Steam trap diagnostic device

Abstract

To provide a diagnostic device that can determine leakage even in steam traps whose diagnostic logic is unknown, by using an estimated diagnostic logic. [Solution] The diagnostic device comprises a storage means for storing diagnostic logic for gas leakage determination for each type of steam trap, a determination means for performing leakage determination of the target by applying the diagnostic logic to the target based on pressure and vibration values, an input receiving means for receiving input of trap information including characteristic information regarding the discharge flow rate of drain, and a generation means for generating estimated diagnostic logic for the target using diagnostic information estimated by applying estimation logic based on the trap information of the target. The estimation logic is determined by a learning model trained on training data including trap information and diagnostic information used in the diagnostic logic, and the determination means performs leakage determination by applying the estimated diagnostic logic according to the fulfillment of predetermined conditions based on pressure and vibration values.

Description

Technical Field

[0001] This invention relates to a diagnostic device for steam traps and the like.

Background Art

[0002] In a steam piping system such as a steam plant, steam condenses to generate drain (condensate). The generated drain is discharged to the outside or the like using a steam trap or the like. In a steam trap, abnormalities such as steam leakage may occur. Therefore, an inspector moves to each steam trap and performs a diagnosis using a diagnostic device.

[0003] As a diagnostic device, there is a pass / fail determination device that detects vibrations associated with the operation of a steam trap and determines whether the operation is normal or not, such as steam leakage (for example, see Patent Document 1). In the pass / fail determination device described in Patent Document 1, the relationship between the vibration level and the pressure in the steam system is stored in advance, and the presence or absence of steam leakage is determined (automatically determined) by comparing the detected vibration level with the above relationship. In the automatic determination of steam leakage described in Patent Document 1, the relationship between the vibration level and the pressure in the steam system is stored in advance for each steam trap. Therefore, it is difficult to perform automatic determination for a steam trap for which the above relationship is not stored.

[0004] There is also a diagnostic device that uses the vibration characteristics of a certain steam trap to diagnose the performance related to steam leakage of a steam trap that has the same operation type (valve opening / closing method type) as this steam trap but a different model (maximum drain discharge amount) (for example, see Patent Document 2). That is, the diagnostic device described in Patent Document 2 has a configuration in which, for a steam trap, if the operation type is the same, diagnosis is possible even if the models are different.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

[0006] Even with the configuration of a diagnostic device for steam traps of the same operating type but different models as described above, the diagnostics are limited to steam traps of the same operating type, and it is difficult to diagnose steam traps of operating types for which vibration characteristics have not been pre-defined.

[0007] The purpose of this invention is to provide a diagnostic device, etc., that can perform leak detection using estimated diagnostic logic, even in the case of a steam trap whose diagnostic logic for gas leak detection is unknown. [Means for solving the problem]

[0008] A steam trap diagnostic device provided by the first aspect of the present invention comprises: measuring means for measuring the vibration value of a steam trap to be diagnosed; acquiring means for acquiring the pressure value of the steam trap to be diagnosed; storage means for storing a diagnostic logic for determining gas leakage from the steam trap for each type of steam trap; determination means for performing a leakage determination of the steam trap by applying the diagnostic logic of the steam trap to the steam trap based on the pressure value and vibration value; input receiving means for receiving input of trap information including characteristic information relating to at least the discharge flow rate of drain; and generation means for generating an estimated diagnostic logic for the steam trap using diagnostic information estimated by applying an estimation logic based on the trap information of the steam trap. The estimation logic is determined by a learning model trained on training data including trap information of the steam trap and diagnostic information used in the steam trap diagnostic logic, and the determination means performs a leakage determination of the steam trap by applying the estimated diagnostic logic of the steam trap to the pressure value and vibration value, depending on whether predetermined conditions are met.

[0009] The above predetermined conditions may also include the condition that the diagnostic logic to be diagnosed is not stored.

[0010] The above trap information also includes information on the operation type of the steam trap, and the estimation logic is determined for each operation type by a learning model trained on the training data, and the generation means may generate the estimation diagnostic logic for the target by applying the estimation logic corresponding to the operation type to be diagnosed.

[0011] In the above diagnostic logic, the presence or absence of leakage is calculated from a determination formula using diagnostic parameters, which are diagnostic information corresponding to the object to be diagnosed, based on the pressure value and vibration value. In the estimation diagnostic logic, the presence or absence of leakage is calculated from a determination formula using estimated diagnostic parameters, which are estimated diagnostic information corresponding to the object to be diagnosed, based on the pressure value and vibration value. The generation means may generate a determination formula for the estimation diagnostic logic of the object to be diagnosed using estimated diagnostic parameters estimated by applying estimation logic based on the characteristic information of the object to be diagnosed.

[0012] In the estimation logic described above, estimated diagnostic parameters may be calculated from a calculation formula determined by a learning model, based on the characteristic information of the subject to be diagnosed.

[0013] The characteristic information of the object to be diagnosed above is a combination of the pressure value of the steam trap to be diagnosed and the discharge flow rate at that pressure value, and may include multiple sets of pressure values ​​and discharge flow rates.

[0014] The above sets of pressure values ​​and discharge flow rates may include combinations of the maximum pressure value and the maximum discharge flow rate, which is the discharge flow rate at that maximum pressure value.

[0015] The steam trap diagnostic program provided by the second aspect of the present invention causes the computer of the steam trap diagnostic device to function as: measuring means for measuring the vibration value of the steam trap to be diagnosed; acquiring means for acquiring the pressure value of the steam trap to be diagnosed; determining means for performing a leak determination of the steam trap by applying the diagnostic logic of the steam trap to be diagnosed from among the gas leak determination diagnostic logics stored for each type of steam trap, based on the pressure value and the vibration value; input receiving means for receiving input of trap information including characteristic information relating to at least the discharge flow rate of drain; and generating means for generating an estimated diagnostic logic of the steam trap using the diagnostic information estimated by applying an estimation logic based on the trap information of the steam trap. The estimation logic is determined by a learning model learned with training data including the trap information of the steam trap and the diagnostic information used in the steam trap diagnostic logic, and the determining means performs a leak determination of the steam trap by applying the estimated diagnostic logic of the steam trap to the pressure value and vibration value, depending on whether predetermined conditions are met.

[0016] A steam trap diagnostic system provided by a third aspect of the present invention comprises: measuring means for measuring the vibration value of a steam trap to be diagnosed; acquiring means for acquiring the pressure value of the steam trap to be diagnosed; storage means for storing diagnostic logic for determining gas leakage from the steam trap for each type of steam trap; determination means for performing a leakage determination of the steam trap by applying the diagnostic logic of the steam trap to the steam trap based on the pressure value and vibration value; input receiving means for receiving input of trap information including characteristic information relating to at least the discharge flow rate of drain; and generation means for generating an estimated diagnostic logic for the steam trap using diagnostic information estimated by applying an estimation logic based on the trap information of the steam trap. The estimation logic is determined by a learning model trained on training data including trap information of the steam trap and diagnostic information used in the steam trap diagnostic logic, and the determination means performs a leakage determination of the steam trap by applying the estimated diagnostic logic of the steam trap to the pressure value and vibration value, depending on whether predetermined conditions are met.

