Drying estimation device

Abstract

This invention provides a drying estimation device that can suppress energy consumption during vacuum drying using a vacuum degreasing and cleaning device. [Solution] The drying estimation device 66 is applied to a vacuum degreasing and cleaning apparatus that performs a vacuum cleaning process and a vacuum drying process on a workpiece 14, and comprises an acquisition unit 71 that acquires cleaning information, which is information relating to the cleaning state of the workpiece 14 in the vacuum cleaning process, and an estimation unit 73 that estimates the drying result of the workpiece 14 in the vacuum drying process performed after the vacuum cleaning process based on the cleaning information.

Description

Technical Field

[0001] The present invention relates to a drying estimation device applied to a vacuum degreasing and cleaning device.

Background Art

[0002] For example, when cleaning a workpiece using a vacuum degreasing and cleaning device, it may be used for pre-cleaning or post-cleaning in a heat treatment process.

[0003] For example, in Patent Document 1, it is described that in a vacuum degreasing and cleaning device, shower cleaning, immersion bubbling cleaning, high-temperature spraying cleaning, and vacuum drying are performed.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, since vacuum drying is performed by evacuating, it is impossible to confirm whether the drying of the workpiece is completed unless the operation of vacuum drying ends. For example, after the operation of vacuum drying ends, the drying state of the workpiece is visually confirmed by an operator. If the drying result in vacuum drying is unknown, the vacuum drying may be set for an unnecessarily long time, and the energy consumption may increase.

[0006] In view of the above problems, an object of the present invention is to provide a drying estimation device capable of suppressing the energy consumption in vacuum drying using a vacuum degreasing and cleaning device.

Means for Solving the Problems

[0007] To solve the above problems, the drying estimation device according to the present invention is a drying estimation device applied to a vacuum degreasing and cleaning apparatus that performs a vacuum cleaning process and a vacuum drying process on a workpiece, and comprises an acquisition unit that acquires cleaning information which is information relating to the cleaning state of the workpiece in the vacuum cleaning process, and an estimation unit that estimates the drying result of the workpiece in the vacuum drying process performed after the vacuum cleaning process based on the cleaning information.

[0008] Furthermore, in the drying estimation device, the estimation unit estimates the probability that drying will be completed and the time required for drying as the drying result.

[0009] Furthermore, in the drying estimation apparatus, the cleaning information is information relating to at least one of the following indicators: the amount of temperature change of the cleaning agent supplied to the workpiece in the vacuum cleaning process, and the immersion time during which the workpiece is immersed in the cleaning agent in the vacuum cleaning process.

[0010] Furthermore, in the drying estimation apparatus, the acquisition unit acquires work information, which is information relating to at least one of the following indicators: the mass of the workpiece, the volume of the workpiece, the surface area of ​​the workpiece, and the number of workpieces. The estimation unit estimates the drying result based on the work information and the cleaning information.

[0011] Furthermore, the drying estimation device further includes a learning unit that trains a model on the drying results of the workpiece corresponding to the vacuum drying process based on the inflection point of the pressure in the space where the workpiece is placed during the vacuum drying process, and the estimation unit uses the trained model to estimate the drying results of the workpiece.

[0012] Furthermore, in the drying estimation device, the learning unit uses the drying result as an estimated value of the probability that drying will be completed, obtained from the pressure fluctuations of the pressure, when there is an inflection point during the steam exhaust stage in the vacuum drying process, and uses the drying result as a drying failure when there is no inflection point during the steam exhaust stage, and teaches the model accordingly. [Effects of the Invention]

[0013] According to the drying estimation apparatus of the present invention, it is possible to suppress energy consumption in vacuum drying using a vacuum degreasing and cleaning apparatus. [Brief explanation of the drawing]

[0014] [Figure 1] This figure schematically shows an example of the overall configuration of a vacuum degreasing and cleaning apparatus equipped with a drying estimation device according to an embodiment of the present invention. [Figure 2] Figure 1 is a cross-sectional view showing an example of the main unit's configuration. [Figure 3] This figure shows an example configuration of the drying estimation system in the vacuum degreasing and cleaning apparatus shown in Figure 1. [Figure 4] This figure shows an example of a specific configuration of the first model in the estimation section. [Figure 5] This figure shows an example of a specific configuration of the second model in the estimation section of Figure 3. [Figure 6] This figure shows an example of the pressure and pressure fluctuations inside the cleaning chamber during the vacuum drying process of the vacuum degreasing and cleaning apparatus shown in Figure 1. [Figure 7] This figure shows another example of the pressure and pressure fluctuations inside the cleaning chamber during the vacuum drying process of the vacuum degreasing and cleaning apparatus shown in Figure 1. [Figure 8] Figure 3 is a flowchart showing an example of the process flow for estimating drying results in the drying estimation device. [Figure 9] Figure 3 is a flowchart showing an example of the learning process flow in the drying estimation device. [Modes for carrying out the invention]

[0015] Embodiments of the present invention will be described below with reference to the attached drawings. In order to facilitate understanding of the explanation, the same reference numerals will be used for identical components in each drawing as much as possible, and redundant explanations will be omitted as appropriate.

