processing device 30 includes an ultrasonic transmission / reception circuit 32 , a data conversion circuit 33 , a data calculation section 34 , and an image data storage section 35 . The ultrasonic transmission reception circuit 32 generates an electrical signal for transmitting an ultrasonic wave, and outputs the generated electrical signal to the ultrasonic sensor 20. The ultrasonic transmission reception circuit 32 receives the electrical signal output by the ultrasonic sensor 20. The ultrasonic transmission reception circuit 32 outputs the received electrical signal to the data conversion circuit 33. The data conversion circuit 33 acquires the electrical signal output by the ultrasonic transmission / reception circuit 32. The data conversion circuit 33 amplifies the acquired electrical signal. The data conversion circuit 33 performs masking processing on the amplified electrical signal. Based on the result of masking processing on the amplified electrical signal, the data conversion circuit 33 converts the signal strength into a digital signal by digital processing. For example, the data conversion circuit 33 converts the electrical signal into, for example, 256 gradations based on the signal strength. The data conversion circuit 33 outputs the digital signal to the data calculation unit 34.

[0023] The data calculation unit 34 acquires the digital signal output by the data conversion circuit 33. The data calculation unit 34 also acquires temperature data from a thermocouple (not shown) installed in the ultrasonic sensor 20. The data calculation unit 34 performs a correction calculation of the speed of sound traveling through water using the acquired temperature data. The data calculation unit 34 also expresses the position (= distance) of the signal as a function of time based on the acquired digital signal. The data calculation unit 34 also calculates the change in reflection intensity (signal strength) over time since the ultrasonic sensor 20 transmitted ultrasonic waves based on the acquired digital signal. The data calculation unit 34 associates the signal strength with the position information based on the result of expressing the position (= distance) of the signal as a function of time and the result of calculating the change in reflection intensity (signal strength) over time since the ultrasonic sensor 20 transmitted ultrasonic waves. The data calculation unit 34 associates the signal strength with the position information and temporarily stores (stocks) it. The data calculation unit 34 calculates the position (depth) of the interface 26 between the suspended matter sediment layer 23 and the supernatant water 24. For example, the data calculation unit 34 derives the elapsed time from when the ultrasonic sensor 20 transmits ultrasonic waves to when the reflection intensity suddenly increases beyond a predetermined threshold, based on the elapsed time of the reflection intensity of the ultrasonic waves. The data calculation unit 34 calculates the distance to the interface 26 (the position of the interface 26) based on the derived elapsed time. The data calculation unit 34 outputs digital data of the numerical value of the interface level to the image data storage unit 35.

[0024] The data calculation unit 34 outputs the stored information associating the signal intensity with the position information, and the digital data of the numerical value of the interface level to the image data storage unit 35. Here, if the data calculation unit 34 is unable to determine the interface level, it may output information indicating a determination error to the image data storage unit 35. The data calculation unit 34 may output to the image data storage unit 35 the information associating the signal intensity with the position information, and information associating each of the digital data of the numerical value of the interface level with temperature data. Fig. 4 is a diagram showing an example of the operation of the monitoring system according to this embodiment. Fig. 4 shows an example of data output from the data calculation unit 34 to the image data storage unit 35. The example of data output from the data calculation unit 34 to the image data storage unit 35 is expressed as an instantaneous value. Fig. 4 shows the instantaneous value output from the data calculation unit 34 to the image data storage unit 35 in two dimensions. In the data processing device 30, the data calculation unit 34 transmits the digital signal to the monitoring device 50 via the gateway device 31. Returning to FIG.

[0025] (Monitoring device 50) The monitoring device 50 is realized by a device such as a personal computer, a server, or an industrial computer. The monitoring device 50 includes a communication device 51, a recording device 52, an information processing device 53, and a bus line BL such as an address bus or a data bus for electrically connecting the components as shown in FIG. The communication device 51 is realized by a communication module. The communication device 51 communicates with other devices such as the data processing device 30 and the information processing device 40 via the network NW. The communication device 51 receives a digital signal transmitted by the data processing device 30. For example, the communication device 51 receives data measured during a past predetermined time at each predetermined time interval. Specifically, the communication device 51 receives data for the past hour once every hour. The communication device 51 also receives a monitoring image request sent by the terminal device 45 to request a monitoring image. Here, the monitoring image is an image showing the change in reflection intensity (reception intensity) over time since ultrasonic waves were transmitted to the solid-liquid separation tank. In response to the received monitoring image request, the communication device 51 transmits a monitoring image response output by the information processing unit 53 to the terminal device 45. In response to the transmitted monitoring image response, the communication device 51 receives a diagnosis result notification sent by the terminal device 45. The diagnosis result notification includes information showing the monitoring image and information showing the diagnosis result of the internal condition of the solid-liquid separation tank. Furthermore, the communication device 51 receives the tank internal status information request transmitted by the information processing device 40. The communication device 51 transmits the tank internal status information response output by the information processing unit 53 to the information processing device 40. The communication device 51 acquires the status notification information output by the information processing unit 53, and transmits the acquired status notification information to the information processing device 40.

[0026] The recording device 52 is realized by, for example, a RAM, a ROM, a HDD, a flash memory, or a hybrid storage device in which a plurality of these are combined. The recording device 52 stores a program (monitoring application) executed by the monitoring device 50. The recording device 52 also stores pixel data output by the information processing unit 53. The recording device 52 stores information indicating the monitoring image and diagnostic result training data associating the information indicating the monitoring image with the diagnostic result of the inside of the solid-liquid separation tank (inside the tank) based on the monitoring image, and a learning model of the diagnostic result obtained by machine learning the relationship between the monitoring image and the internal state of the solid-liquid separation tank based on the diagnostic result training data. Here, the monitoring image will be described. 5 shows an example of a monitoring image. The data processing device 30 transmits ultrasonic waves from the ultrasonic sensor 20 downward in water (toward the bottom), receives the reflected waves that hit objects within the range of the ultrasonic waves, and converts the received reflected waves into digital signals. The information processing unit 53 acquires the digital signal transmitted by the data processing device 30. Based on the acquired digital signal, the information processing unit 53 converts the intensity of the reflected wave into a color tone, converts the time it takes for the reflected wave to return into a distance to provide position information, and combines the color tone and distance to continuously plot the distance from the sensor (=water depth) vertically and over time (=time series) horizontally. This continuous plot is the monitoring image. As shown in Fig. 5, the monitoring image shows images corresponding to the bottom of the solid-liquid separation tank, the bottom of the solid-liquid separation tank, the rake, the sludge sediment layer, the sludge interface, and the supernatant water. Fig. 6 is a diagram showing an example of teacher data. Fig. 6 shows teacher data of diagnosis results. The teacher data of diagnosis results is data that associates a monitoring image with a diagnosis result of the internal state of the solid-liquid separation tank based on the monitoring image. In the present embodiment, as an example, for each of a plurality of monitoring images, one of "normal", "abnormal", or "malfunction" is associated as the diagnosis result. In Figure 6, (1) is diagnosed as normal because the supernatant water and the sludge sediment layer are separated, and solid-liquid separation is in good condition. (2) is diagnosed as abnormal because the sludge's settling ability has deteriorated and the sludge is floating up. (3) is diagnosed as malfunctioning because the sedimented sludge is seen flying up. Let's return to Figure 1 and continue the explanation.

[0027] The information processing unit 53 functions as, for example, a graphics unit 54, a current state determination unit 55, and a learning unit 56. The graphic conversion unit 54 acquires the digital signal received by the communication device 51. The graphic conversion unit 54 converts the value of the acquired digital signal into pixel data. The graphic conversion unit 54 stores the pixel data after the conversion of the digital signal in the recording device 52. The graphic conversion unit 54 acquires the surveillance image request received by the communication device 51. The graphic conversion unit 54 acquires pixel data stored in the recording device 52 based on the acquired surveillance image request. The graphic conversion unit 54 creates a surveillance image based on the acquired pixel data. The graphic conversion unit 54 includes information indicating the created surveillance image. The graphic conversion unit 54 creates a surveillance image response addressed to the information processing device 40. The graphic conversion unit 54 outputs the created surveillance image response to the communication device 51. The graphic conversion unit 54 acquires the tank interior status information request received by the communication device 51. The graphic conversion unit 54 acquires pixel data stored in the recording device 52 based on the acquired tank interior status information request, and creates a monitoring image based on the acquired pixel data. The graphic conversion unit 54 includes information indicating the created monitoring image. The graphic conversion unit 54 creates an tank interior status information response addressed to the information processing device 40. The graphic conversion unit 54 outputs the created tank interior status information response to the communication device 51.

[0028] The current status determination unit 55 acquires pixel data stored in the recording device 52, and creates a monitoring image based on the acquired pixel data. The current status determination unit 55 acquires a learning model of the diagnosis result stored in the recording device 52. The current status determination unit 55 judges the internal state of the solid-liquid separation tank of the created monitoring image based on the learning model of the acquired diagnosis result. When the judgment result of the internal state of the solid-liquid separation tank is malfunction or abnormal, the current status determination unit 55 creates status notification information addressed to the information processing device 40, including information indicating the judgment result of the internal state of the solid-liquid separation tank. The current status determination unit 55 outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the current status determination unit 55, and transmits the acquired status notification information to the information processing device 40. When creating a monitoring image, the current status determination unit 55 may use the measured data as is, or may include data measured over a long period of time in a limited display width by thinning out the data. By including data measured over a long period of time in a limited display width, it is possible to monitor changes over a longer period of time. If it were a still image, pixel data could be picked up at any suitable interval and switched for display, but in this embodiment, measurements are constantly performed and new data is added, so if pixel data is picked up at any suitable interval and switched for display, there is a risk of delay or obstruction in data processing, and if the measurement becomes unstable due to image display, it would be counterproductive. Therefore, in this embodiment, several preset display time widths are prepared, and data storage areas for time widths corresponding to each of the multiple display time widths are created. In this embodiment, an interval (interval) for adding new data is specified, and an image database (data storage area (address)) corresponding to each of the multiple intervals is created.

[0029] An operation to switch the display is performed on the monitoring device 50, and a display time width is selected. The current situation determination unit 55 acquires data from a database corresponding to the selected time display width, and creates a monitoring image using the acquired data. If an operation to switch the time display width is performed, data is acquired from a database corresponding to the selected time display width, and a monitoring image is created using the acquired data. By configuring in this way, smooth switching can be achieved without processing data in the database in which the data is stored, and without a time lag in creating a monitoring image. The learning unit 56 acquires the diagnosis result notification received by the communication device 51, and stores in the recording device 52 the teacher data of the diagnosis result in which the information indicating the monitoring image included in the acquired diagnosis result notification is associated with the diagnosis result of the internal state (inside the tank) of the solid-liquid separation tank based on the monitoring image. The learning unit 56 acquires the teacher data of the diagnosis result stored in the recording device 52. The learning unit 56 generates a learning model of the diagnosis result in which the monitoring image and the internal state of the solid-liquid separation tank are associated with each other by machine learning (supervised learning) of the monitoring image and the diagnosis result of the internal state of the solid-liquid separation tank based on the acquired teacher data of the diagnosis result. For example, the learning unit 56 recognizes the monitoring image using a convolutional neural network (CNN). The learning model of the diagnosis result classifies the monitoring image into one of normal, malfunction, and abnormal as the internal state of the solid-liquid separation tank based on the information indicating the monitoring image. The learning unit 56 stores the generated learning model of the diagnosis result in the recording device 52. All or a part of the information processing unit 53 is a functional unit (hereinafter referred to as a software functional unit) that is realized by a processor such as a CPU (Central Processing Unit) executing a program such as a monitoring application stored in the recording device 52. All or a part of the information processing unit 53 may be realized by hardware such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array), or may be realized by a combination of a software functional unit and hardware.

[0030] (Information processing device 40) The information processing device 40 is realized by a device such as a personal computer, a server, an industrial computer, etc. One example of the information processing device 40 is installed in a monitoring center that monitors the sewage treatment facility 10 from a remote location. When the information processing device 40 receives the status information notification transmitted by the monitoring device 50, it displays the judgment result of the solid-liquid separation tank included in the received status information notification. In addition, the information processing device 40 creates an in-tank status information request addressed to the monitoring device 50, including information requesting the status inside the tank, based on an operation by the operator to obtain information on the status inside the solid-liquid separation tank. The information processing device 40 transmits the created in-tank status information request to the communication device 51. The information processing device 40 receives the tank interior status information response transmitted by the monitoring device 50 in response to the tank interior status information request. The information processing device 40 acquires the monitoring image included in the received tank interior status information response. The information processing device 40 displays the acquired monitoring image.

[0031] (Terminal Device 45) The terminal device 45 is realized by a device such as a personal computer, a server, an industrial computer, etc. An example of the terminal device 45 is installed in a monitoring center that monitors the sewage treatment facility 10. When diagnosing the internal condition of the solid-liquid separation tank, the user operates the terminal device 45 to create a monitoring image request addressed to the monitoring device 50, the request including information requesting a monitoring image. The terminal device 45 creates the monitoring image request based on the user's operation. The terminal device 45 transmits the created monitoring image request to the monitoring device 50. The terminal device 45 receives the monitoring image response transmitted by the monitoring device 50 in response to the monitoring image request transmitted to the monitoring device 50. The terminal device 45 displays the monitoring image included in the monitoring image response. The user refers to the monitoring image displayed by the terminal device 45 and diagnoses the internal condition of the solid-liquid separation tank included in the monitoring image. The user operates the terminal device 45 to create a diagnosis result notification addressed to the monitoring device 50, the diagnosis result notification including information showing a monitoring image and a diagnosis result of the internal condition of the solid-liquid separation tank. The terminal device 45 creates the diagnosis result notification based on the user's operation. The terminal device 45 transmits the created diagnosis result notification to the monitoring device 50.