[0017] The learning model generation method provided by the fourth aspect of the present invention includes acquisition processing for acquiring a plurality of pieces of teacher data, which is teacher data including trap information containing characteristic information regarding the discharge flow rate of the drain of a steam trap and diagnostic information used in the diagnostic logic for determining gas leakage of the steam trap, output processing for inputting the trap information of the teacher data and outputting diagnostic information estimated by applying estimation logic, and correction processing for performing learning by correcting the parameters of the estimation logic so that the diagnostic information estimated in the output processing approximates the diagnostic information of the teacher data.

Advantages of the Invention

[0018] According to this invention, when a predetermined condition such as a case where the diagnostic logic of the steam trap to be diagnosed is not stored is satisfied, an estimated diagnostic logic is generated and the presence or absence of leakage of the diagnostic target is determined by estimation. Therefore, even for a steam trap with unknown diagnostic logic for gas leakage determination, highly accurate leakage determination is possible.

Brief Description of the Drawings

[0019] [Figure 1] It is a configuration diagram of a diagnostic system according to an embodiment of this invention. [Figure 2] It is an external view of a diagnostic machine according to an embodiment of this invention. [Figure 3] It is a schematic diagram showing the state during diagnosis of a diagnostic machine according to an embodiment of this invention. [Figure 4] It is a graph showing an example of a function of a determination threshold value of a steam trap. [Figure 5] It is a graph for explaining the relationship (characteristics) between the pressure value and the discharge flow rate of a steam trap. [Figure 6] It is a graph showing an example of an estimation result of diagnostic information by the estimation logic according to an embodiment of this invention. [Figure 7] It is a flowchart showing the diagnostic process according to an embodiment of this invention. [Figure 8] It is a diagram showing an overview of a learning model according to an embodiment of this invention.

Embodiment for Carrying Out the Invention

[0020] A diagnostic system (diagnostic device, diagnostic program) and a method for generating a learning model according to an embodiment of the present invention will be described with reference to the drawings. Note that the configuration of the present invention is not limited to the embodiment. Also, the order of various processes constituting the flow described below may be changed arbitrarily as long as there is no contradiction in the processing content.

[0021] FIG. 1 is a schematic diagram of a diagnostic system 100 according to an embodiment of the present invention. The diagnostic system 100 includes a diagnostic machine 10, a terminal device (management terminal) 20, a server device 30, and the like. The diagnostic machine 10 and the terminal device 20 can perform wireless communication within a communicable range by a communication method such as Bluetooth (registered trademark). The terminal device 20 and the server device 30 are communicably connected via a communication network N such as the Internet.

[0022] The diagnostic system 100 diagnoses a steam trap. In the diagnosis, a leakage determination (automatic determination) of the steam (gas) of the steam trap is performed. The steam trap is disposed in the steam piping system of the steam plant. The steam trap is configured to automatically let the drain (condensate) generated by the condensation of the steam flow out to the downstream side, but to confine the steam in the steam trap so that it does not flow out. The leakage determination of the diagnostic system 100 determines whether the steam in the steam trap leaks to the outside. In the diagnosis, after measuring the vibration value and the like of the steam trap, the leakage determination is executed. For example, when a diagnostic request is input by an inspector using the diagnostic machine 10, the diagnosis is performed. Hereinafter, the steam trap to be diagnosed may be simply referred to as the diagnostic target.

[0023] There are several types of steam traps. These types are classified by factors such as operating type, model, and manufacturer. The type of steam trap can be identified by its product number. Steam traps of the same type are assigned the same product number. Furthermore, the operating type, model, and manufacturer can be identified from the product number. The operating type is classified based on the valve opening and closing mechanism, such as float type, bucket type, or disc type. The model is classified based on the maximum drain discharge rate. The manufacturer is classified based on the steam trap manufacturer. Note that classification based on at least the operating type and model is sufficient.

[0024] In this embodiment, the diagnostic device 10 performs leak detection. The diagnostic device 10 performs leak detection based on the ultrasonic vibration (vibration value) and pressure value of the steam trap. In the leak detection, the result is "normal" if it is determined that no steam leak has occurred, and "leak" if it is determined that a steam leak has occurred. The vibration value is measured by the diagnostic device 10.

[0025] In this embodiment, a data leak determination is basically performed by applying a diagnostic logic corresponding to the target of the diagnosis. However, if the diagnostic logic corresponding to the target of the diagnosis is not stored (registered), for example, an estimated diagnostic logic for the target of the diagnosis is generated, and a data leak determination is performed by applying this estimated diagnostic logic. Details will be described later.

[0026] <Configuration of the diagnostic machine> Figure 2 is an external view of the diagnostic device 10. The diagnostic device (diagnostic apparatus) 10 determines steam leakage from the steam trap by executing a diagnostic program. The diagnostic device 10 includes a vibration measuring unit 11, a temperature measuring unit 12, a storage unit 13, a calculation unit 14, a communication unit 15, input buttons 16, and a liquid crystal display 17, etc.

[0027] The vibration measurement unit 11 measures the ultrasonic vibration (vibration) of the steam trap. As shown in Figure 2, the vibration measurement unit 11 is located at the tip of a rod-shaped probe 10a. By bringing the probe 10a into contact with the steam trap to be diagnosed (see, for example, Figure 3), vibration measurement by the vibration measurement unit 11 becomes possible. The vibration measurement unit 11 includes a known ultrasonic detection element (e.g., a piezoelectric element). In this embodiment, one vibration measurement is performed once for a predetermined period (e.g., 15 seconds) of ultrasonic vibration. The measurement data measured by the vibration measurement unit 11 is stored in the storage unit 13. The calculation unit 14 calculates the RMS (Root mean square) of the vibration as the vibration value from the measurement data. Note that if it can be used for leakage detection, etc., a value other than RMS may be applied as the vibration value.

[0028] The temperature measuring unit 12 measures the temperature (surface temperature) of the steam trap. As shown in Figure 2, the temperature measuring unit 12 is located at the tip of a rod-shaped probe 10a. By bringing the probe 10a into contact with the steam trap (see, for example, Figure 3), temperature measurement by the temperature measuring unit 12 becomes possible. The temperature measuring unit 12 includes a known temperature measuring element (e.g., a thermocouple). The measurement data measured by the temperature measuring unit 12 is stored in the storage unit 13. The calculation unit 14 calculates the temperature value from the measurement data.