[0016] ===Implementation Method=== ≪Overall Structure≫ FIG. 1 is a diagram schematically showing an example of the overall configuration of a vacuum degreasing and cleaning apparatus 100 including a drying estimation device 66 according to an embodiment of the present invention. FIG. 2 is a cross-sectional view showing a configuration example of the main body 10 of FIG. 1.

[0017] Before and after heat treatment such as quenching, cleaning is performed using the vacuum degreasing and cleaning apparatus 100.

[0018] As shown in FIGS. 1 and 2, the vacuum degreasing and cleaning apparatus 100 is provided with a main body 10. The main body 10 has a cleaning chamber 1 having opening / closing doors 3 at both ends, and a hollow cleaning agent tank 4 surrounding the cleaning chamber 1. The workpiece 14 can be carried in and out of the cleaning chamber 1 through the opening / closing door 3. For example, a plurality of workpieces 14 can be carried into the cleaning chamber 1. Inside the cleaning chamber 1, three-stage cleaning of shower cleaning, immersion cleaning, and high-temperature spray cleaning is possible. Also, vacuum drying is possible inside the cleaning chamber 1. That is, cleaning and drying are performed in the same chamber.

[0019] Furthermore, a shower cleaning nozzle 5 is positioned at the top of the cleaning chamber 1, and a bubbling nozzle 6 connected to an N2 gas supply device 19 (note that the "2" in N2 gas is a subscript) is positioned at the bottom. A workpiece stand 18 is provided on the bottom surface of the cleaning chamber 1, and workpieces 14 are brought in and placed on it through the opening / closing door 3. The cleaning agent tank 4 contains a cleaning agent 17 as a cleaning liquid and is connected to the shower cleaning nozzle 5 via a shower line 16. The cleaning agent tank 4 is connected to a pump 15 via piping 30, and to the top of the cleaning agent tank 4 via piping 37 and piping 35. The lower part of the cleaning chamber 1 is connected to the cleaning agent tank 4 via piping 31 and an on-off valve 51. The lower part of the cleaning chamber 1 is connected to the waste liquid tank 11 via piping 31 and an on-off valve 52. The waste liquid tank 11 is connected to the distillation regenerator 9 via piping 32. The distillation regenerator 9 is connected to the regeneration tank 8 via piping 33. The regeneration tank 8 is connected to the shower cleaning nozzle 5 via piping 21. The distillation regenerator 9 is connected to the top of the detergent tank 4 via piping 36 and piping 35. The top of the cleaning chamber 1 is connected to the vacuum pump 7 via piping 34. The detergent tank 4 and the distillation regenerator 9 are connected to the heat exchanger 13 via piping 35, piping 36, and piping 37.

[0020] The vacuum degreasing and cleaning apparatus 100 performs an exhaust process, a primary shower process, an immersion process, a secondary shower process, and a vacuum drying process to clean and dry the workpiece 14. In particular, the primary shower process, the immersion process, and the secondary shower process are cleaning processes (triple cleaning processes). As will be described later, the cleaning process is performed under vacuum exhaust, and therefore it is a vacuum cleaning process (reduced pressure cleaning process).

[0021] During the exhaust process, the workpiece 14 is brought into the washing chamber 1 through the opening / closing door 3. Then, the opening / closing door 3 is closed, and the washing chamber 1 is evacuated by the vacuum pump 7.

[0022] In the primary shower process, the cleaning agent 17 from the cleaning agent tank 4 is sprayed from the upper shower cleaning nozzle 5 through the shower line 16 into the depressurized cleaning chamber 1, and the workpiece 14 is shower-cleaned. The cleaning agent 17 is, for example, a hydrocarbon-based cleaning agent.