[0032] (Operation of monitoring system) Fig. 7 is a diagram showing a first example of the operation of the monitoring system according to the present embodiment. With reference to Fig. 7, a process will be described in which monitoring device 50 accumulates the diagnosis results of the internal condition of the solid-liquid separation tank contained in the diagnosis result notification transmitted by terminal device 45, and performs machine learning based on the accumulated diagnosis results of the internal condition of the solid-liquid separation tank to generate a learning model of the diagnosis results. (Step S1-1) In the data processing device 30 , an ultrasonic transmission / reception circuit 32 generates an electrical signal for transmitting ultrasonic waves, and outputs the generated electrical signal to the ultrasonic sensor 20 . (Step S2-1) In the data processing device 30 , an ultrasonic transmission / reception circuit 32 receives the electrical signal output by the ultrasonic sensor 20 . (Step S3-1) In the data processing device 30, the ultrasonic transmission / reception circuit 32 outputs the received electrical signal to the data conversion circuit 33. The data conversion circuit 33 acquires the electrical signal output by the ultrasonic transmission / reception circuit 32. The data conversion circuit 33 amplifies the acquired electrical signal. The data conversion circuit 33 performs masking processing on the amplified electrical signal. Based on the result of masking processing on the amplified electrical signal, the data conversion circuit 33 converts the signal into a digital signal by digital processing the signal intensity. The data calculation unit 34 acquires the digital signal from the data conversion circuit 33, and performs temperature correction calculation related to position (distance) information and interface level determination calculation on the acquired digital signal. (Step S4-1) In the data processing device 30, the data calculation unit 34 transmits to the monitoring device 50 via the gateway device 31 a digital signal that has been subjected to temperature correction calculation related to the position (distance) information and calculation for determining the interface level. (Step S5-1) In the monitoring device 50, the communication device 51 receives the digital signal transmitted by the data processing device 30. The graphic conversion unit 54 acquires the digital signal received by the communication device 51. The graphic conversion unit 54 converts the value of the acquired digital signal into pixel data. (Step S6-1) In the monitoring device 50, the graphics generating unit 54 stores the pixel data after conversion into a digital signal in the recording device 52. (Step S7-1) The terminal 45 generates a surveillance image request.

[0033] (Step S8-1) The terminal device 45 transmits the created surveillance image request to the surveillance device 50 . (Step S9-1) In the monitoring device 50, the communication device 51 receives the monitoring image request sent by the terminal device 45. The graphic conversion unit 54 acquires the monitoring image request received by the communication device 51. The graphic conversion unit 54 acquires the pixel data stored in the recording device 52 based on the acquired monitoring image request. The graphic conversion unit 54 creates a monitoring image based on the acquired pixel data. The graphic conversion unit 54 includes information indicating the created monitoring image. A monitoring image response addressed to the terminal device 45 is created. (Step S10-1) In the monitoring device 50, the graphic conversion unit 54 outputs the created monitoring image response to the communication device 51. The communication device 51 acquires the monitoring image response output by the graphic conversion unit 54, and transmits the acquired monitoring image response to the terminal device 45. (Step S11-1) The terminal device 45 receives the surveillance image response transmitted by the surveillance device 50. The terminal device 45 displays the surveillance image by performing image processing on the information indicating the surveillance image included in the received surveillance image response. The terminal device 45 creates a diagnosis result notification including the information indicating the surveillance image and the result of diagnosing the surveillance image. (Step S12-1) The terminal device 45 transmits the created diagnosis result notification to the monitoring device 50. (Step S13-1) In the monitoring device 50, the communication device 51 receives the diagnosis result notification transmitted by the terminal device 45. The learning unit 56 acquires the diagnosis result notification received by the communication device 51, and stores in the recording device 52 teacher data of the diagnosis result that associates information indicating the monitoring image included in the acquired diagnosis result notification with the diagnosis result of the state of the inside (inside the tank) of the solid-liquid separation tank based on the monitoring image. (Step S14-1) In the monitoring device 50, the learning unit 56 acquires the teacher data of the diagnosis results stored in the recording device 52. The learning unit 56 performs machine learning of the monitoring images and the diagnosis results of the internal state of the solid-liquid separation tank based on the monitoring images, based on the acquired teacher data of the diagnosis results, to generate a learning model of the diagnosis results that associates the monitoring images with the internal state of the solid-liquid separation tank. (Step S15-1) In the monitoring device 50, the learning unit 56 stores the generated learning model of the diagnosis results in the recording device 52.

[0034] Fig. 8 is a diagram showing a second example of the operation of the monitoring system according to this embodiment. With reference to Fig. 8, the monitoring device 50 acquires the digital signal transmitted by the data processing device 30, and creates a monitoring image based on the acquired digital signal. The process of determining the internal state of the solid-liquid separation tank by the monitoring device 50 based on the created monitoring image will be described. Steps S1-2 to S6-2 can be implemented by steps S1-1 to S6-1 in FIG. 7, so a description thereof will be omitted here. (Step S7-2) In the monitoring device 50, the current state determining unit 55 acquires the pixel data stored in the recording device 52, and creates a monitoring image based on the acquired pixel data. (Step S8-2) In the monitoring device 50, the current state determination unit 55 acquires a learning model of the diagnosis results stored in the recording device 52. (Step S9-2) In the monitoring device 50, the current state determining unit 55 determines the internal state of the solid-liquid separation tank from the created monitoring image based on a learning model of the acquired diagnosis results. (Step S10-2) In monitoring device 50, current state determination unit 55 determines whether the determination result of the internal state of the solid-liquid separation tank is malfunction or abnormal. If current state determination unit 55 determines that the determination result of the internal state of the solid-liquid separation tank is neither malfunction nor abnormal, that is, that the tank is normal, the process ends. (Step S11-2) In the monitoring device 50, when the current status determination unit 55 determines that the determination result of the internal condition of the solid-liquid separation tank is malfunction or abnormal, it creates status notification information addressed to the information processing device 40, including information indicating the determination result of the internal condition of the solid-liquid separation tank. (Step S12-2) In the monitoring device 50, the current state determination unit 55 outputs the created state notification information to the communication device 51. The communication device 51 acquires the state notification information output by the current state determination unit 55, and transmits the acquired state notification information to the information processing device 40.

[0035] Fig. 9 is a diagram showing a third example of the operation of the monitoring system according to the present embodiment. A process in which the monitoring device 50 transmits information indicating a monitoring image based on a request for information on a tank status transmitted by the information processing device 40 will be described with reference to Fig. 9. Steps S1-3 to S6-3 can be implemented by steps S1-1 to S6-1 in FIG. 7, so a description thereof will be omitted here. (Step S7-3) The information processing device 40 generates a request for tank internal status information based on a user's operation. (Step S8-3) The information processing device 40 transmits the created request for tank internal status information to the monitoring device 50. (Step S9-3) In the monitoring device 50, the communication device 51 receives the tank interior status information request transmitted by the information processing device 40. The graphic generating unit 54 acquires the tank interior status information request received by the communication device 51. The graphic generating unit 54 acquires pixel data stored in the recording device 52 based on the acquired tank interior status information request, and creates a monitoring image based on the acquired pixel data. (Step S10-3) In the monitoring device 50, the graphic generating unit 54 generates a tank internal status information response addressed to the information processing device 40, which includes information indicating the generated monitoring image. (Step S11-3) In the monitoring device 50, the graphic conversion unit 54 outputs the created tank internal condition information response to the communication device 51. The communication device 51 acquires the tank internal condition information response output by the graphic conversion unit 54, and transmits the acquired tank internal condition information response to the information processing device 40. After step S11-3, the information processing device 40 receives the tank internal status information response transmitted by the monitoring device 50, and acquires information indicating a monitoring image contained in the received tank internal status information response. The information processing device 40 displays the monitoring image by image processing the acquired information indicating the monitoring image. With this configuration, a user of the information processing device 40 can check the internal status of the solid-liquid separation tank.

[0036] In the above-described embodiment, as an example, the ultrasonic sensor 20 is installed in the post-sedimentation tank 17, and the internal state of the post-sedimentation tank 17 is determined. However, the present invention is not limited to this example. For example, the ultrasonic sensor 20 may be installed in the pre-sedimentation tank 11, and the internal state of the pre-sedimentation tank 11 may be determined, or the ultrasonic sensor 20 may be installed in the concentration tank 12, and the internal state of the concentration tank 12 may be determined. In other words, the ultrasonic sensor 20 is installed in at least one of the post-sedimentation tank 17, the pre-sedimentation tank 11, and the concentration tank 12, and the internal state is determined. In the above embodiment, the monitoring system 100 is connected to one sewage treatment facility 10, but the present invention is not limited to this example. For example, the monitoring system 100 may be connected to multiple sewage treatment facilities 10, or multiple monitoring systems 100 may be connected to one sewage treatment facility 10. If the monitoring system 100 is connected to multiple sewage treatment facilities 10, when an unexperienced unsteady state occurs in facility A, if facility B has an experience of the unsteady state occurring, it is highly likely to be determined as an "abnormality" and output. In other words, the monitoring device 50 can learn more, and therefore the number of cases that can be used for determination can be increased. Therefore, the number of unsteady states that can be determined as abnormal or malfunction can be increased. In the above-described embodiment, the monitoring device 50 performs machine learning, but the present invention is not limited to this example. For example, the device performing machine learning may be realized by a device other than the monitoring device 50. In this case, the learning device acquires, from the monitoring device 50, a monitoring image which is an image showing the inside of a solid-liquid separation tank for separating wastewater into solid and liquid, and a diagnosis result based on the monitoring image of the inside of the solid-liquid separation tank. The learning device has a learning unit which generates a learning model of the diagnosis result showing the relationship between the monitoring image and the internal state of the solid-liquid separation tank by machine learning (supervised machine learning) based on the acquired monitoring image and the diagnosis result based on the monitoring image of the internal state of the solid-liquid separation tank. In the above embodiment, a case has been described in which the judgment result of the internal state of the solid-liquid separation tank is judged to be normal, abnormal, or malfunctioning based on the monitoring image, but this is not limited to the example. For example, the judgment result of the internal state of the solid-liquid separation tank may be judged to be normal or abnormal based on the monitoring image, or the judgment result of the internal state of the solid-liquid separation tank may be classified into four or more types based on the monitoring image.

[0037] In the above-described embodiment, when the current status determination unit 55 determines that there has been a change as a result of comparing the created monitoring image with a past monitoring image of a normal state, the current status determination unit 55 may determine the internal status of the solid-liquid separation tank using a learning model of the diagnosis results stored in the recording device 52. In the above-described embodiment, the data processing device 30 may be provided with a display switching operation unit 36 ​​and an image data display unit 37. Fig. 10 is a diagram showing another example of the data processing device according to the present embodiment. An example of the display switching operation unit 36 ​​is configured by a display switching button. An example of the image data display unit 37 is a display. The image data display unit 37 may display the measured data as is, or may display long-term data in a limited display width by thinning out the measured data. By displaying long-term data in a limited display width, it is possible to monitor changes over a long period of time. If a still image is displayed on the image data display unit 37, pixel data can be picked up at any suitable interval and switched between displays. However, if measurements are constantly being taken and new data is being added as in this embodiment, there is a risk of delays or interruptions in data processing when pixel data is picked up and switched between displays, and it would be counterproductive if measurements became unstable due to the image display. For this reason, in this embodiment, several preset display time widths are prepared, a data storage area for each of the multiple display time widths is created, and an interval (interval) for adding new data is specified to create an image database corresponding to each of the multiple display time widths. If the image data display unit 37 is configured with 200 pixels in the vertical direction (= depth direction) and 240 pixels in the horizontal direction (= time series), when data is stored every second, the display data will be 4 minutes long in the horizontal direction, and when data is stored every 10 seconds, the display data will be 40 minutes long in the horizontal direction. By deleting the oldest data at the same time as adding new data, it is possible to operate with a limited data area. In addition, an area is provided to store more than 240 pieces of data necessary for the display width, and by scrolling and displaying the data, it is possible to observe changes from the past as if they were a video. In this embodiment, these two data storage and display functions are possible.

[0038] In the above-described embodiment, the image data storage unit 35 of the data processing device 30 may be accessed from an external terminal to retrieve the data stored in the image data storage unit 35 . In the above-described embodiment, the data calculation unit 34 of the data processing device 30 may be accessed from an external terminal. In this case, the data stored in the data calculation unit 34 may be retrieved and output to the external terminal in response to a command from the external terminal. By configuring in this way, the data output by the data calculation unit 34 can be displayed on the external terminal, making online monitoring possible. In this case, the latest data may be output from the data calculation unit 34 to the external terminal in response to a command from the external terminal. The interval at which data is output can be set by the external terminal. With this configuration, data can be displayed on the external terminal in sequence, enabling remote live image monitoring. Specifically, a unique identification number (=password) assigned to the data calculation unit 34 by dedicated software installed on the external terminal is checked sequentially between the data calculation unit and the external terminal. One-to-one communication is possible between the external terminal and the data calculation unit, preventing eavesdropping of data by installing a signal distributor or the like.