[0029] The storage unit 13 is, for example, a semiconductor memory (such as flash memory). The storage unit 13 stores the diagnostic program described above. The storage unit 13 also stores information related to the diagnostic logic for each type of steam trap (diagnostic logic information). In this embodiment, the storage unit 13 stores diagnostic logic information for some of the types of steam traps that can be used in the diagnostic system 100 to determine leakage. For example, the aforementioned some traps are those included in the trap list information described later. The diagnostic logic information is, for example, information related to the determination expression (function) used in the diagnostic logic (function information).

[0030] Furthermore, the storage unit 13 stores trap list information of types that can be used for leak detection by the diagnostic logic in the diagnostic system 100. The trap list information is, for example, a product number that can identify the type. In this embodiment, the storage unit 13 stores trap list information for some of the types of steam traps that can be used for leak detection by the diagnostic logic in the diagnostic system 100. The trap list information and diagnostic logic information for the above-mentioned some traps are, for example, information about the steam trap to be diagnosed, and are transmitted from the terminal device 20 to the diagnostic machine 10 before the diagnosis. For example, when an inspector selects the steam trap to be diagnosed at the terminal device 20, the above-mentioned trap list information, etc., is transmitted from the terminal device 20 to the diagnostic machine 10. Note that all types of trap list information and diagnostic logic information may be stored in the storage unit 13 in advance.

[0031] Furthermore, the memory unit 13 stores information (generated information) for generating estimation diagnostic logic, such as information related to estimation logic.

[0032] The calculation unit 14 includes a CPU, memory (RAM), etc. The calculation unit 14 performs processing related to the diagnosis of the steam trap. Specifically, the calculation unit 14 calculates vibration values ​​from vibration measurement data measured by the vibration measurement unit 11. The calculation unit 14 also calculates temperature values ​​from measurement data measured by the temperature measurement unit 12. Since the calculation of vibration values ​​and temperature values ​​is by known methods, a detailed explanation is omitted.

[0033] Furthermore, the calculation unit 14 applies diagnostic logic corresponding to the object to be diagnosed based on the vibration value and pressure value to determine whether a leak exists in the object to be diagnosed. The pressure value is, for example, a value indicating the current pressure (inlet pressure) inside the object to be diagnosed (steam trap). The pressure value used is, for example, a value input by an inspector via the terminal device 20. For example, the inspector may input a pressure value measured by another device or an expected pressure value.

[0034] In the diagnostic logic of this embodiment, the presence or absence of leakage is calculated from a determination formula using diagnostic information (diagnostic parameters) corresponding to the object to be diagnosed, based on pressure and vibration values. In other words, the diagnostic logic is a diagnostic method that performs leakage determination using a determination formula. In the diagnostic logic of this embodiment, for example, leakage determination is performed using the following two determination formulas (1) and (2) (functions).

number

[0035] "k", "l", and "m" are functions, and "r", "s", and "d" are the diagnostic parameters mentioned above. The numerical values ​​of the diagnostic parameters (r, s, d) vary depending on the type of steam trap. That is, the judgment formulas (1) and (2) are applied to the diagnostic logic in common, regardless of the type of steam trap. The values ​​of the diagnostic parameters (r, s, d) are those prepared for each type of steam trap. The diagnostic parameters (r, s, d) are values ​​obtained through various experiments. The judgment formulas and the diagnostic parameters (r, s, d) for each type of steam trap are stored in the storage unit 13 as diagnostic logic information.

[0036] In the diagnostic logic, the measured vibration value is input to the variable x in the judgment formulas (1) and (2), which use the diagnostic parameters of the device being diagnosed, and the result of the leak detection is output. Specifically, if f1 > 0 and f2 > 0, it is determined to be a "leak". Otherwise, it is determined to be "normal". Although details are omitted, the pressure value is also used in judgment formula (2).

[0037] The diagnostic logic using judgment formula (1) refers to a diagnostic logic that uses the relationship between the diagnostic parameter (r,s) and the judgment threshold N (vibration value) as shown in Figure 4(A). Figure 4(A) is a graph showing the function of the judgment threshold N. The horizontal axis of the graph shows the vibration value, and the vertical axis shows the diagnostic parameter (r,s). In Figure 4(A), for example, if the diagnostic parameter (r,s) of the object to be diagnosed is (r1,s1), the judgment threshold N1 is determined. Then, it is determined whether the measured vibration value V1 of the object to be diagnosed exceeds the judgment threshold N1. That is, this is equivalent to the process of determining whether f1>0 in the diagnostic logic using judgment formula (1). Furthermore, the diagnostic logic using judgment formula (2) refers to a diagnostic logic that uses the relationship between the pressure value (P) and the diagnostic parameter (d) and the judgment threshold M (vibration value) as shown in Figure 4(B). Figure 4(B) is a graph showing the function of the judgment threshold M. The horizontal axis of the graph represents the vibration value, and the vertical axis represents the pressure value (P) and the diagnostic parameter (d), which is (P,d). In Figure 4(B), for example, if the target of diagnosis is (P1,d1), the judgment threshold M1 is determined. Then, it is determined whether the measured vibration value V1 of the target of diagnosis exceeds the judgment threshold M1. In other words, this is equivalent to the process of determining whether f2 > 0 in the diagnostic logic using judgment formula (2).

[0038] Figures 4(A) and 4(B) are examples shown for illustrative purposes. Furthermore, since the diagnostic logic using the above-described judgment formulas (1) and (2) is a well-known method, a detailed explanation of the configuration of judgment formulas (1) and (2) is omitted. Also, judgment formulas (1) and (2) are examples and are not limiting.

[0039] Furthermore, if the predetermined conditions are met, the calculation unit 14 applies the estimated diagnostic logic of the target to perform a data leak determination. That is, if the predetermined conditions are not met, the data leak determination is performed by applying the diagnostic logic of the target to perform a data leak determination. On the other hand, if the predetermined conditions are met, the data leak determination is performed by applying the estimated diagnostic logic of the target to perform a data leak determination.

[0040] The predetermined condition is, for example, that the diagnostic logic corresponding to the target of diagnosis is not stored (registered). Specifically, the predetermined condition is met if the diagnostic logic information for the diagnostic logic corresponding to the target of diagnosis is not stored in the diagnostic system 100, including the storage unit 13. In this embodiment, the predetermined condition is met if the diagnostic parameters (r, s, d) included in the diagnostic logic information are not stored. In other words, in this embodiment, if leakage determination cannot be made using the diagnostic logic, leakage determination is made using the estimated diagnostic logic.

[0041] Whether or not a trap is registered can be determined, for example, from the trap list. The trap list contains information (e.g., product number) of traps for which diagnostic logic can be executed. Specifically, when an inspector requests a steam trap diagnosis, the trap list is displayed on the liquid crystal display 17. The inspector can then use the input button 16 to select the type of trap to be diagnosed from the trap list. In this case, if there is no matching type in the trap list, the inspector can select "Not Registered". When "Not Registered" is selected, a predetermined condition is met.