[0023] During the immersion process, valves 52 and 53 are closed, and valve 51 is opened. Then, the lower part of the cleaning chamber 1 and the cleaning agent tank 4 are connected via pipes 30 and 31, and the cleaning agent 17 fills the depressurized cleaning chamber 1 to a set position, and the workpiece 14 is immersed. The workpiece 14 is heated by the immersion. Then, valve 51 is closed, and the bubbling nozzle 6 connected to the N2 gas supply device 19 is activated, and N2 gas is injected from the bubbling nozzle 6. The cleaning agent 17 is agitated by the bubbles, and the workpiece 14 is immersed for a predetermined time while bubbling cleaning is performed. After the predetermined time has elapsed, valve 51 is opened, and the cleaning agent 17 that has filled the cleaning chamber 1 is returned to the cleaning agent tank 4 via pipes 30 and 31 while the N2 gas supply device 19 is activated.

[0024] In the secondary shower process, a high-temperature, high-purity cleaning agent 17 from a regeneration tank 8 is sprayed from a shower cleaning nozzle 5 into a depressurized cleaning chamber 1 via piping 21, and the workpiece 14 is given a final cleaning by high-temperature spray cleaning.

[0025] In the vacuum drying process, the workpiece 14 to which the cleaning agent 17 is attached and the inside of the cleaning chamber 1 are evacuated by a vacuum pump 7, and vacuum drying is performed.

[0026] The vacuum-dried workpiece 14 is removed through the opening / closing door 3. The above exhaust process, primary shower process, immersion process, secondary shower process, and vacuum drying process constitute one cycle, and the cycle is repeatedly executed.

[0027] Figure 3 shows an example configuration of a drying estimation system 60 applied to a vacuum degreasing and cleaning apparatus 100. As shown in Figure 3, the drying estimation system 60 includes a load cell 61, a camera 62, a pressure sensor 63, a temperature sensor 64, a timer 65, and a drying estimation device 66. The timer 65 may be mounted on the drying estimation device 66.

[0028] The load cell 61 is a device that measures the mass W of a workpiece 14 before it is brought into the washing chamber 1. For example, the load cell 61 measures the mass W of one workpiece 14. Alternatively, the load cell 61 may measure the total mass of multiple workpieces 14 that are brought into the washing chamber 1. The measurement result of the mass W by the load cell 61 is output to the drying estimation device 66.

[0029] Camera 62 is a device that photographs the workpiece 14 before it is brought into the washing chamber 1. The image of the workpiece 14 that is photographed is output to the drying estimation device 66.

[0030] The pressure sensor 63 is a device that measures the pressure p inside the washing chamber 1. Specifically, the pressure sensor 63 measures the pressure p inside the washing chamber 1 during the vacuum drying process. The measurement result of pressure p is output to the drying estimation device 66.

[0031] The temperature sensor 64 is a device that measures the temperature of the cleaning agent 17 relative to the cleaning chamber 1. Specifically, the temperature sensor 64 measures the temperature change (temperature change amount ΔT) of the cleaning agent 17 supplied to the workpiece 14 during the immersion process. For example, the temperature sensor 64 measures the temperature of the cleaning agent 17 before it is supplied to the workpiece 14 during the immersion process (temperature of the cleaning agent 17 supplied to the cleaning chamber 1) and the temperature of the cleaning agent 17 after it is supplied to the workpiece 14 during the immersion process (temperature of the cleaning agent 17 discharged from the cleaning chamber 1). In this case, the temperature change amount ΔT is the difference between the temperature of the cleaning agent 17 before it is supplied to the workpiece 14 and the temperature of the cleaning agent 17 after it is supplied to the workpiece 14. For example, the temperature sensor 64 is installed at the supply port and discharge port of the cleaning agent 17 in the cleaning chamber 1. Note that the specific measurement method and measurement location are not limited as long as the temperature change of the cleaning agent 17 related to the amount of heat exchange between the cleaning agent 17 and the workpiece 14 can be measured. For example, since heat is transferred from the cleaning agent 17 to the workpiece 14, the temperature change ΔT of the cleaning agent 17 becomes the temperature drop. The temperature measurement result is output to the drying estimation device 66.

[0032] The timer 65 measures the time (immersion time t) that the workpiece 14 is immersed in the cleaning agent 17 during the immersion process. The immersion time t may be the time from the start to the end of the immersion process, or it may be the time while the workpiece 14 is actually immersed. The measurement result of the immersion time t is output to the drying estimation device 66.

[0033] The drying estimation device 66 is an information processing device (computer) that estimates the drying result of the workpiece 14 in the vacuum degreasing and cleaning device 100. In this embodiment, the drying estimation device 66 estimates the drying result using AI. However, estimation is not limited to the use of AI. For example, the drying estimation device 66 is composed of a CPU, memory, communication interface, storage device, operating device, and display device.