[0039] In the above-described embodiment, the data calculation unit 34 of the data processing device 30 may be set to transmit data to an external terminal. The data calculation unit 34 may transmit data to the external terminal based on the setting for transmitting data. By configuring in this way, the external terminal can be used for remote monitoring of the sewage treatment facility 10. In the above-described embodiment, the data calculation unit 34 of the data processing device 30 may be set to output data to an external data server or recording medium. The data calculation unit 34 outputs data to the external data server or recording medium based on the setting. The external data server creates a database based on the data output by the data calculation unit 34. The external data server may display an image or process data based on the created database. In the above-described embodiment, when the data calculation unit 34 of the data processing device 30 transmits data to an external terminal, the data may be transmitted according to, for example, the RS232C standard. When the data calculation unit 34 of the data processing device 30 transmits data to an external terminal, the data may be transmitted directly between the external terminal, or a signal converter may be provided to convert the data into a transmission protocol such as RS422 or RS485 standard, USB (Universal Serial Bus), LAN, or optical fiber before transmission. Data may be transmitted from the external terminal to a server on a regular or irregular basis. In this case, the central monitoring device may request data from the external terminal (such as a small PC) and have the external terminal output the data. In this case, the external terminal may have the functions of the monitoring device 50 described above. The external terminal may acquire the data output by the data calculation unit 34, judge the internal state of the solid-liquid separation tank from the monitoring images based on the acquired data, and output the judgment result to the server. The server may also have the functions of the current state judgment unit 55 of the monitoring device described above. The external terminal acquires the data output by the data calculation unit 34, and transmits the acquired data to the server. The server acquires the data transmitted by the external device, and judges the internal state of the solid-liquid separation tank from the monitoring images based on the acquired data.

[0040] According to the monitoring system 100 of this embodiment, the monitoring device 50 has a judgment unit serving as a current status judgment unit 55 that judges the internal state of the solid-liquid separation tank from a monitored image showing the inside of the solid-liquid separation tank that is the subject of diagnosis, using a first learning model as a learning model of the diagnostic result that has learned the relationship between the monitored image and the internal state of the solid-liquid separation tank based on a monitored image, which is an image showing the inside of a solid-liquid separation tank for separating wastewater into solid and liquid, and a diagnostic result based on the monitored image of the inside of the solid-liquid separation tank, and an output unit that outputs information identifying the internal state of the solid-liquid separation tank judged by the judgment unit using the monitored image of the solid-liquid separation tank that is the subject of diagnosis and the first learning model. With this configuration, the monitoring device 50 can use the first learning model that has learned the relationship between the monitoring image and the diagnosis result of the inside of the solid-liquid separation tank to determine the internal condition of the solid-liquid separation tank from the monitoring image showing the inside of the solid-liquid separation tank that is the target of diagnosis, thereby making it possible to monitor the internal condition of the solid-liquid separation tank. By being able to determine the internal condition of the solid-liquid separation tank from the monitoring image showing the inside of the solid-liquid separation tank that is the target of diagnosis using the first learning model, human experience is not required and variation in diagnosis results can be reduced compared to when a person diagnoses the inside of the solid-liquid separation tank based on experience. If the system is configured to be self-contained at the site, the dimensions and characteristics of the tank will not change, so past and present can be compared simply, making it easy to distinguish between steady and unsteady states or the occurrence of an abnormality. By acquiring case collections, latest cases, etc. from outside (either online or offline) and updating the acquired case collections, latest cases, etc., when an abnormality is detected, it is possible to determine what the abnormality is and present countermeasures, even if there has been no experience of the abnormality occurring in the past and there is no learning history. By making it possible to refer to an accessible database when an abnormality occurs, it is possible to estimate what the abnormality is and to obtain countermeasures. It also reduces mistakes in making judgments that would not occur at the site (equipment). This reduces the risk of error judgments and the time required for judgments. It also reduces the risk of data hacking and attacks on the system.

[0041] Furthermore, the diagnostic result is generated based on either or both of the accumulated state of solid matter and the floating state of solid matter contained in the monitoring image. By configuring in this manner, the monitoring device 50 can determine the internal state of the solid-liquid separation tank, which is the target of diagnosis, from the monitoring image showing the inside of the solid-liquid separation tank, using the first learning model that has learned the relationship between the monitoring image and the diagnostic result generated based on either or both of the accumulated state of solid matter and the floating state of solid matter contained in the monitoring image. Furthermore, the determination unit determines whether the internal state of the solid-liquid separation tank is normal, malfunction, or abnormal from the monitoring image showing the interior of the solid-liquid separation tank that is the subject of diagnosis. By configuring in this manner, the monitoring device 50 can determine whether the internal state of the solid-liquid separation tank is normal, malfunction, or abnormal from the monitoring image showing the interior of the solid-liquid separation tank that is the subject of diagnosis, using the first learning model that has learned the relationship between the monitoring image and the diagnosis result based on the monitoring image of the interior of the solid-liquid separation tank. Furthermore, the device further includes a notification unit that notifies the user that the internal condition of the solid-liquid separation tank is either in a malfunction or an abnormal state when the determination unit determines that the internal condition of the solid-liquid separation tank is either in a malfunction or an abnormal state. By configuring in this manner, the device can notify the user that the internal condition of the solid-liquid separation tank is either in a malfunction or an abnormal state when the determination unit determines that the internal condition of the solid-liquid separation tank is either in a malfunction or an abnormal state, thereby making it possible to notify the user that some action is required regarding the internal condition of the solid-liquid separation tank.

[0042] [First Modification of the Embodiment] (Surveillance System) 11 is a diagram showing a configuration example of a monitoring system according to a first modification of the embodiment of the present invention. A monitoring system 100a according to the first modification of the embodiment diagnoses a sludge accumulation state in a solid-liquid separation tank such as a settling tank or a thickening tank. In the first modification of the embodiment, as in the embodiment, a sewage treatment facility 10 is applied as an example of a facility equipped with a solid-liquid separation tank.

[0043] (Monitoring system 100a) The monitoring system 100a includes an ultrasonic sensor 20, a data processing device 30, a gateway device 31, an information processing device 40a, a terminal device 45a, and a monitoring device 50a. The gateway device 31, the information processing device 40a, the terminal device 45a, and the monitoring device 50a are connected via a network NW. In the data processing device 30, the data calculation unit 34 transmits the digital signal via the gateway device 31 to the monitoring device 50a. (Monitoring device 50a) The monitoring device 50a is realized by a device such as a personal computer, a server, or an industrial computer. The monitoring device 50a includes a communication device 51, a recording device 52, an information processing device 53a, and a bus line BL such as an address bus or a data bus for electrically connecting the components as shown in FIG. The recording device 52 stores a program (monitoring application) executed by the monitoring device 50a. The recording device 52 also stores pixel data output by the information processing unit 53a. The recording device 52 stores training data of diagnostic results that associates information indicating the monitoring image with the diagnosis results of the internal condition of the solid-liquid separation tank based on the monitoring image, and a learning model of the diagnostic results obtained by machine learning the relationship between the monitoring image and the internal condition of the solid-liquid separation tank based on the training data of the diagnostic results. The recording device 52 stores cause training data that associates information indicating the monitoring image with information identifying the cause that results in the diagnosis of the internal condition of the solid-liquid separation tank based on the monitoring image, and a cause learning model obtained by machine learning the relationship between the monitoring image and the information identifying the cause of the internal condition of the solid-liquid separation tank based on the cause training data.

[0044] Fig. 12 is a diagram showing an example of teacher data. Fig. 12 shows teacher data of diagnosis results and teacher data of causes. The teacher data of diagnosis results is data that associates a monitoring image with a diagnosis result of the internal state of the solid-liquid separation tank based on the monitoring image. The teacher data of causes is data that associates a monitoring image with information that identifies the cause of the diagnosis result of the internal state of the solid-liquid separation tank based on the monitoring image. In the first modification of the embodiment, as an example, each of the multiple monitoring images is associated with one of "normal", "abnormal", and "malfunction" as a diagnosis result, as in the embodiment. Furthermore, each of the multiple monitoring images is associated with an estimated result of the cause of the diagnosis result based on the diagnosis result. In Fig. 12, (1) is diagnosed as normal because the supernatant water and the sludge sediment layer are separated and solid-liquid separation is in good condition. If it is diagnosed as normal, the estimated result of the cause of the diagnosis result is not stored. (2) is diagnosed as an abnormality because the settling property of the sludge has deteriorated and the sludge has floated to the surface. In this case, bulking is stored as an example of an estimated result of the cause of the diagnosis. Bulking refers to the phenomenon in which the settling property of the sludge deteriorates and it becomes difficult to obtain supernatant water. (3) is diagnosed as being out of order because the accumulated sludge is seen to be flying up. In this case, the following are stored as examples of the estimated results of the diagnosis: the sludge input speed is fast, the sludge input amount is large, and the sludge interface is high. Returning to Figure 11, we will continue the explanation.

[0045] The information processing unit 53a functions as, for example, a graphics generating unit 54, a current state determining unit 55a, a learning unit 56a, and a cause determining unit 57. The current state determination unit 55a acquires pixel data stored in the recording device 52, and creates a monitoring image based on the acquired pixel data. The current state determination unit 55a acquires a learning model of the diagnosis results stored in the recording device 52. The current state determination unit 55a judges the internal state of the solid-liquid separation tank in the created monitoring image based on the acquired learning model of the diagnosis results. The learning unit 56a has the following functions in addition to the functions of the learning unit 56. The learning unit 56a acquires a diagnosis result notification received by the communication device 51, and stores in the recording device 52 cause teacher data that associates information indicating a monitoring image included in the acquired diagnosis result notification with an estimated result of a cause that becomes a diagnosis result. The learning unit 56a acquires the cause teacher data stored in the recording device 52. The learning unit 56a generates a cause learning model that associates the monitoring image with information that identifies the cause of the internal state of the solid-liquid separation tank by machine learning (supervised learning) of the monitoring image and the estimated result of the cause that becomes the diagnosis result of the internal state of the solid-liquid separation tank based on the acquired cause teacher data. For example, the learning unit 56a recognizes the monitoring image using a convolutional neural network. The cause learning model classifies the monitoring image into any of information that identifies the cause of the internal state of the solid-liquid separation tank based on the information indicating the monitoring image. The learning unit 56a stores the generated cause learning model in the recording device 52.

[0046] The cause determination unit 57 acquires information showing the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55a. When the acquired determination result of the internal state of the solid-liquid separation tank is malfunction or abnormal, the cause determination unit 57 acquires a learning model of the cause stored in the recording device 52. The cause determination unit 57 judges information specifying the cause of the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the cause. The cause determination unit 57 creates status notification information addressed to the information processing device 40a, including information showing the monitoring image, information showing the internal state of the solid-liquid separation tank, and information showing the determination result of the information specifying the cause of the internal state of the solid-liquid separation tank. The cause determination unit 57 outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the cause determination unit 57 and transmits the acquired status notification information to the information processing device 40a. All or a part of the information processing unit 53a is a functional unit (hereinafter referred to as a software functional unit) that is realized by, for example, a processor such as a CPU executing a program such as a monitoring application stored in the recording device 52. All or a part of the information processing unit 53a may be realized by hardware such as an LSI, an ASIC, or an FPGA, or may be realized by a combination of a software functional unit and hardware. The information processing device 40 can be applied to the information processing device 40a.

[0047] (Terminal device 45a) The terminal device 45a is realized by a device such as a personal computer, a server, an industrial computer, etc. An example of the terminal device 45a is installed in a monitoring center that monitors the sewage treatment facility 10. When diagnosing the internal condition of the solid-liquid separation tank, the user operates the terminal device 45a to create a monitoring image request addressed to the monitoring device 50a, including information requesting a monitoring image. The terminal device 45a creates the monitoring image request based on the user's operation. The terminal device 45a transmits the created monitoring image request to the monitoring device 50a. The terminal device 45a receives the monitoring image response sent by the monitoring device 50a in response to the monitoring image request sent to the monitoring device 50a. The terminal device 45a displays the monitoring image included in the monitoring image response. The user refers to the monitoring image displayed by the terminal device 45a, diagnoses the internal state of the solid-liquid separation tank included in the monitoring image, and further estimates the cause of the internal state of the solid-liquid separation tank. The user operates the terminal device 45a to create a diagnosis result notification addressed to the monitoring device 50a, which includes information showing the monitoring image, a diagnosis result of the internal state of the solid-liquid separation tank, and information identifying the cause of the diagnosis result. The terminal device 45a creates the diagnosis result notification based on the user's operation. The terminal device 45a transmits the created diagnosis result notification to the monitoring device 50a.

[0048] (Operation of monitoring system) Fig. 13 is a diagram showing a first example of the operation of the monitoring system according to the first modified example of the embodiment. Referring to Fig. 13, a process will be described in which monitoring device 50a accumulates the diagnosis result of the internal state of the solid-liquid separation tank included in the diagnosis result notification transmitted by terminal device 45a and information identifying the cause of the diagnosis result, and performs machine learning based on the accumulated diagnosis result of the internal state of the solid-liquid separation tank and the information identifying the cause of the diagnosis result to generate a learning model of the diagnosis result and a learning model of the cause. Steps S1-4 to S10-4 can be implemented by steps S1-1 to S10-1 in FIG. 7, so a description thereof will be omitted here. (Step S11-4) The terminal device 45a receives the monitoring image response transmitted by the monitoring device 50a. The terminal device 45a displays the monitoring image by image processing information indicating the monitoring image included in the received monitoring image response. The terminal device 45a creates a diagnosis result notification addressed to the monitoring device 50a, which includes the diagnosis result of the internal condition of the solid-liquid separation tank and information identifying the cause of the diagnosis result. (Step S12-4) The terminal device 45a transmits the created diagnosis result notification to the monitoring device 50a.