[0042] In the estimation diagnostic logic of this embodiment, the presence or absence of leakage is calculated from a determination formula using estimated diagnostic information (estimated diagnostic parameters) corresponding to the diagnostic target, based on pressure and vibration values. In other words, the estimation diagnostic logic is a diagnostic method that performs leakage determination using a determination formula, similar to the diagnostic logic described above. Furthermore, the determination formula used in the estimation diagnostic logic is the same as the determination formula used in the diagnostic logic. In the estimation diagnostic logic of this embodiment, leakage determination is performed using the two determination formulas (1) and (2) (functions) described above. However, in the estimation diagnostic logic, diagnostic information calculated (estimated) by the calculation unit 14 is used, rather than the diagnostic information (diagnostic parameters (r,s,d)) used in the diagnostic logic. The calculation unit 14 estimates the diagnostic information by applying the estimation logic based on the trap information of the diagnostic target. The estimated diagnostic information will be described as estimated diagnostic parameters (r,s,d).

[0043] As described above, the estimated diagnostic parameters (r, s, d) are estimated by applying estimation logic based on the trap information of the target being diagnosed. First, let's explain the trap information.

[0044] The trap information in this embodiment includes characteristic information regarding the discharge flow rate of the drain to be diagnosed, and identification information for the operating type of the drain to be diagnosed. The characteristic information regarding the discharge flow rate of the drain is, for example, a combination of the pressure value (operating pressure difference) of the steam trap to be diagnosed and the discharge flow rate (drain volume) at this pressure value, and includes multiple sets of pressure values ​​and discharge flow rates. In this embodiment, three sets of (pressure value, discharge flow rate) are included in the characteristic information. In addition, in this embodiment, one of the three sets is (maximum pressure value, maximum discharge flow rate). (Maximum pressure value, maximum discharge flow rate) is a combination of the maximum pressure value of the steam trap and the maximum discharge flow rate, which is the discharge flow rate at this maximum pressure value. The discharge flow rate is the amount of drain discharged per predetermined time. The three sets of (pressure value, discharge flow rate) can be obtained from existing pressure value and discharge flow rate relationship (characteristic) data, rather than by measuring the drain to be diagnosed. Such data is publicly available for each steam trap, and is published as a graph like the one in Figure 5.

[0045] Figure 5 is a graph illustrating the relationship (characteristics) between the pressure value and discharge flow rate of a steam trap. The horizontal axis of the graph shows the operating pressure difference (pressure value), and the vertical axis shows the discharge flow rate. Lines A to E are graphs of steam traps from the same manufacturer and of the same operating type, but different models. For example, line A represents a steam trap model with a maximum discharge flow rate of 250 kg / h. For the steam trap on line A, when the pressure value (maximum pressure value) is 0.3 MPa, the discharge flow rate (maximum discharge flow rate) is 250 kg / h. Line B represents a steam trap model with a maximum pressure value of 0.5 MPa and a maximum discharge flow rate of 210 kg / h.

[0046] For example, in the case of a steam trap with a straight line A, the inspector only needs to acquire three sets of (pressure value, discharge flow rate) as characteristic information, including the combination (maximum pressure value, maximum discharge flow rate) = (0.3, 250). The remaining two sets of (pressure value, discharge flow rate) can be selected arbitrarily by referring to Figure 4. The characteristic information is then input into the diagnostic machine 10 by the inspector, for example.

[0047] Next, the estimation logic will be explained. In the estimation logic of this embodiment, diagnostic information (estimated diagnostic parameters (r, s, d)) is calculated (estimated) from a calculation formula based on the trap information of the target to be diagnosed. In other words, the estimation logic is a calculation method that calculates (estimates) diagnostic information using a calculation formula. In the estimation logic of this embodiment, for example, the diagnostic information is calculated (estimated) using the following four calculation formulas (3) to (6). In other words, the estimated diagnostic parameters (r, s, d) are calculated using the following four calculation formulas (3) to (6).

[0048]

number

[0049] The above (A1,B1,C1,D1,Z1), (A2,B2,C2,D2,E2,Z2), and (A3,B3,C3,D3,E3,Y3,Z3) are parameters (numerical values). Hereafter, the parameters (A1,B1,C1,D1,Z1), (A2,B2,C2,D2,E2,Z2), and (A3,B3,C3,D3,E3,Y3,Z3) may be referred to as calculated parameters. The values ​​of each calculated parameter differ depending on the operating type of the steam trap.

[0050] In the estimation logic, (α,β) is first calculated from at least two sets of (pressure value, discharge flow rate) from the three sets of (pressure value, discharge flow rate) characteristic information of the item to be diagnosed, and from calculation formula (3). Subsequently, the estimated diagnostic parameters (r,s,d) are calculated from the (maximum pressure value, maximum discharge flow rate) of the item to be diagnosed, (α,β), and calculation formulas (4) to (6) using calculation parameters corresponding to the operating type of the item to be diagnosed.

[0051] Furthermore, information regarding the estimation logic, such as calculation formulas (3) to (6) and calculation parameters for each operation type, is stored in the storage unit 13 as generated information. In this embodiment, the calculation parameters are values ​​predetermined using a learning model. The learning model will be described later.

[0052] The calculation unit 14 calculates the estimated diagnostic parameters (r, s, d) as described above, and then sets the estimated diagnostic parameters (r, s, d) in the determination expressions (1) and (2) to generate the estimated diagnostic logic. The calculation unit 14 then performs a leak determination on the target of the diagnosis by applying the estimated diagnostic logic. The estimated diagnostic logic uses the same method as the diagnostic logic described above, so a detailed explanation is omitted.

[0053] The calculation formulas (3) to (6) described above are merely examples and are not the only ones that can be used. Various formulas can be used in calculation formulas (3) to (6) as long as the diagnostic information (estimated diagnostic parameters (r, s, d)) is calculated based on the (pressure value, discharge flow rate) caused by the vibration.

[0054] Figures 6(A) to 6(C) are graphs showing examples of estimation results for (estimated diagnostic parameters (r,s,d)) using estimation logic. Figures 6(A) to 6(C) show the estimation results of calculating estimated diagnostic parameters (r,s,d) using estimation logic for multiple float-type steam traps in which diagnostic parameters (r,s,d) used in the diagnostic logic exist. The estimation logic uses the calculation formulas (3) to (6) described above. Figure 6(A) plots the estimated value and target value of estimated diagnostic parameter (r). Figure 6(B) plots the estimated value and target value of estimated diagnostic parameter (s). Figure 6(C) plots the estimated value and target value of estimated diagnostic parameter (d). In each graph, the horizontal axis shows the estimated value and the vertical axis shows the target value.