[0034] ≪Functional configuration≫ Figure 3 shows an example of the various functions of the drying estimation device 66. As shown in Figure 3, the drying estimation device 66 mainly comprises an acquisition unit 71, an image analysis unit 72, an estimation unit 73, a notification unit 74, a calculation unit 75, and a learning unit 76.

[0035] The acquisition unit 71 acquires work information and cleaning information.

[0036] The work information is information about the workpiece 14 before cleaning. Specifically, the acquisition unit 71 acquires information about at least one of the following indicators: the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of the workpiece 14. The mass W of the workpiece 14, the volume V of the workpiece 14, and the surface area A of the workpiece 14 are indicators corresponding to a single workpiece 14, for example, but they may also be indicators that represent multiple workpieces 14 together. In this embodiment, the acquisition unit 71 acquires the measurement result of the mass W of the workpiece 14 by the load cell 61 and an image captured by the camera 62 as work information. The measurement result of the mass W by the load cell 61 is information about the mass W of the workpiece 14. The image captured by the camera 62 is information about the volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of the workpiece 14. The information about the mass W of the workpiece 14 is output to the estimation unit 73. The image captured by the camera 62 is output to the image analysis unit 72.

[0037] The cleaning information is information regarding the cleaning state of the workpiece 14 during the vacuum cleaning process. Specifically, the cleaning information is information regarding at least one of the following indicators: the temperature change ΔT of the cleaning agent 17 supplied to the workpiece 14 during the vacuum cleaning process, and the immersion time t during which the workpiece 14 is immersed in the cleaning agent 17 during the vacuum cleaning process. In this embodiment, the acquisition unit 71 acquires the measurement results of the temperature sensor 64 and the measurement results of the timer 65 as cleaning information. The measurement result of the temperature sensor 64 indicates the temperature change ΔT of the cleaning agent 17 supplied to the workpiece 14 during the immersion process in the vacuum cleaning process. The measurement result of the timer 65 indicates the time during which the workpiece 14 was immersed in the cleaning agent 17 during the immersion process in the vacuum cleaning process. Information regarding the temperature change ΔT and the immersion time t is output to the estimation unit 73. Note that the temperature change ΔT and the immersion time t are related to the amount of heat exchange between the cleaning agent 17 and the workpiece 14. In other words, the temperature change ΔT and immersion time t are related to the temperature state of the workpiece 14 at the end of the vacuum cleaning process (or at the start of the vacuum drying process). Therefore, any other information besides the temperature change ΔT and immersion time t may be used as cleaning information, as long as it is a parameter related to the temperature state of the workpiece 14 at the end of the vacuum cleaning process.

[0038] Furthermore, the acquisition unit 71 acquires the measurement result of the pressure p inside the cleaning chamber 1 from the pressure sensor 63. The measurement result is time-series data of the pressure p inside the cleaning chamber 1. The measurement result of pressure p is output to the calculation unit 75.

[0039] The image analysis unit 72 analyzes the images captured by the camera 62 acquired by the acquisition unit 71. Specifically, the image analysis unit 72 analyzes the images of the workpiece 14 to identify the volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of the workpiece 14. Multiple types of images may be acquired from the camera 62 for this analysis. The information relating to the identified volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of the workpiece 14 is output to the estimation unit 73.

[0040] The estimation unit 73 estimates the drying result of the workpiece 14 in the vacuum drying process performed after the vacuum cleaning process, based on the workpiece information and cleaning information. Specifically, the estimation unit 73 estimates the drying result using the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number N of workpieces 14, the temperature change ΔT, and the immersion time t.

[0041] The estimation unit 73 estimates the probability that drying will be completed and the time required for drying as drying results. For this purpose, the estimation unit 73 has a first model M1 and a second model M2.

[0042] The first model M1 is a model that estimates the probability that a workpiece 14 that has completed the vacuum cleaning process will be successfully dried (completed to a predetermined drying state) in the vacuum drying process. Figure 4 shows an example of the specific configuration of the first model M1. As shown in Figure 4, the first model M1 is composed of a neural network. That is, the first model M1 has an input layer, an intermediate layer, and an output layer. In the input layer, the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number of workpieces N, the temperature change ΔT, and the immersion time t are input. Then, the probability that drying will be completed is output from the output layer via the machine learning-trained intermediate layer. The output layer outputs information indicating the probability, such as 0 to 1, using an activation function, etc. When the output is 0, it indicates that the probability of drying being completed is 0%, and when the output is 1, it indicates that the probability of drying being completed is 100%. Although Figure 4 describes an example of the configuration of the first model M1, the first model M1 is not limited to the configuration shown in Figure 4.