[0049] (Step S13-4) In the monitoring device 50a, the communication device 51 receives the diagnosis result notification transmitted by the terminal device 45a. The learning unit 56a acquires the diagnosis result notification received by the communication device 51, and acquires information indicating the monitoring image included in the acquired diagnosis result notification, the diagnosis result of the internal state of the solid-liquid separation tank, and information identifying the cause of the diagnosis result. The learning unit 56a stores, in the recording device 52, teacher data of a diagnosis result that associates information indicative of the acquired monitoring image with a diagnosis result of the internal state of the solid-liquid separation tank. The learning unit 56a stores, in the recording device 52, teacher data of a cause that associates information indicative of the acquired monitoring image with information identifying the cause of the diagnosis result of the internal state of the solid-liquid separation tank. (Step S14-4) In the monitoring device 50a, the learning unit 56a acquires teacher data of the diagnosis results stored in the recording device 52. The learning unit 56a performs machine learning of the monitoring images and the diagnosis results of the internal state of the solid-liquid separation tank based on the monitoring images, based on the acquired teacher data of the diagnosis results, to generate a learning model of the diagnosis results that associates the monitoring images with the internal state of the solid-liquid separation tank. The learning unit 56a acquires the teacher data of the cause stored in the recording device 52. The learning unit 56a performs machine learning on the monitoring image and information identifying the cause that results in the diagnosis of the internal state of the solid-liquid separation tank based on the acquired teacher data of the cause, thereby generating a learning model of the cause that associates the monitoring image with the information identifying the cause that results in the diagnosis of the internal state of the solid-liquid separation tank. (Step S15-4) In the monitoring device 50a, the learning unit 56a stores the generated learning model of the diagnosis result in the recording device 52. The learning unit 56a stores the generated learning model of the cause in the recording device 52.

[0050] Fig. 14 is a diagram showing a second example of the operation of the monitoring system according to the first modification of the embodiment. Referring to Fig. 14, the monitoring device 50a acquires a digital signal transmitted by the data processing device 30, and creates a monitoring image based on the acquired digital signal. The process of the monitoring device 50a determining the internal state of the solid-liquid separation tank based on the created monitoring image will be described. Steps S1-5 to S6-5 can be implemented by steps S1-1 to S6-1 in FIG. 7, so a description thereof will be omitted here. (Step S7-5) In the monitoring device 50a, the current state determining unit 55a obtains the pixel data stored in the recording device 52, and creates a monitoring image based on the obtained pixel data. (Step S8-5) In the monitoring device 50a, the current state determining unit 55a obtains a learning model of the diagnosis results stored in the recording device 52. (Step S9-5) In the monitoring device 50a, the current state determining unit 55a determines the internal state of the solid-liquid separation tank from the created monitoring image based on a learning model of the acquired diagnosis results. (Step S10-5) In monitoring device 50a, cause determination unit 57 acquires the determination result of the internal state of the solid-liquid separation tank from current state determination unit 55a. Cause determination unit 57 determines whether the acquired determination result of the internal state of the solid-liquid separation tank is a malfunction or an abnormality. If cause determination unit 57 determines that the acquired determination result of the internal state of the solid-liquid separation tank is neither a malfunction nor an abnormality, the process ends.

[0051] (Step S11-5) In the monitoring device 50a, when the cause determination unit 57 determines that the acquired determination result of the internal condition of the solid-liquid separation tank is malfunction or abnormal, it acquires a learning model of the cause stored in the recording device 52. (Step S12-5) In the monitoring device 50a, the cause determination unit 57 determines information that identifies the cause of the internal state of the solid-liquid separation tank in the acquired monitoring image, based on the acquired learning model of the cause. (Step S13-5) In the monitoring device 50a, the cause determination unit 57 creates status notification information addressed to the information processing device 40a, and includes information indicating a monitoring image, information indicating the determination result of the internal condition of the solid-liquid separation tank, and information indicating the determination result of the cause of the internal condition of the solid-liquid separation tank. (Step S14-5) In the monitoring device 50a, the cause determination unit 57 outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the cause determination unit 57, and transmits the acquired status notification information to the information processing device 40a. The process in which the monitoring device 50a transmits information showing a monitoring image based on a request for information on an inside tank condition transmitted by the information processing device 40a can be described with reference to FIG. 9, and therefore a description thereof will be omitted.

[0052] In the first modification of the embodiment described above, the case where the monitoring system 100a is connected to one sewage treatment facility 10 has been described, but the present invention is not limited to this example. For example, the monitoring system 100a may be connected to multiple sewage treatment facilities 10, or multiple monitoring systems 100a may be connected to one sewage treatment facility 10. If the monitoring system 100a is connected to multiple sewage treatment facilities 10, when an unexperienced unsteady state occurs in facility A, if the unsteady state has occurred in facility B, it is highly likely that the unsteady state is judged as an "abnormality" and information identifying the cause of the abnormality is determined and output. In other words, the monitoring device 50a is capable of learning more, and therefore the number of cases that can be used for judgment can be increased. Therefore, the number of unsteady states that can be judged as abnormal or malfunction can be increased. In the first modified example of the embodiment described above, the monitoring device 50a performs machine learning, but the present invention is not limited to this example. For example, the device performing machine learning may be realized by a device other than the monitoring device 50a. In this case, in the learning device described in the embodiment, the learning device acquires the monitoring image and information for identifying the cause of the diagnosis result based on the monitoring image from the monitoring device 50a. The learning unit of the learning device generates a second learning model representing the relationship between the monitoring image and the information for identifying the cause of the internal state of the solid-liquid separation tank by machine learning (supervised machine learning) based on the monitoring image and the information for identifying the cause of the diagnosis result based on the monitoring image. In the first modified example of the embodiment described above, a case has been described in which the result of the determination of the internal state of the solid-liquid separation tank is determined to be normal, abnormal, or malfunction based on the monitoring image, and further, based on whether the result of the determination of the internal state of the solid-liquid separation tank is abnormal or malfunction, the facts such as bulking, a high sludge input rate, a large amount of sludge input, and a high sludge interface are stored, but this is not limited to the example. For example, based on whether the result of the determination of the internal state of the solid-liquid separation tank is abnormal or malfunction, the cause may be classified into one or more causes.

[0053] According to the monitoring system 100a of the first modified embodiment, the monitoring device 50a includes a cause determination unit 57 that determines information identifying a cause of the internal state of the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, using a second learning model as a learning model of causes that has learned the relationship between the monitoring image and the information identifying a cause that results in a diagnosis result based on the monitoring image and the information identifying a cause that results in a diagnosis result of the inside of the solid-liquid separation tank, based on the monitoring image and the information identifying a cause that results in a diagnosis result based on the monitoring image. The output unit further outputs information identifying the cause of the internal state of the solid-liquid separation tank determined by the cause determination unit using the monitoring image of the solid-liquid separation tank that is the target of diagnosis and the second learning model. With this configuration, the monitoring device 50a can use the second learning model that has learned the relationship between the monitoring image and information identifying the cause of the diagnosis result inside the solid-liquid separation tank to determine information identifying the cause of the internal state of the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, thereby making it possible to monitor the causes of the internal state of the solid-liquid separation tank. By being able to determine the cause of the internal state of the solid-liquid separation tank from the monitoring image showing the interior of the solid-liquid separation tank that is the target of diagnosis using the second learning model, human experience is not required and variation in the diagnosis results can be reduced compared to when a person diagnoses the cause of the internal state of the solid-liquid separation tank based on experience.

[0054] [Modification 2 of the embodiment] (Surveillance System) 15 is a diagram showing a configuration example of a monitoring system according to a second modification of the embodiment of the present invention. A monitoring system 100b according to the second modification of the embodiment diagnoses a sludge accumulation state in a solid-liquid separation tank such as a settling tank or a thickening tank. In the second modification of the embodiment, as in the embodiment, a sewage treatment facility 10 is applied as an example of a facility equipped with a solid-liquid separation tank.

[0055] (Monitoring system 100b) The monitoring system 100b includes an ultrasonic sensor 20, a data processing device 30, a gateway device 31, an information processing device 40b, a terminal device 45b, and a monitoring device 50b. The gateway device 31, the information processing device 40b, the terminal device 45b, and the monitoring device 50b are connected via a network NW. In the data processing device 30, the data calculation unit 34 transmits the digital signal via the gateway device 31 to the monitoring device 50b.

[0056] (Monitoring device 50b) The monitoring device 50b is realized by a device such as a personal computer, a server, or an industrial computer. The monitoring device 50b includes a communication device 51, a recording device 52, an information processing unit 53b, and a bus line BL such as an address bus or a data bus for electrically connecting the components as shown in FIG. The recording device 52 stores a program (monitoring application) executed by the monitoring device 50b. The recording device 52 also stores pixel data output by the information processing unit 53b. The recording device 52 stores training data of diagnostic results that associates information indicating the monitoring image with the diagnostic results of the interior of the solid-liquid separation tank based on the monitoring image, and a learning model of the diagnostic results obtained by machine learning the relationship between the monitoring image and the internal condition of the solid-liquid separation tank based on the training data of the diagnostic results. The recording device 52 stores cause training data that associates information indicating the monitoring image with information identifying the cause of the diagnosis result of the interior of the solid-liquid separation tank based on the monitoring image, and a cause learning model obtained by machine learning the relationship between the monitoring image and the information identifying the cause of the internal condition of the solid-liquid separation tank based on the cause training data. The recording device 52 stores training data on countermeasures that associates information indicating a monitoring image with information identifying a method of dealing with the diagnosis results of the solid-liquid separation tank based on the monitoring image, and a learning model of the countermeasures obtained by machine learning the relationship between the monitoring image and information identifying a method of dealing with the internal condition of the solid-liquid separation tank based on the training data on the countermeasures.

[0057] FIG. 16 is a diagram showing an example of teacher data. FIG. 16 shows teacher data of diagnosis results, teacher data of causes, and teacher data of countermeasures. The teacher data of diagnosis results is data that associates a monitoring image with a diagnosis result of the internal state of the solid-liquid separation tank based on the monitoring image. The teacher data of causes is data that associates a monitoring image with information that identifies the cause of the diagnosis result of the internal state of the solid-liquid separation tank based on the monitoring image. The teacher data of countermeasures is data that associates a monitoring image with information that identifies a countermeasure for the internal state of the solid-liquid separation tank based on the monitoring image. In the second modification of the embodiment, as an example, each of the multiple monitoring images is associated with one of "normal", "abnormal" and "malfunction" as a diagnostic result, as in the embodiment. Furthermore, each of the multiple monitoring images is associated with an estimated result of the cause of the diagnostic result, based on the diagnostic result. Furthermore, each of the multiple monitoring images is associated with information that specifies a method of dealing with the diagnostic result, based on the diagnostic result. In Fig. 16, (1) is diagnosed as normal because the supernatant water and the sludge sediment layer are separated and solid-liquid separation is in good condition. If it is diagnosed as normal, the estimated result of the diagnosis and the countermeasure are not stored. (2) is diagnosed as an abnormality because the settling property of the sludge has deteriorated and the sludge has floated to the surface. In this case, bulking is stored as the estimated result of the diagnosis. Furthermore, promotion of sludge extraction and addition of a sludge settling agent, etc. are stored as countermeasures for the estimated result of the diagnosis. (3) is diagnosed as malfunctioning because the accumulated sludge is seen to be flying up. In this case, a high sludge input speed, a large amount of sludge input, and a high sludge interface are stored as examples of inferred causes that will result in a diagnosis. Furthermore, slowing down the input speed, reducing the input amount, and promoting sludge extraction are stored as examples of ways to deal with the inferred causes that will result in a diagnosis. Returning to FIG. 15, the explanation will continue.

[0058] The information processing unit 53b functions as, for example, a graphic generating unit 54, a current state determining unit 55a, a learning unit 56b, a cause determining unit 57b, and a countermeasure determining unit 58. The learning unit 56b has the following functions in addition to the functions of the learning unit 56a: The learning unit 56b acquires a diagnosis result notification received by the communication device 51, and stores in the recording device 52 teacher data on a countermeasure method in which information indicating a monitoring image included in the acquired diagnosis result notification is associated with information specifying a countermeasure method for the diagnosis result. The learning unit 56b acquires the teacher data of the countermeasures stored in the recording device 52. The learning unit 56b generates a learning model of the countermeasures that associates the monitoring image with the countermeasures for the internal state of the solid-liquid separation tank by performing machine learning (supervised learning) on ​​the monitoring image and information that specifies the countermeasures for the diagnosis result of the internal state of the solid-liquid separation tank based on the acquired teacher data of the countermeasures. For example, the learning unit 56b recognizes the monitoring image using a convolutional neural network. The learning model of the countermeasures classifies the monitoring image into any of the information that specifies the countermeasures for the internal state of the solid-liquid separation tank based on the information indicating the monitoring image. The learning unit 56b stores the generated learning model of the countermeasures in the recording device 52.

[0059] The cause determination unit 57b acquires information indicating the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55a. If the acquired determination result of the internal state of the solid-liquid separation tank is malfunction or abnormal, the cause determination unit 57b acquires a learning model of the cause stored in the recording device 52. The cause determination unit 57b determines the cause of the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the cause. The countermeasure determination unit 58 acquires information indicating the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55a. When the acquired determination result of the internal state of the solid-liquid separation tank is a malfunction or abnormality, the countermeasure determination unit 58 acquires a learning model of the countermeasure stored in the recording device 52. The countermeasure determination unit 58 determines a countermeasure for the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the countermeasure. The countermeasure determination unit 58 acquires information identifying the cause of the internal state of the solid-liquid separation tank in the monitoring image from the cause determination unit 57b. The countermeasure determination unit 58 creates status notification information addressed to the information processing device 40b, the status notification information including information showing the monitoring image, information identifying the cause of the internal state of the solid-liquid separation tank, and information identifying a countermeasure for the internal state of the solid-liquid separation tank. The countermeasure determination unit 58 outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the cause determination unit 57b, and transmits the acquired status notification information to the information processing device 40b. All or a part of the information processing unit 53b is a functional unit (hereinafter referred to as a software functional unit) that is realized by, for example, a processor such as a CPU executing a program such as a monitoring application stored in the recording device 52. All or a part of the information processing unit 53b may be realized by hardware such as an LSI, an ASIC, or an FPGA, or may be realized by a combination of a software functional unit and hardware. The information processing device 40b can be implemented as the information processing device 40.