[0055] Furthermore, Figures 6(A) to 6(C) show the approximate straight line (dashed line) and the coefficient of determination (R) for each plotted point. 2The coefficient of determination (R) is also shown. It can be seen that the linear function of each approximation line is approximately y=x. 2 A value of approximately 0.6, close to 1, was obtained. Therefore, it can be seen that the estimated diagnostic parameters (r,s,d) have high estimation accuracy. Consequently, the estimation diagnostic logic using the estimated diagnostic parameters (r,s,d) can also perform highly accurate data leakage detection.

[0056] Furthermore, the calculation unit 14 controls the display on the liquid crystal display 17. For example, it displays the judgment result, etc., on the liquid crystal display 17.

[0057] The vibration measurement unit 11 and calculation unit 14 described above correspond to measurement means for measuring vibration values. Furthermore, the calculation unit 14 corresponds to a determination means for performing leakage detection and a generation means for generating estimation diagnostic logic.

[0058] The communication unit 15 has a communication circuit compatible with Bluetooth® and the like, and sends and receives various types of information to and from the terminal device (communication unit 24). For example, it transmits information related to the performed diagnosis (vibration value, temperature value, pressure value, vibration and temperature measurement data, judgment result, etc.) to the terminal device 20 (communication unit 24). Also, for example, the communication unit 15 receives trap list information, diagnostic logic information related to steam traps included in the trap list information, etc. from the terminal device 24. The received information is stored in the storage unit 13.

[0059] While this system utilizes Bluetooth® (registered trademark) communication, it is not limited to this method. Various communication methods, such as Wi-Fi, can be used as long as they are capable of communicating with the terminal device 20.

[0060] The input button (input means) 16 accepts input of various operations from the inspector. For example, it accepts input of a steam trap diagnostic request, input of trap information, etc. The input button 16 also serves as an acquisition means for accepting input of the current pressure value of the object to be diagnosed. The liquid crystal display 17 displays various information necessary for the work. For example, when a steam trap diagnostic request is received, the liquid crystal display 17 displays a list of traps (product number list) to allow the inspector to select the type of object to be diagnosed. After the diagnosis is completed, the liquid crystal display 17 displays the diagnostic results screen.

[0061] <Configuration of the terminal device> The terminal device 20 is, for example, a portable tablet computer. The terminal device 20 is carried by the inspector together with the diagnostic machine 10. The terminal device 20 transmits various information necessary for diagnosis to the diagnostic machine 10 by executing an application program in the storage unit 22. The terminal device 20 also transmits (uploads) information such as the leakage judgment result to the server device 30. The terminal device 20 has a touch panel 21, a storage unit 22, a calculation unit 23, and a communication unit 24, etc.

[0062] The touch panel 21 consists of a display unit such as an LCD panel and an input unit such as a touchpad. The touch panel 21 functions as an input / output interface for the inspector. The touch panel 21 also displays the diagnostic results screen for the steam trap. The diagnostic results screen displays information such as the leak detection result.

[0063] The memory unit 22 is a semiconductor memory (e.g., flash memory). The application program described above is stored in the memory unit 22. Diagnostic logic information is also stored in the memory unit 22 for each type of steam trap. Furthermore, the memory unit 22 stores a list of steam traps of the types that can be used for leakage detection by the diagnostic logic in the diagnostic system 100. The terminal device 20 obtains the latest diagnostic logic information from the server device 30 each time it is updated. The memory unit 22 also stores various diagnostic information, such as the judgment results received from the diagnostic machine 10.

[0064] The arithmetic unit 23 controls the display on the touch panel 21, etc. For example, the arithmetic unit 23 displays the diagnostic results screen on the touch panel 21. The arithmetic unit 23 includes a CPU, memory (RAM), etc.

[0065] The communication unit 24 has a communication circuit that supports Bluetooth® and Wi-Fi, etc. The communication unit 24 sends and receives various information with the diagnostic device 10 (communication unit 15). For example, if an inspector selects a steam trap, the communication unit 24 sends trap list information and diagnostic logic information related to the selected steam trap to the diagnostic device 10. Also, for example, the communication unit 24 receives information related to the performed diagnosis (vibration value, temperature value, pressure value, drain amount, vibration and temperature measurement data, judgment result, etc.) from the diagnostic device 10. The terminal device 20 stores the received information in the storage unit 22. The communication unit 24 also sends and receives various information with the server device 30 (communication unit 33) via the network N. For example, the communication unit 24 sends diagnostic information such as the result of a leak judgment to the server device 30. For example, information related to a day's worth of diagnoses is sent to the server device 30 all at once after the completion of the day's diagnostic work.

[0066] <Server configuration> The server device 30 transmits and receives various information with the terminal device 20 (communication unit 24) via the communication unit 33, and stores and manages various information of the steam trap, such as the judgment results. The server device 30 is, for example, a cloud server that exists on a network (on a cloud environment). The server device 30 has a storage unit 31, a processing unit 32, a communication unit 33, etc.

[0067] The storage unit 31 is a large-capacity storage device such as a hard disk. The storage unit 31 stores various types of steam trap information, such as diagnostic information. In this embodiment, in the diagnostic system 100, the information stored in the storage unit 31 of the server device 30 has the status of master data. Various types of steam trap information are stored, for example, associated with the steam trap management number.

[0068] The arithmetic unit 32 updates various types of information stored in the memory unit 31. For example, if it receives a new steam trap determination result, it adds the new determination result to the corresponding steam trap information. The arithmetic unit 32 includes a CPU, memory (RAM), etc.

[0069] The communication unit 33 has a communication circuit that supports Wi-Fi, etc. The communication unit 33 sends and receives various types of information to and from the terminal device 20 (communication unit 24) via the network N.

[0070] Furthermore, the server device 30 can be accessed from terminal devices other than the terminal device 20 (not shown), and various types of information stored in the storage unit 31 can be viewed from these other terminal devices.

[0071] <How to operate the diagnostic system> During the diagnostic process, the inspector moves around carrying the diagnostic device 10 and terminal device 20, diagnosing (determining leaks) each steam trap installed within the steam plant one by one. The following describes the diagnostic process, primarily focusing on tasks related to leak determination. Note that the memory unit 13 of the diagnostic device 10 already stores the trap list information received from the terminal device 20.

[0072] First, let's explain the case where the diagnostic logic to be diagnosed is registered.

[0073] When the inspector reaches the steam trap to be diagnosed, they first input a diagnostic request into the diagnostic device 10. This displays a list of traps (product number list) on the diagnostic device 10. The inspector uses the input button 16 to select the type of trap corresponding to the one to be diagnosed, and then inputs the pressure value of the trap to be diagnosed. The product number of the trap to be diagnosed is indicated, for example, on the outer surface of the trap. At this point, the diagnostic device 10 is ready to begin the diagnosis.