[0060] (Terminal device 45b) The terminal device 45b is realized by a device such as a personal computer, a server, an industrial computer, etc. An example of the terminal device 45b is installed in a monitoring center that monitors the sewage treatment facility 10. When diagnosing the internal condition of the solid-liquid separation tank, the user operates the terminal device 45b to create a monitoring image request addressed to the monitoring device 50b, including information requesting a monitoring image. The terminal device 45b creates the monitoring image request based on the user's operation. The terminal device 45b transmits the created monitoring image request to the monitoring device 50b. The terminal device 45b receives the monitoring image response sent by the monitoring device 50b in response to the monitoring image request sent to the monitoring device 50b. The terminal device 45b displays the monitoring image included in the monitoring image response. The user refers to the monitoring image displayed by the terminal device 45b, diagnoses the internal state of the solid-liquid separation tank included in the monitoring image, and further estimates the cause of the internal state of the solid-liquid separation tank. The user operates the terminal device 45b to create a diagnosis result notification addressed to the monitoring device 50b, which includes information showing the monitoring image, a diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, and information identifying a method of dealing with the diagnosis result. The terminal device 45b creates the diagnosis result notification based on the user's operation. The terminal device 45b transmits the created diagnosis result notification to the monitoring device 50b.

[0061] (Operation of monitoring system) Fig. 17 is a diagram showing a first example of the operation of the monitoring system according to the second modification of the embodiment. With reference to Fig. 17, a process in which monitoring device 50b accumulates the diagnosis result of the internal state of the solid-liquid separation tank included in the diagnosis result notification transmitted by terminal device 45b, information identifying the cause of the diagnosis result, and information identifying a method of dealing with the diagnosis result will be described. A process in which monitoring device 50b performs machine learning based on the accumulated diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, and information identifying a method of dealing with the diagnosis result, and generates a learning model of the diagnosis result, a learning model of the cause, and a learning model of the method of dealing with the diagnosis result will be described. Steps S1-6 to S10-6 can be implemented by steps S1-1 to S10-1 in FIG. 7, so a description thereof will be omitted here. (Step S11-6) The terminal device 45b receives the monitoring image response transmitted by the monitoring device 50b. The terminal device 45b displays the monitoring image by performing image processing on the information indicating the monitoring image contained in the received monitoring image response. The terminal device 45b creates a diagnosis result notification addressed to the monitoring device 50b, which includes the diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, and information identifying a method of dealing with the diagnosis result. (Step S12-6) The terminal device 45b transmits the created diagnosis result notification to the monitoring device 50b.

[0062] (Step S13-6) In the monitoring device 50b, the communication device 51 receives the diagnosis result notification transmitted by the terminal device 45b. The learning unit 56b acquires the diagnosis result notification received by the communication device 51, and acquires information indicating the monitoring image included in the acquired diagnosis result notification, the diagnosis result of the internal state of the solid-liquid separation tank, and information specifying a method of dealing with the diagnosis result. The learning unit 56b stores in the recording device 52 the following teacher data of diagnosis results that associates information indicative of the acquired monitoring image with the diagnosis result of the internal condition of the solid-liquid separation tank, teacher data of causes that associates information indicative of the monitoring image with information identifying the cause of the diagnosis result of the internal condition of the solid-liquid separation tank, and teacher data of countermeasures that associates information indicative of the monitoring image with information identifying the method of dealing with the diagnosis result. (Step S14-6) In the monitoring device 50b, the learning unit 56b acquires teacher data of the diagnosis results stored in the recording device 52. The learning unit 56b performs machine learning of the monitoring images and the diagnosis results of the internal state of the solid-liquid separation tank based on the monitoring images, based on the acquired teacher data of the diagnosis results, to generate a learning model of the diagnosis results that associates the monitoring images with the internal state of the solid-liquid separation tank. The learning unit 56b acquires the teacher data of the cause stored in the recording device 52. The learning unit 56b performs machine learning on the monitoring image and information identifying the cause that results in the diagnosis of the internal state of the solid-liquid separation tank based on the acquired teacher data of the cause, thereby generating a learning model of the cause that associates the monitoring image with the information identifying the cause that results in the diagnosis of the internal state of the solid-liquid separation tank. The learning unit 56b acquires the teacher data of the countermeasures stored in the recording device 52. The learning unit 56b performs machine learning on the monitoring image and information specifying a countermeasure for the internal state of the solid-liquid separation tank based on the acquired teacher data of the countermeasures, thereby generating a learning model of the countermeasures that associates the monitoring image with the information specifying a countermeasure for the internal state of the solid-liquid separation tank. (Step S15-6) In the monitoring device 50b, the learning unit 56b stores in the recording device 52 the generated learning model of the diagnosis result, the learning model of the cause, and the learning model of the countermeasure method.

[0063] Fig. 18 is a diagram showing a second example of the operation of the monitoring system according to the second modification of the embodiment. With reference to Fig. 18, the monitoring device 50b acquires a digital signal transmitted by the data processing device 30, and creates a monitoring image based on the acquired digital signal. A process of determining the internal state of the solid-liquid separation tank by the monitoring device 50b based on the created monitoring image will be described. Steps S1-7 to S6-7 can be implemented by steps S1-1 to S6-1 in FIG. 7, so a description thereof will be omitted here. (Step S7-7) In the monitoring device 50b, a current state determining unit 55a obtains the pixel data stored in the recording device 52, and creates a monitoring image based on the obtained pixel data. (Step S8-7) In the monitoring device 50b, the current state determining unit 55a obtains a learning model of the diagnosis results stored in the recording device 52. (Step S9-7) In the monitoring device 50b, the current state determining unit 55a determines the internal state of the solid-liquid separation tank from the created monitoring image based on a learning model of the acquired diagnosis results. (Step S10-7) In monitoring device 50b, cause determination unit 57b acquires the determination result of the internal state of the solid-liquid separation tank from current state determination unit 55a. Cause determination unit 57b determines whether the acquired determination result of the internal state of the solid-liquid separation tank is a malfunction or an abnormality. If cause determination unit 57b determines that the acquired determination result of the internal state of the solid-liquid separation tank is neither a malfunction nor an abnormality, the process ends.

[0064] (Step S11-7) In the monitoring device 50b, when the cause determination unit 57b determines that the acquired determination result of the internal condition of the solid-liquid separation tank is malfunction or abnormal, it acquires a learning model of the cause stored in the recording device 52. (Step S12-7) In the monitoring device 50b, the cause determining unit 57b determines the cause of the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the cause. (Step S13-7) In monitoring device 50b, countermeasure determination unit 58 acquires information indicating the monitored image and the judgment result of the internal state of the solid-liquid separation tank from current state determination unit 55a. If the acquired judgment result of the internal state of the solid-liquid separation tank is malfunction or abnormal, countermeasure determination unit 58 acquires a learning model of the countermeasure stored in recording device 52. (Step S14-7) In the monitoring device 50b, the countermeasure determination unit 58 determines a countermeasure to the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the countermeasure. (Step S15-7) In monitoring device 50b, countermeasure determination unit 58 acquires information identifying the cause of the internal state of the solid-liquid separation tank in the monitoring image from cause determination unit 57b. Countermeasure determination unit 58 creates status notification information addressed to information processing device 40b, which includes information showing the monitoring image, information identifying the cause of the internal state of the solid-liquid separation tank, and information identifying a countermeasure for the internal state of the solid-liquid separation tank. (Step S16-7) In the monitoring device 50b, the countermeasure determination unit 58 outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the countermeasure determination unit 58, and transmits the acquired status notification information to the information processing device 40b. The process in which the monitoring device 50b transmits information showing a monitoring image based on a request for information on an inside of the tank transmitted by the information processing device 40b can be described with reference to FIG. 9, and therefore a description thereof will be omitted.

[0065] In the above-described second modification of the embodiment, the case where the monitoring system 100b is connected to one sewage treatment facility 10 has been described, but the present invention is not limited to this example. For example, the monitoring system 100b may be connected to multiple sewage treatment facilities 10, or multiple monitoring systems 100b may be connected to one sewage treatment facility 10. If the monitoring system 100b is connected to multiple sewage treatment facilities 10, when an unexperienced unsteady state occurs in facility A, if the unsteady state has occurred in facility B, it is highly likely that the unsteady state is judged as an "abnormality" and information specifying the cause of the abnormality and information specifying a countermeasure are determined and output. In other words, the monitoring device 50b can learn more, and therefore the number of cases that can be used for judgment can be increased. Therefore, the number of unsteady states that can be judged as abnormal or malfunction can be increased. In the above-described second modification of the embodiment, the monitoring device 50b performs machine learning, but the present invention is not limited to this example. For example, the device performing machine learning may be realized by a device other than the monitoring device 50b. In this case, the learning device, in the learning device described in the first modification of the embodiment, acquires the monitoring image and information specifying a method of dealing with the diagnosis result based on the monitoring image from the monitoring device 50b. The learning unit of the learning device generates a third learning model by machine learning (supervised machine learning) that represents the relationship between the monitoring image and the information specifying a method of dealing with the diagnosis result of the inside of the solid-liquid separation tank, based on the monitoring image and the information specifying a method of dealing with the diagnosis result based on the monitoring image. In the second variant of the embodiment, the information processing device 40b may notify the operator of the sewage treatment equipment 10 of the countermeasure method included in the status notification, or may create control information for causing the equipment control device 19 to execute the countermeasure method and transmit the created control information to the equipment control device 19. In the second modification of the embodiment described above, a case has been described in which the result of the determination of the internal state of the solid-liquid separation tank is determined based on the monitoring image as being normal, abnormal, or malfunctioning, and further, depending on whether the result of the determination of the internal state of the solid-liquid separation tank is abnormal or malfunctioning, promotion of sludge removal, addition of a sludge settling agent or the like, slowing down the addition rate, reducing the amount of input, and promotion of sludge removal are stored, but this is not limited to the example. For example, depending on whether the result of the determination of the internal state of the solid-liquid separation tank is abnormal or malfunctioning, the countermeasures may be classified into one or more methods. In the above-described second modification of the embodiment, the first modification of the embodiment further includes a process for determining information for identifying a method for dealing with the internal state of the solid-liquid separation tank from a monitoring image of the solid-liquid separation tank, but the present invention is not limited to this example. For example, the embodiment may further include a process for determining information for identifying a method for dealing with the internal state of the solid-liquid separation tank from a monitoring image of the solid-liquid separation tank.

[0066] According to a monitoring system 100b according to a second modification of the embodiment, the monitoring device 50b includes a countermeasure determination unit 58 that determines information specifying a countermeasure for the internal state of the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, using a third learning model as a learning model of countermeasures that has learned the relationship between the monitoring image and the information specifying a countermeasure for the diagnosis result of the inside of the solid-liquid separation tank, based on the monitoring image and information specifying a countermeasure for the diagnosis result based on the monitoring image. The output unit further outputs information specifying a countermeasure for the internal state of the solid-liquid separation tank that is determined by the countermeasure determination unit 58 using the monitoring image of the solid-liquid separation tank that is the target of diagnosis and the third learning model. With this configuration, the monitoring device 50b can use the third learning model that has learned the relationship between the monitoring image and information specifying how to deal with the diagnosis result of the inside of the solid-liquid separation tank to determine information specifying how to deal with the internal condition of the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis, and can therefore monitor how to deal with the internal condition of the solid-liquid separation tank. By being able to determine how to deal with the internal condition of the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank that is the target of diagnosis using the third learning model, human experience is not required and variation in the diagnosis results can be reduced compared to when a person diagnoses how to deal with the internal condition of the solid-liquid separation tank based on experience.

[0067] [Modification 3 of the embodiment] (Surveillance System) 19 is a diagram showing a configuration example of a monitoring system according to a third modification of the embodiment of the present invention. The monitoring system 100c according to the third modification of the embodiment not only diagnoses the sludge deposition state in a solid-liquid separation tank such as a settling tank or a thickening tank, but also detects signs of change. In the third modification of the embodiment, as in the embodiment, a sewage treatment facility 10 is applied as an example of a facility equipped with a solid-liquid separation tank.

[0068] (Monitoring system 100c) The monitoring system 100c includes an ultrasonic sensor 20, a data processing device 30, a gateway device 31, an information processing device 40c, a terminal device 45c, and a monitoring device 50c. The gateway device 31, the information processing device 40c, the terminal device 45c, and the monitoring device 50c are connected via a network NW. In the data processing device 30, the data calculation unit 34 transmits the digital signal via the gateway device 31 to the monitoring device 50c.