[0074] Subsequently, the diagnostic process begins when the inspector presses the probe 10a of the diagnostic device 10 against the steam trap being diagnosed. For example, as shown in Figure 3, the probe 10a of the diagnostic device 10 is pressed against the primary side of the steam trap ST. This causes the diagnostic device 10 to start measuring vibration, temperature, etc., and applies diagnostic logic to ultimately determine whether there is a leak (normal, leak). The result of the determination is then displayed on the liquid crystal display 17. The result of the determination is also stored in the storage unit 13 and transferred (transmitted) to the terminal device 20.

[0075] Next, we will explain what happens when the diagnostic logic to be diagnosed is not registered.

[0076] When the inspector reaches the steam trap to be diagnosed, they first input a diagnostic request into the diagnostic device 10. This displays a list of traps (product number list) on the diagnostic device 10. The inspector uses the input button 16 to select "Not Registered" from the trap list. This prompts the diagnostic device 10 to request the inspector to input the trap information to be diagnosed. The inspector uses the input button 16 to input the trap information to be diagnosed, and then inputs the pressure value to be diagnosed. For example, the inspector can launch a web browser on the terminal device 20 and search for data such as graphs related to the object to be diagnosed based on the type of object to be diagnosed (product number), as shown in Figure 5, to identify the trap information. The diagnostic device 10 then applies estimation logic to finally generate an estimated diagnostic logic. This allows the diagnostic device 10 to start the diagnosis.

[0077] Subsequently, as described above, the inspection begins when the inspector presses the probe 10a of the diagnostic device 10 against the steam trap to be inspected. This causes the diagnostic device 10 to start measuring vibration, temperature, etc., and applies the estimation diagnostic logic to finally determine whether there is a leak (normal, leak). The determination result is then displayed on the liquid crystal display 17. In this case, for example, the display may be made visible to indicate that the determination was made using estimation diagnostic logic. The determination result is also stored in the storage unit 13 and transferred (transmitted) to the terminal device 20. In this case, information that identifies the determination as being made using estimation diagnostic logic may be included in the determination result.

[0078] Figure 7 shows a flowchart of the diagnostic process. The diagnostic process mainly describes the steam trap leakage determination using diagnostic logic and estimated diagnostic logic. The calculation unit 14 starts executing the diagnostic process, for example, when a diagnostic request is received from an inspector. The storage unit 13 of the diagnostic machine 10 already stores trap list information and other data received from the terminal device 20.

[0079] Upon input of a diagnostic request, the diagnostic device 10 displays a list of traps (product number lists) as described above. The calculation unit 14 waits until the inspector inputs (selects) the type of item to be diagnosed (step S10). The inspector simply selects the type of item to be diagnosed from the list of traps. If the inspector selects the type of item to be diagnosed (step S10: YES), the calculation unit 14 determines whether the selection is "not registered" or not (step S11).

[0080] If "Not Registered" is not selected (Step S11: NO), the calculation unit 14 waits until the inspector inputs the current pressure value (Step S12). Although not shown in the diagram, the diagnostic device 10 displays a pressure value input screen.

[0081] If the inspector inputs the pressure value to be diagnosed (step S12: YES), the calculation unit 14 performs a calculation process (step S13). Specifically, the calculation unit 14 instructs the vibration measurement unit 11 to measure vibration and calculates the vibration value from the acquired measurement data, etc. As described above, the vibration measurement unit 11 starts measuring with the probe 10a of the diagnostic device 10 pressed against the steam trap.

[0082] Subsequently, the calculation unit 14 executes a determination process (step S14). In the determination process, a leak determination of the target to be diagnosed is performed by applying the diagnostic logic based on the acquired pressure value, vibration value, etc. In the diagnostic logic, as described above, a leak determination is performed by determination expressions (1) and (2) using the diagnostic parameters corresponding to the target to be diagnosed. The calculation unit 14 only needs to acquire diagnostic logic information related to the target to be diagnosed from the storage unit 13 based on the type of target to be diagnosed entered by the inspector.

[0083] Subsequently, the arithmetic unit 14 performs output processing (step S15) and terminates the diagnostic process. During output processing, information regarding the performed diagnosis is stored in the storage unit 13. Also, during output processing, the diagnostic results screen is displayed on the liquid crystal display 17.

[0084] Once the diagnostic process is complete, the diagnostic device 10 transmits diagnostic information, including the leak detection result, to the terminal device 20. The terminal device 20 receives this information and stores the date and time of receipt as the diagnostic date in the storage unit 22.

[0085] Furthermore, if the process returns to step S11 and "No registration" is selected (step S11: YES), the calculation unit 14 waits until the inspector inputs trap information (step S17). Although not shown in the diagram, the diagnostic machine 10 displays a trap information input screen. If the inspector inputs the trap information to be diagnosed (step S16: YES), the calculation unit 14 executes the generation process of the estimated diagnostic logic (step S17). In the generation process, diagnostic information (estimated diagnostic parameters) is estimated (calculated) by applying the estimation logic based on the trap information input by the inspector. In the estimation logic, as described above, the estimated diagnostic parameters are calculated by calculation formulas (3) to (6) using calculation parameters corresponding to the operation type of the diagnostic target. In addition, the generation process generates the estimated diagnostic logic using the calculated estimated diagnostic parameters. That is, judgment formulas (1) and (2) using the calculated estimated diagnostic parameters are generated.

[0086] Next, the calculation unit 14 waits for the current pressure value to be input from the inspector, similar to the process in step S12 (step S18). If the inspector inputs the pressure value to be diagnosed (step S18: YES), the calculation unit 14 performs the vibration value calculation process, similar to the process in step S13 (step S19).

[0087] Next, the calculation unit 14 executes a determination process (step S20). In the determination process, a leak determination of the target to be diagnosed is performed by applying the estimated diagnostic logic generated in the generation process above, based on the acquired pressure value, vibration value, etc. That is, a leak determination is performed by determination formulas (1) and (2) using the estimated diagnostic parameters corresponding to the target to be diagnosed.

[0088] Subsequently, the arithmetic unit 14 performs output processing (step S21) and terminates the diagnostic process. During output processing, information regarding the performed diagnosis is stored in the storage unit 13. Also, during output processing, the diagnostic results screen is displayed on the liquid crystal display 17.

[0089] Once the diagnostic process is complete, the diagnostic device 10 transmits diagnostic information, including the leak detection result, to the terminal device 20. The terminal device 20 receives this information and stores the date and time of receipt as the diagnostic date in the storage unit 22.

[0090] <Generating a calculation formula> In this embodiment, the calculation parameters of calculation formulas (4) to (6) of the estimation logic described above are determined using a learning model. Specifically, the calculation parameters are determined by training the learning model using training data. In this embodiment, the calculation parameters are determined for each type of steam trap operation.

[0091] For example, in this embodiment, the learning model program module is stored in the storage unit 31 of the server device 30, and the arithmetic unit 32 of the server device 30 functions as the learning model.