[0069] (Monitoring device 50c) The monitoring device 50c is realized by a device such as a personal computer, a server, or an industrial computer. The monitoring device 50c includes a communication device 51, a recording device 52, an information processing unit 53c, and a bus line BL such as an address bus or a data bus for electrically connecting the components as shown in FIG. The recording device 52 stores a program (monitoring application) executed by the monitoring device 50c. The recording device 52 also stores pixel data output by the information processing unit 53c. The recording device 52 stores training data of diagnostic results that associates information indicating the monitoring image with the diagnosis results of the internal condition of the solid-liquid separation tank based on the monitoring image, and a learning model of the diagnostic results obtained by machine learning the relationship between the monitoring image and the internal condition of the solid-liquid separation tank based on the training data of the diagnostic results. The recording device 52 stores cause training data that associates information indicating the monitoring image with information identifying the cause that results in the diagnosis of the internal condition of the solid-liquid separation tank based on the monitoring image, and a cause learning model obtained by machine learning the relationship between the monitoring image and the information identifying the cause of the internal condition of the solid-liquid separation tank based on the cause training data. The recording device 52 stores training data on countermeasures that associates information indicating a monitoring image with information identifying a method of dealing with the diagnosis results of the internal condition of the solid-liquid separation tank based on the monitoring image, and a learning model of the countermeasures obtained by machine learning the relationship between the monitoring image and the information identifying a method of dealing with the internal condition of the solid-liquid separation tank based on the training data on the countermeasures. The recording device 52 stores change training data that associates information indicating a monitoring image with information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, and a change learning model obtained by machine learning the relationship between the monitoring image and the information identifying the change in the internal state of the solid-liquid separation tank based on the change training data.

[0070] The information processing unit 53c functions as, for example, a graphic generating unit 54, a current state determining unit 55a, a learning unit 56c, a cause determining unit 57b, a countermeasure determining unit 58c, and a change sign deriving unit 59. The learning unit 56c has the following functions in addition to the functions of the learning unit 56b. The learning unit 56c acquires the teacher data of the change stored in the recording device 52. The learning unit 56c generates a learning model of the change that associates the monitoring image with the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, by performing machine learning (supervised learning) on ​​the monitoring image and the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, based on the acquired teacher data of the change. For example, the learning unit 56c recognizes the monitoring image using a convolutional neural network. The learning model of the change classifies the monitoring image into any of the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, based on the information indicating the monitoring image. The learning unit 56c stores the generated learning model of the change in the recording device 52. The countermeasure determination unit 58c acquires information indicating the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55a. When the acquired determination result of the internal state of the solid-liquid separation tank is a malfunction or abnormality, the countermeasure determination unit 58c acquires a learning model of the countermeasure stored in the recording device 52. The countermeasure determination unit 58c determines a countermeasure for the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the countermeasure. The change sign deriving unit 59 acquires information indicating a monitoring image from the current state determining unit 55a. The change sign deriving unit 59 acquires a learning model of change stored in the recording device 52. The change sign deriving unit 59 derives a sign of change in the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of change. All or a part of the information processing unit 53c is a functional unit (hereinafter referred to as a software functional unit) that is realized by, for example, a processor such as a CPU executing a program such as a monitoring application stored in the recording device 52. All or a part of the information processing unit 53c may be realized by hardware such as an LSI, an ASIC, or an FPGA, or may be realized by a combination of a software functional unit and hardware. The information processing device 40c can be implemented as the information processing device 40.

[0071] (Terminal device 45c) The terminal device 45c is realized by a device such as a personal computer, a server, an industrial computer, etc. An example of the terminal device 45c is installed in a monitoring center that monitors the sewage treatment facility 10. When diagnosing the internal condition of the solid-liquid separation tank, the user operates the terminal device 45c to create a monitoring image request including information requesting a monitoring image and addressed to the monitoring device 50c. The terminal device 45c creates the monitoring image request based on the user's operation. The terminal device 45c transmits the created monitoring image request to the monitoring device 50c. The terminal device 45c receives the monitoring image response sent by the monitoring device 50c in response to the monitoring image request sent to the monitoring device 50c. The terminal device 45c displays the monitoring image included in the monitoring image response. The user refers to the monitoring image displayed by the terminal device 45c, diagnoses the internal state of the solid-liquid separation tank included in the monitoring image, estimates the cause of the internal state of the solid-liquid separation tank, identifies how to deal with the internal state of the solid-liquid separation tank, predicts the internal state of the solid-liquid separation tank after the monitoring image is obtained, and identifies any changes therein. The user operates the terminal device 45c to create a diagnosis result notification addressed to the monitoring device 50c, the diagnosis result including information showing the monitoring image, the diagnosis result of the internal state of the solid-liquid separation tank, information specifying the cause of the diagnosis result, information specifying the method of dealing with the internal state of the solid-liquid separation tank, and information specifying the change in the internal state of the solid-liquid separation tank after the monitoring image was obtained. The terminal device 45c creates the diagnosis result notification based on the user's operation. The terminal device 45c transmits the created diagnosis result notification to the monitoring device 50c.

[0072] (Operation of monitoring system) Fig. 20 is a diagram showing a first example of the operation of the monitoring system according to the third modified example of the embodiment. Referring to Fig. 20, a process will be described in which monitoring device 50c accumulates information indicating a monitoring image included in a diagnosis result notification transmitted by terminal device 45c, a diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, information identifying a method for dealing with the diagnosis result, and information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, and performs machine learning based on the accumulated information indicating the monitoring image, the diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, information identifying a method for dealing with the diagnosis result, and information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, to generate a learning model of the diagnosis result, a learning model of the cause, a learning model of the method for dealing with the diagnosis, and a learning model of the change. Steps S1-8 to S10-8 can be implemented by steps S1-1 to S10-1 in FIG. 7, so a description thereof will be omitted here. (Step S11-8) The terminal device 45c receives the monitoring image response transmitted by the monitoring device 50c. The terminal device 45c displays the monitoring image by image processing information indicating the monitoring image included in the received monitoring image response. The terminal device 45c creates a diagnosis result notification addressed to the monitoring device 50c, which includes information indicating the monitoring image, a diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, information identifying a method of dealing with the diagnosis result, and information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image was obtained.

[0073] (Step S12-8) The terminal device 45c transmits the created diagnosis result notification to the monitoring device 50c. (Step S13-8) In the monitoring device 50c, the communication device 51 receives the diagnosis result notification transmitted by the terminal device 45c. The learning unit 56c acquires the diagnosis result notification received by the communication device 51, and acquires information indicating the monitoring image included in the acquired diagnosis result notification, the diagnosis result of the internal state of the solid-liquid separation tank, information specifying how to deal with the diagnosis result, and information specifying a change in the internal state of the solid-liquid separation tank after the monitoring image was obtained. The learning unit 56c stores in the recording device 52 the following teacher data on diagnosis results that associates information indicative of the acquired monitoring image with the diagnosis result of the internal condition of the solid-liquid separation tank; teacher data on causes that associates information indicative of the monitoring image with information identifying the cause of the diagnosis result of the internal condition of the solid-liquid separation tank; teacher data on countermeasures that associates information indicative of the monitoring image with information identifying the method of dealing with the diagnosis result; and teacher data on changes that associates information indicative of the monitoring image with information identifying the change in the internal condition of the solid-liquid separation tank after the monitoring image was obtained. (Step S14-8) In the monitoring device 50c, the learning unit 56c acquires teacher data of the diagnosis results stored in the recording device 52. The learning unit 56c performs machine learning of the monitoring images and the diagnosis results of the internal state of the solid-liquid separation tank based on the monitoring images, based on the acquired teacher data of the diagnosis results, to generate a learning model of the diagnosis results that associates the monitoring images with the internal state of the solid-liquid separation tank. The learning unit 56c acquires the teacher data of the cause stored in the recording device 52. The learning unit 56c performs machine learning on the monitoring image and information identifying the cause that results in the diagnosis of the internal state of the solid-liquid separation tank based on the acquired teacher data of the cause, thereby generating a learning model of the cause that associates the monitoring image with the information identifying the cause that results in the diagnosis of the internal state of the solid-liquid separation tank. The learning unit 56c acquires the teacher data of the countermeasures stored in the recording device 52. The learning unit 56c performs machine learning on the monitoring image and information specifying a countermeasure for the internal state of the solid-liquid separation tank based on the acquired teacher data of the countermeasures, thereby generating a learning model of the countermeasures that associates the monitoring image with the information specifying a countermeasure for the internal state of the solid-liquid separation tank. The learning unit 56c acquires the teacher data of the change stored in the recording device 52. The learning unit 56c generates a learning model of the change that associates the monitoring image with the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, by machine learning the monitoring image and the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, based on the acquired teacher data of the change.

[0074] (Step S15-8) In the monitoring device 50c, the learning unit 56c stores in the recording device 52 the generated learning model of the diagnosis result, the learning model of the cause, the learning model of the countermeasure method, and the learning model of the change. Fig. 21 is a diagram showing a second example of the operation of the monitoring system according to the third modification of the embodiment. With reference to Fig. 21, the monitoring device 50c acquires a digital signal transmitted by the data processing device 30, and creates a monitoring image based on the acquired digital signal. The process of determining the internal state of the solid-liquid separation tank based on the created monitoring image by the monitoring device 50c will be described. Steps S1-9 to S6-9 can be implemented by steps S1-1 to S6-1 in FIG. 7, so a description thereof will be omitted here. (Step S7-9) In the monitoring device 50c, a current state determining unit 55a obtains the pixel data stored in the recording device 52, and creates a monitoring image based on the obtained pixel data. (Step S8-9) In the monitoring device 50c, the current state determination unit 55a acquires a learning model of the diagnosis results stored in the recording device 52. (Step S9-9) In the monitoring device 50c, a current state determining unit 55a determines the internal state of the solid-liquid separation tank from the created monitoring image based on a learning model of the acquired diagnosis results. (Step S10-9) In monitoring device 50c, cause determination unit 57b acquires the determination result of the internal state of the solid-liquid separation tank from current state determination unit 55a. Cause determination unit 57b determines whether the acquired determination result of the internal state of the solid-liquid separation tank is a malfunction or an abnormality. If cause determination unit 57b determines that the acquired determination result of the internal state of the solid-liquid separation tank is neither a malfunction nor an abnormality, the process proceeds to step S15-9. (Step S11-9) In the monitoring device 50c, when the cause determination unit 57b determines that the acquired determination result of the internal condition of the solid-liquid separation tank is malfunction or abnormal, it acquires a learning model of the cause stored in the recording device 52. (Step S12-9) In the monitoring device 50c, the cause determining unit 57b determines the cause of the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the cause. (Step S13-9) In the monitoring device 50c, the countermeasure determination unit 58c acquires information indicating the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55a. If the acquired determination result of the internal state of the solid-liquid separation tank is malfunction or abnormal, the countermeasure determination unit 58c acquires a learning model of the countermeasure stored in the recording device 52.

[0075] (Step S14-9) In the monitoring device 50c, a countermeasure determination unit 58c determines a countermeasure to the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the countermeasure. (Step S15-9) In the monitoring device 50c, the change sign deriving unit 59 acquires the learning model of change stored in the recording device 52. In the monitoring device 50c, the change sign deriving unit 59 detects a sign of change in the internal state of the solid-liquid separation tank in the acquired monitoring image, based on the acquired learning model of change. (Step S16-9) In the monitoring device 50c, the change sign deriving unit 59 obtains information indicative of the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55a. The change sign deriving unit 59 obtains information identifying the cause of the internal state of the solid-liquid separation tank in the monitoring image from the cause determination unit 57b. The change sign deriving unit 59 creates status notification information addressed to the information processing device 40c, which includes information indicative of the monitoring image, the determination result of the internal state of the solid-liquid separation tank in the image, information identifying the cause of the internal state of the solid-liquid separation tank, information identifying a method of dealing with the internal state of the solid-liquid separation tank, and the detection result of a sign of change in the internal state of the solid-liquid separation tank. (Step S17-9) In the monitoring device 50c, the change sign deriving unit 59 outputs the created state notification information to the communication device 51. The communication device 51 acquires the state notification information output by the countermeasure method determining unit 58c, and transmits the acquired state notification information to the information processing device 40c. The process in which the monitoring device 50c transmits information showing a monitoring image based on a request for information on an inside of the tank transmitted by the information processing device 40c can be described with reference to FIG. 9, and therefore a description thereof will be omitted.

[0076] In the third modification of the embodiment described above, the case where the monitoring system 100c is connected to one sewage treatment facility 10 has been described, but the present invention is not limited to this example. For example, the monitoring system 100c may be connected to multiple sewage treatment facilities 10, or multiple monitoring systems 100c may be connected to one sewage treatment facility 10. If the monitoring system 100c is connected to multiple sewage treatment facilities 10, when an unexperienced unsteady state occurs in facility A, if the unsteady state has occurred in facility B, the unsteady state is judged as an "abnormality", and information specifying the cause of the abnormality, information specifying a countermeasure, and information specifying a change in the state inside the solid-liquid separation tank after the monitoring image is obtained are likely to be determined and output. In other words, the monitoring device 50c is capable of learning more, and therefore the number of cases that can be used for judgment can be increased. Therefore, the number of unsteady states that can be judged as abnormal or malfunction can be increased. In the above-described third modification of the embodiment, the monitoring device 50c performs machine learning, but the present invention is not limited to this example. For example, the device performing machine learning may be realized by a device other than the monitoring device 50c. In this case, the learning device, in the learning device described in the second modification of the embodiment, acquires a monitoring image and information specifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained from the monitoring device 50c. The learning unit of the learning device generates a fourth learning model expressing a relationship between the monitoring image and the information specifying a change in the internal state of the solid-liquid separation tank by machine learning (supervised machine learning) based on the monitoring image and the information specifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained. In the third modification of the embodiment, the information processing device 40c may notify an operator of the sewage treatment facility 10 of information that specifies a change in the internal state of the solid-liquid separation tank, which is included in the state notification.