[0092] For example, as shown in Figure 8, the learning model DL is a neural network composed of an input layer DL11, a hidden layer DL12, and an output layer DL13. The input layer DL11 receives trap information from steam traps. The DL13 output layer outputs estimated diagnostic information (estimated diagnostic parameters) when it receives 1 trap information. In the DL12 hidden layer, various parameters (weights, etc.) are set so that the estimated diagnostic information (estimated diagnostic parameters) obtained by applying estimation logic is output from the output layer.

[0093] The training data used to train the above learning model includes trap information for the steam trap and diagnostic information used in the diagnostic logic for determining leaks in the steam trap. The trap information includes characteristic information regarding the drain discharge flow rate of the steam trap, and the operating type of the steam trap (e.g., identification information). The characteristic information regarding the drain discharge flow rate is a combination of the steam trap's pressure value (operating pressure difference) and the discharge flow rate (drain amount) at this pressure value, and includes multiple sets of pressure values ​​and discharge flow rates. In this embodiment, three sets of (pressure value, discharge flow rate) are included in the characteristic information. In addition, in this embodiment, one of the three sets is (maximum pressure value, maximum discharge flow rate). The diagnostic information is the values ​​of the diagnostic parameters (r, s, d) of the steam trap.

[0094] The characteristic information of the training data can be obtained from a graph (characteristic data) like the one shown in Figure 5 above. The diagnostic parameters (r, s, d) of the training data should be those actually used in the diagnostic logic. Then, one training data set can be generated by combining the steam trap operation type (identification information), three sets of (pressure value, discharge flow rate), and diagnostic parameters (r, s, d). In this embodiment, multiple training data sets are prepared for each operation type. The training data can be stored, for example, in the storage unit 31.

[0095] Next, we will describe the configuration in which the learning model performs a learning process using the aforementioned training data. In the learning process, the calculation unit 32 first acquires the training data. Next, the calculation unit 32 performs a generation process. In the generation process, the learning model DL is generated. The calculation unit 32 inputs the trap information of the training data into the learning model and obtains estimated diagnostic parameters as output (output process). In addition, in the generation process, the obtained estimated diagnostic parameters are compared with the correct values ​​(diagnostic parameters), and the parameters such as the weights of the hidden layer DL 12 are optimized (corrected) (correction process). That is, the obtained estimated diagnostic parameters are optimized to approximate the correct values. Specifically, the parameters of the calculation formulas (4) to (6) described above are optimized according to the operation type included in the input trap information. When the optimization is completed for all the training data in the generation process, the learning of the learning model is completed. As a result, in the learning model DL, the calculation formulas (4) to (6) are optimized for each operation type. Therefore, the calculation parameters are also optimized for each operation type. Then, in the diagnostic system 100 described above, calculation formulas (4) to (6) and calculation parameters for each operation type are stored as generated information in the diagnostic device 10 (storage unit 13).

[0096] In this embodiment, the learning model DL is used to determine (optimize) the calculation parameters of calculation formulas (4) to (6), but it is not used for calculating diagnostic information (estimated diagnostic parameters) by the diagnostic machine 10. However, it may be used. For example, the diagnostic machine 10 can be made to function with the aforementioned trained learning model DL, and the learning model DL can be used to calculate diagnostic information (estimated diagnostic parameters) in the generation process shown in Figure 7.

[0097] Furthermore, in this embodiment, the training data includes the operation type, but it does not have to be included. That is, training data including characteristic information and diagnostic information (diagnostic parameters) may be used. For example, multiple training data sets for only one operation type, such as a float operation, can be prepared and used to train the learning model. This generates a trained learning model corresponding to one operation type, and the calculation formulas (4) to (6) are optimized for that one operation type. The calculation parameters of the calculation formulas (4) to (6) optimized for that one operation type can then be stored in the memory. The same procedure can be followed for other operation types. Alternatively, multiple training data sets including characteristic information and diagnostic information (diagnostic parameters) can be prepared without distinguishing between operation types and used to train the learning model. This generates a trained learning model corresponding to multiple operation types, and the calculation formulas (4) to (6) corresponding to multiple operation types are optimized.

[0098] Furthermore, for the learning model DL, models based on various learning algorithms such as CNN (Convolutional Neural Network), multiple regression, Random Forest regression, and SVR (Support Vector Regression) may be applied.

[0099] As described above, if certain conditions are met, such as when the diagnostic logic for the steam trap to be diagnosed is not stored (registered), an estimated diagnostic logic is generated and the presence or absence of leakage in the target is determined by estimation. Therefore, even for steam traps where the diagnostic logic for gas leakage detection is unknown, highly accurate leakage detection is possible.

[0100] In the above embodiment, the predetermined condition was exemplified as the diagnostic logic not being stored (registered), but it is not limited to this. For example, the predetermined condition may be that the inspector requests (decides) to perform a diagnosis by estimation. In this case, even if the diagnostic logic is stored (registered), if the inspector requests it, the estimation diagnostic logic will be applied and the leakage determination will be performed. Furthermore, the predetermined condition may include multiple conditions. For example, the predetermined condition may be that the diagnostic logic is not stored (registered) AND the inspector requests (decides) to perform a diagnosis by estimation.

[0101] In the above-described embodiment, leakage is determined based on pressure and vibration values, but leakage may also be determined by considering the current drain volume (discharge flow rate). The drain volume (discharge flow rate) can be calculated by applying known methods to the measured vibration values, temperature, etc.

[0102] In the above embodiment, three sets of (pressure value, discharge flow rate) are used as characteristic information regarding the drain discharge flow rate, but two sets may also be used.

[0103] In the above embodiment, the characteristic information included in the trap information is input by the inspector, but it is not limited to this. The inspector may input image data of a graph as described above that corresponds to the object to be diagnosed. The diagnostic machine may then acquire the necessary characteristic information from the image data. In this case, the diagnostic machine may acquire the necessary characteristic information from the image data by applying a learning model. The learning model is training data that includes image data and the characteristic information acquired from this image data, and can be trained with training data of multiple types of steam traps.

[0104] In the above-described embodiment, various inputs related to the diagnosis, such as the current pressure value, are made to the diagnostic machine, but inputs may also be made to a terminal device. For example, inputs can be made by operating the control unit of the terminal device.

[0105] In the above-described embodiment, the inspector inputs the current pressure value of the steam trap, but this is not the only way to do so. For example, the diagnostic device may obtain the value by communicating with a pressure measuring device.

[0106] In the above-described embodiment, diagnostic logic information and the like are transmitted from the terminal device to the diagnostic machine, but they may also be stored in the memory of the diagnostic machine beforehand.

[0107] In the above embodiment, the diagnostic device performs the leak detection, but the system is not limited to this. The detection may also be performed by a terminal device or server device that constitutes the diagnostic system. In this case, for example, the diagnostic device transmits measurement data of ultrasonic vibration (vibration value), etc., to a device that performs leak detection. The device that performs leak detection can then perform leak detection by applying diagnostic logic and estimation diagnostic logic.

[0108] In the above-described embodiment, the diagnostic machine and the terminal device were distinguished, but the diagnostic machine and the terminal device may be integrated into a single device.

[0109] In the above embodiment, the diagnosis is performed using a diagnostic device carried by the inspector, but it does not have to be a portable configuration. For example, the diagnostic device may be installed in the steam trap and the diagnosis may be performed periodically. In this case, the judgment results and other information can be transmitted from the diagnostic device to the server device via a communication network.

[0110] In the above-described embodiment, a diagnostic system configuration having a server device, a terminal device, and a diagnostic device is explained. However, a configuration in which the diagnostic device functions as a standalone diagnostic device without a server device and a terminal device is also possible. In this case, for example, the diagnostic can be performed by operating the input buttons on the diagnostic device to input the type of steam trap, pressure value, etc., and the diagnostic information, such as the judgment result, can be stored in the memory of the diagnostic device. In this case, the diagnostic logic information, etc., can also be stored in the memory of the diagnostic device in advance. [Industrial applicability]

[0111] Even with steam traps where the diagnostic logic for gas leak detection is unknown, this method is useful for performing leak detection using the estimated diagnostic logic. [Explanation of Symbols]

[0112] 10 Diagnostic machines 11. Vibration measurement section 12 Temperature measurement part 13 Storage section 14 Arithmetic section 15 Communications Department 17 LCD display 20 Terminal devices 30 Server Devices 100 diagnostic systems DL Learning Model

Claims

1. A measuring means for measuring the vibration value of the steam trap that is the subject of diagnosis. means for obtaining the pressure value of the subject to diagnosis, A storage means for storing the diagnostic logic for determining gas leakage from a steam trap, for each type of steam trap. A determination means that performs a leak determination on the target to be diagnosed by applying the diagnostic logic to the target to be diagnosed based on the pressure value and the vibration value, An input receiving means that accepts input of trap information including characteristic information regarding the discharge flow rate of drain, A generation means that generates an estimated diagnostic logic for the target of diagnosis using the estimated diagnostic information obtained by applying an estimation logic based on the trap information of the target of diagnosis. Equipped with, The estimation logic is determined by a learning model trained on training data including the trap information of the steam trap and the diagnostic information used in the diagnostic logic of the steam trap. The determination means, in response to the fulfillment of predetermined conditions, applies the estimation diagnostic logic of the target to be diagnosed based on the pressure value and the vibration value to perform a leak determination of the target to be diagnosed. A diagnostic device for steam traps characterized by the following features.

2. The aforementioned predetermined conditions include the fact that the diagnostic logic of the subject to diagnosis is not stored. A diagnostic device for a steam trap according to claim 1.

3. The aforementioned trap information also includes information on the operating type of the steam trap. The estimation logic is determined for each operation type by a learning model trained on the training data. The generation means generates an estimated diagnostic logic for the target to be diagnosed by applying the estimated logic corresponding to the operation type of the target to be diagnosed. A diagnostic device for a steam trap according to claim 1.

4. In the diagnostic logic, based on the pressure value and the vibration value, the presence or absence of leakage is calculated from a determination formula using the diagnostic parameters, which are the diagnostic information corresponding to the object to be diagnosed. In the estimation diagnostic logic, based on the pressure value and the vibration value, the presence or absence of leakage is calculated from the determination formula using the estimated diagnostic parameters, which are the estimated diagnostic information corresponding to the object to be diagnosed. The generation means generates a determination expression for the estimated diagnostic logic of the target to be diagnosed using the estimated diagnostic parameters estimated by applying the estimation logic based on the characteristic information of the target to be diagnosed. A diagnostic device for a steam trap according to claim 1.

5. In the estimation logic, the estimated diagnostic parameters are calculated from a calculation formula determined by the learning model based on the characteristic information of the subject to be diagnosed. A diagnostic device for a steam trap according to claim 4.

6. The characteristic information of the object to be diagnosed is a combination of the pressure value of the steam trap to be diagnosed and the discharge flow rate at that pressure value, and includes multiple sets of pressure values ​​and discharge flow rates. A diagnostic device for a steam trap according to claim 5.

7. The aforementioned set of pressure values ​​and discharge flow rates includes a combination of the maximum pressure value and the maximum discharge flow rate, which is the discharge flow rate at the maximum pressure value. A diagnostic device for a steam trap according to claim 6.

8. The computer of the steam trap diagnostic device, A measuring means for measuring the vibration value of the steam trap that is the subject of diagnosis. means for obtaining the pressure value of the subject to diagnosis, A determination means that performs a leak determination for a target by applying the diagnostic logic for the target of diagnosis from among the diagnostic logic for gas leak determination stored for each type of steam trap, based on the pressure value and the vibration value. An input receiving means that accepts input of trap information including characteristic information regarding the discharge flow rate of drain, A generation means that generates an estimated diagnostic logic for the target of diagnosis using the estimated diagnostic information obtained by applying an estimation logic based on the trap information of the target of diagnosis. To make it function as, The estimation logic is determined by a learning model trained on training data including the trap information of the steam trap and the diagnostic information used in the diagnostic logic of the steam trap. The determination means, in response to the fulfillment of predetermined conditions, applies the estimation diagnostic logic of the target to be diagnosed based on the pressure value and the vibration value to perform a leak determination of the target to be diagnosed. A diagnostic program for steam traps characterized by the following:

9. A measuring means for measuring the vibration value of the steam trap that is the subject of diagnosis. means for obtaining the pressure value of the subject to diagnosis, A storage means for storing the diagnostic logic for determining gas leakage in steam traps, for each type of steam trap. A determination means that performs a leak determination on the target to be diagnosed by applying the diagnostic logic to the target to be diagnosed based on the pressure value and the vibration value, An input receiving means that accepts input of trap information including characteristic information regarding the discharge flow rate of drain, A generation means that generates an estimated diagnostic logic for the target of diagnosis using the estimated diagnostic information obtained by applying an estimation logic based on the trap information of the target of diagnosis. Equipped with, The estimation logic is determined by a learning model trained on training data including the trap information of the steam trap and the diagnostic information used in the diagnostic logic of the steam trap. The determination means, in response to the fulfillment of predetermined conditions, applies the estimation diagnostic logic of the target to be diagnosed based on the pressure value and the vibration value to perform a leak determination of the target to be diagnosed. A diagnostic system for steam traps characterized by the following features.

10. Training data including trap information containing characteristic information regarding the discharge flow rate of condensate from a steam trap, and diagnostic information used in a diagnostic logic for determining gas leakage from the steam trap, wherein an acquisition process is performed to acquire multiple training data sets. An output process that takes the trap information from the aforementioned training data as input and outputs estimated diagnostic information by applying estimation logic. A modification process is performed to modify the parameters of the estimation logic and learn so that the diagnostic information estimated in the output process approximates the diagnostic information of the training data. A method for generating a learning model that includes [specific features / details].

Images