[0077] According to a monitoring system 100c according to a third modified example of the embodiment, the monitoring device 50c includes a change sign deriving unit 59 that detects a sign of a change in the internal state of the solid-liquid separation tank from a monitoring image of the solid-liquid separation tank that is the target of diagnosis, using a fourth learning model as a learning model of change that has learned the relationship between the monitoring image and the information specifying the change in the internal state of the solid-liquid separation tank based on the monitoring image and information specifying the change in the internal state of the solid-liquid separation tank after the monitoring image was obtained. The output unit further outputs information specifying the sign of a change in the internal state of the solid-liquid separation tank detected by the change sign deriving unit, using the monitoring image of the solid-liquid separation tank that is the target of diagnosis and the fourth learning model. With this configuration, the monitoring device 50c can detect signs of changes in the internal state of the solid-liquid separation tank from monitoring images of the solid-liquid separation tank that is the target of diagnosis using a fourth learning model that has learned the relationship between the monitoring images and information that specifies changes in the internal state of the solid-liquid separation tank, thereby making it possible to monitor changes in the internal state of the solid-liquid separation tank. By being able to detect signs of changes in the internal state of the solid-liquid separation tank from monitoring images of the solid-liquid separation tank that is the target of diagnosis using the fourth learning model, human experience is not required and variation in diagnosis results can be reduced compared to when signs of changes in the internal state of the solid-liquid separation tank are detected based on human experience.

[0078] [Fourth Modification of the Embodiment] (Surveillance System) FIG. 22 is a diagram showing a configuration example of a monitoring system according to a fourth modification of the embodiment of the present invention. The monitoring system 100d according to the fourth modification of the embodiment is obtained by connecting the monitoring device 50d between the data processing device 30d and the gateway device 31 without using the network NW in the third modification of the embodiment. The monitoring system 100d according to the fourth modification of the embodiment diagnoses the sludge deposition state in a solid-liquid separation tank such as a settling tank or a thickening tank, and detects signs of change. In the fourth modification of the embodiment, a sewage treatment facility 10 is applied as an example of a facility equipped with a solid-liquid separation tank, as in the embodiment. The sewage treatment facility 10 is omitted in FIG. 22.

[0079] (Monitoring system 100d) The monitoring system 100d includes an ultrasonic sensor 20, a data processing device 30d, a monitoring device 50d, a gateway device 31, an information processing device 40d, and a terminal device 45d. The gateway device 31, the information processing device 40d, and the terminal device 45d are connected via a network NW. In the data processing device 30d, the data calculation unit 34 transmits the digital signal to the monitoring device 50d.

[0080] (Monitoring device 50d) Fig. 23 is a diagram showing an example of a monitoring device of a monitoring system according to Modification 4 of this embodiment. The monitoring device 50d is realized by a device such as a personal computer, a server, or an industrial computer. The monitoring device 50d includes a communication device 51, a recording device 52, an information processing unit 53d, and a bus line BL such as an address bus or a data bus for electrically connecting the components as shown in Fig. 23. The recording device 52 stores a program (monitoring application) executed by the monitoring device 50d. The recording device 52 also stores pixel data output by the information processing unit 53d. The recording device 52 stores training data of diagnostic results that associates information indicating the monitoring image with the diagnosis results of the internal condition of the solid-liquid separation tank based on the monitoring image, and a learning model of the diagnostic results obtained by machine learning the relationship between the monitoring image and the internal condition of the solid-liquid separation tank based on the training data of the diagnostic results. The recording device 52 stores cause training data that associates information indicating the monitoring image with information identifying the cause that results in the diagnosis of the internal condition of the solid-liquid separation tank based on the monitoring image, and a cause learning model obtained by machine learning the relationship between the monitoring image and the information identifying the cause of the internal condition of the solid-liquid separation tank based on the cause training data. The recording device 52 stores training data on countermeasures that associates information indicating a monitoring image with information identifying a method of dealing with the diagnosis results of the internal condition of the solid-liquid separation tank based on the monitoring image, and a learning model of the countermeasures obtained by machine learning the relationship between the monitoring image and the information identifying a method of dealing with the internal condition of the solid-liquid separation tank based on the training data on the countermeasures. The recording device 52 stores change training data that associates information indicating a monitoring image with information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, and a change learning model obtained by machine learning the relationship between the monitoring image and the information identifying the change in the internal state of the solid-liquid separation tank based on the change training data.

[0081] The information processing unit 53d functions as, for example, a graphic generating unit 54d, a current state determining unit 55d, a learning unit 56d, a cause determining unit 57d, a countermeasure determining unit 58d, and a change sign deriving unit 59d. The graphic conversion unit 54d acquires the digital signal received by the communication device 51. The graphic conversion unit 54d converts the value of the acquired digital signal into pixel data. The graphic conversion unit 54d stores the pixel data after the conversion of the digital signal in the recording device 52. The graphic conversion unit 54d acquires the surveillance image request received by the communication device 51. The graphic conversion unit 54d acquires pixel data stored in the recording device 52 based on the acquired surveillance image request. The graphic conversion unit 54d creates a surveillance image based on the acquired pixel data. The graphic conversion unit 54d includes information indicating the created surveillance image. The graphic conversion unit 54d creates a surveillance image response addressed to the information processing device 40d. The graphic conversion unit 54d outputs the created surveillance image response to the communication device 51. The graphic generating unit 54d acquires the tank interior status information request received by the communication device 51. The graphic generating unit 54d acquires pixel data stored in the recording device 52 based on the acquired tank interior status information request, and creates a monitoring image based on the acquired pixel data. The graphic generating unit 54d creates an tank interior status information response addressed to the information processing device 40d, including information indicating the created monitoring image. The graphic generating unit 54d outputs the created tank interior status information response to the communication device 51.

[0082] The current status determination unit 55d acquires pixel data stored in the recording device 52, and creates a monitoring image based on the acquired pixel data. The current status determination unit 55d acquires a learning model of the diagnosis result stored in the recording device 52. The current status determination unit 55d judges the internal state of the solid-liquid separation tank of the created monitoring image based on the learning model of the acquired diagnosis result. When the judgment result of the internal state of the solid-liquid separation tank is malfunction or abnormal, the current status determination unit 55d creates status notification information addressed to the information processing device 40d, including information indicating the judgment result of the internal state of the solid-liquid separation tank. The current status determination unit 55d outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the current status determination unit 55d, and transmits the acquired status notification information to the information processing device 40d. When creating a monitoring image, the current state determination unit 55d may use the measured data as is, or may thin out the data to include data measured over a long period of time in a limited display width. By including data measured over a long period of time in a limited display width, it is possible to monitor changes over a longer period of time. If it were a still image, pixel data could be picked up at any suitable interval and switched for display, but in the fourth modification of this embodiment, measurements are always performed and new data is added, so if pixel data is picked up at any suitable interval and switched for display, there is a risk of delay or obstruction in data processing, and if the measurement becomes unstable due to image display, it would be counterproductive. Therefore, in the fourth modification of this embodiment, several preset display time widths are prepared, and data storage areas for time widths corresponding to each of the multiple display time widths are created. In this embodiment, an interval (interval) for adding new data is specified, and an image database (data storage area (address)) corresponding to each of the multiple intervals is created.

[0083] An operation to switch the display is performed on the monitoring device 50d, and a display time width is selected. The current situation determination unit 55d acquires data from a database corresponding to the selected time display width, and creates a monitoring image using the acquired data. If an operation to switch the time display width is performed, data is acquired from a database corresponding to the selected time display width, and a monitoring image is created using the acquired data. By configuring in this way, smooth switching can be achieved without processing data in the database in which the data is stored, and without a time lag in creating a monitoring image. The learning unit 56d acquires the diagnosis result notification received by the communication device 51, and stores in the recording device 52 the teacher data of the diagnosis result in which the information indicating the monitoring image included in the acquired diagnosis result notification is associated with the diagnosis result of the internal state (inside the tank) of the solid-liquid separation tank based on the monitoring image. The learning unit 56d acquires the teacher data of the diagnosis result stored in the recording device 52. The learning unit 56d generates a learning model of the diagnosis result in which the monitoring image and the internal state of the solid-liquid separation tank are associated with each other by machine learning (supervised learning) the monitoring image and the diagnosis result of the internal state of the solid-liquid separation tank based on the acquired teacher data of the diagnosis result. For example, the learning unit 56d recognizes the monitoring image using a convolutional neural network. The learning model of the diagnosis result classifies the monitoring image into one of normal, malfunction, and abnormal as the internal state of the solid-liquid separation tank based on the information indicating the monitoring image. The learning unit 56d stores the generated learning model of the diagnosis result in the recording device 52. The learning unit 56d acquires the teacher data of the countermeasures stored in the recording device 52. The learning unit 56d generates a learning model of the countermeasures that associates the monitoring image with the countermeasures for the internal state of the solid-liquid separation tank by performing machine learning (supervised learning) on ​​the monitoring image and information that specifies the countermeasures for the diagnosis result of the internal state of the solid-liquid separation tank based on the acquired teacher data of the countermeasures. For example, the learning unit 56d recognizes the monitoring image using a convolutional neural network. The learning model of the countermeasures classifies the monitoring image into any of the information that specifies the countermeasures for the internal state of the solid-liquid separation tank based on the information indicating the monitoring image. The learning unit 56d stores the generated learning model of the countermeasures in the recording device 52.

[0084] The learning unit 56d acquires the teacher data of the change stored in the recording device 52. The learning unit 56d generates a learning model of the change that associates the monitoring image with the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, by performing machine learning (supervised learning) on ​​the monitoring image and the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, based on the acquired teacher data of the change. For example, the learning unit 56d recognizes the monitoring image using a convolutional neural network. The learning model of the change classifies the monitoring image into any of the information that specifies the change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, based on the information indicating the monitoring image. The learning unit 56d stores the generated learning model of the change in the recording device 52.

[0085] The cause determination unit 57d acquires information showing the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55d. When the acquired determination result of the internal state of the solid-liquid separation tank is malfunction or abnormal, the cause determination unit 57d acquires a learning model of the cause stored in the recording device 52. The cause determination unit 57d judges information specifying the cause of the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the cause. The cause determination unit 57d creates status notification information addressed to the information processing device 40d, including information showing the monitoring image, information showing the internal state of the solid-liquid separation tank, and information showing the determination result of the information specifying the cause of the internal state of the solid-liquid separation tank. The cause determination unit 57d outputs the created status notification information to the communication device 51. The communication device 51 acquires the status notification information output by the cause determination unit 57d and transmits the acquired status notification information to the information processing device 40d.

[0086] The countermeasure determination unit 58d acquires information indicating the monitoring image and the determination result of the internal state of the solid-liquid separation tank from the current state determination unit 55d. When the acquired determination result of the internal state of the solid-liquid separation tank is a malfunction or abnormality, the countermeasure determination unit 58d acquires a learning model of the countermeasure stored in the recording device 52. The countermeasure determination unit 58d determines a countermeasure for the internal state of the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of the countermeasure. The countermeasure determination unit 58d acquires information for identifying the cause of the internal state of the solid-liquid separation tank in the monitoring image from the cause determination unit 57d. The countermeasure determination unit 58d creates status notification information addressed to the information processing device 40d, the status notification information including information showing the monitoring image, information for identifying the cause of the internal state of the solid-liquid separation tank, and information for identifying the method of dealing with the internal state of the solid-liquid separation tank. The countermeasure determination unit 58d outputs the created status notification information to the communication device 51.

[0087] The change sign deriving unit 59d acquires information indicating a monitoring image from the current state determining unit 55d. The change sign deriving unit 59d acquires a learning model of change stored in the recording device 52. The change sign deriving unit 59d derives a sign of change in the solid-liquid separation tank in the acquired monitoring image based on the acquired learning model of change. All or a part of the information processing unit 53d is a functional unit (hereinafter referred to as a software functional unit) that is realized by, for example, a processor such as a CPU executing a program such as a monitoring application stored in the recording device 52. All or a part of the information processing unit 53c may be realized by hardware such as an LSI, an ASIC, or an FPGA, or may be realized by a combination of a software functional unit and hardware. The information processing device 40 can be applied to the information processing device 40d.

[0088] (Terminal device 45d) The terminal device 45d is realized by a device such as a personal computer, a server, an industrial computer, etc. An example of the terminal device 45d is installed in a monitoring center that monitors the sewage treatment facility 10. When diagnosing the internal condition of the solid-liquid separation tank, the user operates the terminal device 45d to create a monitoring image request including information requesting a monitoring image and addressed to the monitoring device 50d. The terminal device 45d creates the monitoring image request based on the user's operation. The terminal device 45d transmits the created monitoring image request to the monitoring device 50d. The terminal device 45d receives the monitoring image response sent by the monitoring device 50d in response to the monitoring image request sent to the monitoring device 50d. The terminal device 45d displays the monitoring image included in the monitoring image response. The user refers to the monitoring image displayed by the terminal device 45d, diagnoses the internal state of the solid-liquid separation tank included in the monitoring image, estimates the cause of the internal state of the solid-liquid separation tank, identifies how to deal with the internal state of the solid-liquid separation tank, predicts the internal state of the solid-liquid separation tank after the monitoring image is obtained, and identifies any changes therein. The user operates the terminal device 45d to create a diagnosis result notification addressed to the monitoring device 50d, the diagnosis result including information showing the monitoring image, the diagnosis result of the internal state of the solid-liquid separation tank, information specifying the cause of the diagnosis result, information specifying the method of dealing with the internal state of the solid-liquid separation tank, and information specifying the change in the internal state of the solid-liquid separation tank after the monitoring image was obtained. The terminal device 45d creates the diagnosis result notification based on the user's operation. The terminal device 45d transmits the created diagnosis result notification to the monitoring device 50d.

[0089] (Operation of monitoring system) Fig. 24 is a diagram showing a first example of the operation of the monitoring system according to the fourth modified example of the embodiment. Referring to Fig. 24, a process will be described in which monitoring device 50d accumulates information indicating a monitoring image included in a diagnosis result notification transmitted by terminal device 45d, a diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, information identifying a method for dealing with the diagnosis result, and information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, and performs machine learning based on the accumulated information indicating the monitoring image, the diagnosis result of the internal state of the solid-liquid separation tank, information identifying the cause of the diagnosis result, information identifying a method for dealing with the diagnosis result, and information identifying a change in the internal state of the solid-liquid separation tank after the monitoring image is obtained, to generate a learning model of the diagnosis result, a learning model of the cause, a learning model of the method for dealing with the diagnosis, and a learning model of the change. Steps S1-1 to S10-1 in FIG. 7 can be applied to steps S1-10 to S10-10, and steps S11-8 to S15-8 in FIG. 20 can be applied to steps S11-10 to S15-10, so a description thereof will be omitted here.

[0090] Fig. 25 is a diagram showing a second example of the operation of the monitoring system according to the fourth modification of the embodiment. Referring to Fig. 25, the monitoring device 50d acquires the digital signal transmitted by the data processing device 30d, and creates a monitoring image based on the acquired digital signal. The process of determining the internal state of the solid-liquid separation tank based on the created monitoring image by the monitoring device 50d will be described. Steps S1-1 to S6-11 can be implemented by steps S1-1 to S6-1 in FIG. 7, and steps S7-11 to S17-11 can be implemented by steps S7-9 to S17-9 in FIG. 21, so a description thereof will be omitted here. The process in which the monitoring device 50d transmits information indicating a monitoring image based on a request for information on an inside of the tank transmitted by the information processing device 40d can be described with reference to FIG. 9, and therefore a description thereof will be omitted.

[0091] In the fourth modification of the embodiment described above, the monitoring device 50d is connected between the data processing device 30d and the gateway device 31 without passing through the network NW in the third modification of the embodiment, but the present invention is not limited to this example. For example, the embodiment can be applied to a case where the monitoring device 50 is connected between the data processing device 30 and the gateway device 31 without passing through the network NW. The first modification of the embodiment can be applied to a case where the monitoring device 50a is connected between the data processing device 30 and the gateway device 31 without passing through the network NW. The second modification of the embodiment can be applied to a case where the monitoring device 50b is connected between the data processing device 30 and the gateway device 31 without passing through the network NW.

[0092] According to the monitoring system 100d according to the fourth modified example of the embodiment, the monitoring device 50d can be installed at the site where the sewage treatment facility 10 is installed by connecting the monitoring device 50d between the data processing device 30 and the gateway device 31 without going through the network NW. Therefore, the data of the data processing device 30d can be monitored in real time, compared to the case where the monitoring device 50d is installed at a position away from the site where the sewage treatment facility 10 is installed via the network NW. Since the monitoring device 50d can perform the judgment and derivation in real time, the time required for the judgment and derivation can be shortened. Since the time required for the judgment and derivation can be shortened, the status can be notified immediately when an abnormality or malfunction is judged. Since the monitoring device 50d can be installed at the site where the sewage treatment facility 10 is installed, the risk of data hacking and attacks on the system can be reduced compared to when the monitoring device 50d is installed at a location away from the site where the sewage treatment facility 10 is installed via the network NW. Since the monitoring device 50d can be installed at the site where the sewage treatment equipment 10 is installed, it is easier to determine the situation specific to the site (equipment), determine the cause, determine the countermeasure, and derive signs, compared to when the monitoring device 50d is installed at a location away from the site where the sewage treatment equipment 10 is installed via the network NW. An information processing device 40d installed in a monitoring center can remotely update information (programs, monitoring apps, learning models, image data, AI analysis results) recorded in a recording device 52 of a monitoring device 50d installed at the site via a network NW and a gateway device 31. Information (mainly pixel data and AI analysis results) recorded in the recording device 52 of the monitoring device 50d installed on-site can be updated to an information processing device 40d installed remotely via the gateway device 31 and the network NW.

[0093] Although an embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and design changes and the like that do not deviate from the gist of the present invention are also included. For example, a computer program for implementing the functions of each of the above-mentioned devices may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer system and executed. Note that the "computer system" referred to here may include hardware such as an OS and peripheral devices. In addition, the term "computer-readable recording medium" refers to writable non-volatile memory such as a flexible disk, an optical magnetic disk, a ROM, or a flash memory, a portable medium such as a DVD (Digital Versatile Disc), or a storage device such as a hard disk built into a computer system.

[0094] Furthermore, the term "computer-readable recording medium" includes devices that retain a program for a certain period of time, such as volatile memory (e.g., DRAM (Dynamic Random Access Memory)) within a computer system that serves as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. The program may be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by a transmission wave in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The program may be for realizing part of the functions described above. Furthermore, the above-mentioned functions may be realized in combination with a program already recorded in the computer system, that is, a so-called differential file (differential program). [Explanation of symbols]

[0095] 10... sewage treatment equipment, 11... pre-settling tank, 12... thickening tank, 13... storage tank, 14... dehydrator, 15... container, 16... aeration tank, 17... post-settling tank, 18... pump, 19... equipment control device, 20... ultrasonic sensor, 21 oscillator, 22... receiver, 23... suspended solids sediment layer, 24... supernatant water, 25... treatment tank, 26... interface, 27... height, 30, 30d... data processing device, 31... gateway device, 32... ultrasonic transmission / reception circuit, 33... data conversion circuit, 34... data calculation unit, 35... image data storage unit, 36... display switching operation unit, 37... image data display unit, 40, 40a, 40b, 40c, 40d...information processing device, 45, 45a, 45b, 45c, 45d...terminal device, 50, 50a, 50b, 50c, 50d...monitoring device, 51...communication device, 52...recording device, 53, 53a, 53b, 53c...information processing unit, 54, 54d...graphics conversion unit, 55, 55a, 55d...current situation determination unit, 56, 56a, 56b, 56c, 56d...learning unit, 57, 57b, 57d...cause determination unit, 58, 58c, 58d...countermeasure determination unit, 59, 59d...change sign derivation unit, 100, 100a, 100b, 100c, 100d...monitoring system

Claims

1. a judgment unit that judges the internal state of each of the multiple solid-liquid separation tanks from the monitoring images showing the interior of each of the multiple solid-liquid separation tanks that are the subject of diagnosis, using a first learning model that learns the relationship between the monitoring images and the internal state of the solid-liquid separation tanks based on monitoring images, which are images showing the interior of each of the multiple solid-liquid separation tanks for separating wastewater into solid and liquid, and a diagnosis result based on the monitoring images of the interior of each of the multiple solid-liquid separation tanks; an output unit that outputs information that specifies the internal state of each of the plurality of solid-liquid separation tanks that are targets of diagnosis, determined by the determination unit using the monitoring image of each of the plurality of solid-liquid separation tanks that are targets of diagnosis and the first learning model; A surveillance system having:

2. A cause determination unit is provided which determines information identifying the cause of the state inside the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank which is the subject of diagnosis, using a second learning model which has learned the relationship between the monitoring image and the information identifying the cause of the diagnosis result inside the solid-liquid separation tank based on a plurality of the monitoring images and information identifying the cause of the diagnosis result based on each of the plurality of the monitoring images, The monitoring system of claim 1, wherein the output unit further outputs information identifying the cause of the state inside the solid-liquid separation tank determined by the cause determination unit using the monitoring image of the solid-liquid separation tank that is the subject of diagnosis and the second learning model.

3. A countermeasure determination unit is provided which determines information specifying how to deal with the condition inside the solid-liquid separation tank from the monitoring image of the solid-liquid separation tank which is the subject of diagnosis, using a third learning model which has learned the relationship between the monitoring image and information specifying how to deal with the diagnosis result inside the solid-liquid separation tank based on a plurality of the monitoring images and information specifying how to deal with the diagnosis result based on each of the plurality of the monitoring images, A monitoring system as described in claim 1 or claim 2, wherein the output unit further outputs information identifying a method of dealing with the condition inside the solid-liquid separation tank determined by the method of dealing with the condition determined by the method of dealing with the condition inside the solid-liquid separation tank using the monitoring image of the solid-liquid separation tank that is the subject of diagnosis and the third learning model.

4. A change prediction deriving unit is provided which detects predictions of changes in the state inside the solid-liquid separation tank from the monitoring images of the solid-liquid separation tank which is the subject of diagnosis, using a fourth learning model which has learned the relationship between the monitoring images and the information which specifies the change in the state inside the solid-liquid separation tank based on the multiple monitoring images and information which specifies the change in the state inside the solid-liquid separation tank after each of the multiple monitoring images is obtained, The monitoring system according to any one of claims 1 to 3, wherein the output unit further outputs information that identifies a sign of change in the state inside the solid-liquid separation tank detected by the change sign derivation unit using the monitoring image of the solid-liquid separation tank that is the subject of diagnosis and the fourth learning model.

5. The monitoring system according to claim 1 , wherein the diagnosis result is generated based on one or both of an accumulation state of solid matter and a floating state of solid matter contained in a monitoring image.

6. The monitoring system according to any one of claims 1 to 5, wherein the judgment unit judges whether the internal condition of the solid-liquid separation tank, which is the subject of diagnosis, is normal, malfunction, or abnormal from the monitoring image showing the interior of the solid-liquid separation tank.

7. a notification unit that notifies the user that the internal state of the solid-liquid separation tank is either a malfunction or an abnormality when the determination unit determines that the internal state of the solid-liquid separation tank is either a malfunction or an abnormality. The surveillance system according to any one of claims 1 to 6, further comprising:

8. A learning device for generating a first learning model for determining an internal state of each of a plurality of solid-liquid separation tanks from a monitoring image showing the interior of each of the plurality of solid-liquid separation tanks to be diagnosed, comprising: a learning unit that generates, by learning, a first learning model that represents a relationship between the monitoring image and the internal state of each of a plurality of solid-liquid separation tanks for separating wastewater into solid and liquid, based on a monitoring image that is an image showing the interior of each of the plurality of solid-liquid separation tanks and a diagnosis result based on the monitoring image of the internal state of each of the plurality of solid-liquid separation tanks. A learning device having the above configuration.

9. the learning unit generates, by learning, a second learning model that represents a relationship between the monitoring images and information that identifies a cause of a diagnosis result inside the solid-liquid separation tank, based on a plurality of the monitoring images and information that identifies a cause of a diagnosis result based on each of the plurality of the monitoring images; The learning device according to claim 8.

10. the learning unit generates, by learning, a third learning model that represents a relationship between the monitoring images and information that specifies a method of dealing with a diagnosis result of the inside of the solid-liquid separation tank, based on the plurality of monitoring images and information that specifies a method of dealing with a diagnosis result based on each of the plurality of monitoring images; The learning device according to claim 8 or 9.

11. A learning device as described in any one of claims 8 to 10, wherein the learning unit generates a fourth learning model representing the relationship between the monitoring images and information specifying changes in the internal state of the solid-liquid separation tank based on a plurality of the monitoring images and information specifying changes in the internal state of the solid-liquid separation tank after each of the plurality of monitoring images is obtained.

12. The learning device according to claim 8 , wherein the diagnosis result is generated based on one or both of an accumulation state of solid matter and a floating state of solid matter included in a monitoring image.

13. a step of determining an internal state of each of a plurality of solid-liquid separation tanks from a monitoring image showing the interior of each of a plurality of solid-liquid separation tanks that are the subject of diagnosis, using a first learning model that has learned the relationship between the monitoring image and the internal state of the solid-liquid separation tank based on a monitoring image that is an image showing the interior of each of a plurality of solid-liquid separation tanks for separating wastewater into solid and liquid, and a diagnosis result based on the monitoring image of the interior of each of the plurality of solid-liquid separation tanks; a step of outputting information for identifying the internal state of each of the plurality of solid-liquid separation tanks that are targets of diagnosis, determined in the determining step using the monitoring image of each of the plurality of solid-liquid separation tanks that are targets of diagnosis and the first learning model; A monitoring method executed by a monitoring system, comprising:

14. A learning method executed by a learning device that generates a first learning model for determining an internal state of each of a plurality of solid-liquid separation tanks to be diagnosed from a monitored image showing the interior of each of the plurality of solid-liquid separation tanks, the method comprising: A step of generating, by learning, a first learning model that represents a relationship between the monitoring image and the internal state of each of a plurality of solid-liquid separation tanks for separating wastewater into solid and liquid, based on a monitoring image that is an image showing the interior of each of the plurality of solid-liquid separation tanks and a diagnosis result based on the monitoring image of the interior of each of the plurality of solid-liquid separation tanks. A learning method executed by a learning device.

15. The monitoring system computer a step of determining an internal state of each of a plurality of solid-liquid separation tanks from a monitoring image showing the interior of each of a plurality of solid-liquid separation tanks that are the subject of diagnosis, using a first learning model that has learned the relationship between the monitoring image and the internal state of the solid-liquid separation tank based on a monitoring image that is an image showing the interior of each of a plurality of solid-liquid separation tanks for separating wastewater into solid and liquid, and a diagnosis result based on the monitoring image of the interior of each of the plurality of solid-liquid separation tanks; a step of outputting information for identifying the internal state of each of the plurality of solid-liquid separation tanks that are targets of diagnosis, determined in the determining step using the monitoring image of each of the plurality of solid-liquid separation tanks that are targets of diagnosis and the first learning model; A program to execute.

16. A computer of a learning device that generates a first learning model for determining an internal state of each of a plurality of solid-liquid separation tanks to be diagnosed from a monitoring image showing the interior of each of the plurality of solid-liquid separation tanks, A step of generating, by learning, a first learning model that represents a relationship between the monitoring image and the internal state of each of a plurality of solid-liquid separation tanks for separating wastewater into solid and liquid, based on a monitoring image that is an image showing the interior of each of the plurality of solid-liquid separation tanks and a diagnosis result based on the monitoring image of the interior of each of the plurality of solid-liquid separation tanks. A program to execute.

Images