[0043] The second model M2 is a model that estimates the time required for a workpiece 14 that has completed the vacuum cleaning process to dry in the vacuum drying process (the time required to reach a predetermined dry state). Figure 5 shows an example of the specific configuration of the second model M2. As shown in Figure 5, the second model M2 is also constructed using a neural network. That is, the second model M2 has an input layer, an intermediate layer, and an output layer. In the input layer, the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number of workpieces N, the temperature change ΔT, and the immersion time t are input. Then, the time required for drying is output from the output layer via the machine learning-processed intermediate layer. The output layer outputs a numerical value as the time required for drying using an identity map or the like. For example, an output of 10 indicates that the time required for drying is 10 minutes. Although Figure 5 describes an example of the configuration of the second model M2, the second model M2 is not limited to the configuration shown in Figure 5.

[0044] Figures 4 and 5 show an example where the first model M1 and the second model M2 are composed of neural networks, but the first model M1 and the second model M2 are not limited to being composed of neural networks.

[0045] In this way, the estimation unit 73 uses various information obtained before the start of the vacuum drying process to estimate the probability of drying being completed and the time required for drying. Preferably, the estimation is performed before the start of the vacuum drying process, or as soon as possible after the start of the vacuum drying process (before the end of the drying process).

[0046] The first model M1 and the second model M2 are trained by the learning unit 76, which will be described later, and the estimation unit 73 uses the trained first model M1 and second model M2 to estimate the drying result of the workpiece 14.

[0047] Returning to Figure 3, the notification unit 74 notifies the estimation result from the estimation unit 73. For example, the notification unit 74 displays the estimation result on a predetermined display device and notifies the worker, etc. Specifically, the notification unit 74 notifies the worker, etc. of the probability that drying will be completed. For example, the notification unit 74 notifies that the workpiece 14 can be dried in the vacuum drying process if the probability of drying being completed estimated by the estimation unit 73 is equal to or greater than a set value. For example, the notification unit 74 notifies "Drying is possible" if the probability is equal to or greater than the set value, and notifies "Drying is not possible" if the probability is less than the set value. The notification unit 74 may also notify the worker, etc. of the probability that drying will be completed.

[0048] Furthermore, the notification unit 74 notifies the worker or other personnel of the time required for drying. Preferably, the time required for drying is notified when it is possible to dry the workpiece 14 in the vacuum drying process (when the probability is above a set value). For example, if the time required for drying estimated by the estimation unit 73 is 10 minutes, the notification unit 74 notifies, "The estimated time required is 10 minutes." In other words, the notification unit 74 notifies, "Drying is possible," and also notifies, "The estimated time required is 10 minutes," when the probability is above a set value. The notification unit 74 may also notify, "Drying is not possible," and provide countermeasures such as "Please stop the cycle" if the probability of drying being completed is below a set value.

[0049] Notifications from the notification unit 74 allow workers to understand the probability of drying being completed (whether drying is possible or not) and the time required for drying. For example, if drying is possible and the time required for drying is shorter than expected, workers can consider shortening the cycle time. If drying is possible and the time required for drying is longer than expected, workers can consider reducing the number of workpieces 14 N. If drying is not possible, workers can take measures such as stopping the equipment. In other words, they can consider vacuum drying tailored to the workpieces 14 or take measures before vacuum drying.

[0050] The calculation unit 75 calculates the pressure fluctuation Δp from the pressure p in the space where the workpiece 14 is placed during the vacuum drying process. Specifically, the calculation unit 75 calculates the pressure fluctuation Δp from the time-series data of the pressure p in the washing chamber 1. The pressure fluctuation Δp (Pa / s) is a parameter that indicates the amount of change in pressure p (Pa) per unit time. Various indicators can be used for the pressure fluctuation Δp as long as they can show the fluctuation (change) of pressure p. The calculation unit 75 may also smooth the pressure fluctuation Δp by taking a moving average.

[0051] Figures 6 and 7 show examples of the pressure p (Pa) and the pressure fluctuation Δp (Pa / s) in the washing chamber 1 during the vacuum drying process. Figure 6 shows the change in state when drying is good (drying is complete) during the vacuum drying process. Figure 7 shows the change in state when drying is poor (drying is not complete) during the vacuum drying process.

[0052] Specifically, the vacuum drying process is divided into the following stages: the exhaust preparation stage, the vacuum exhaust stage, the oil and grease evaporation stage, and the steam exhaust stage. The exhaust preparation stage is the stage in which preparations are made for vacuum exhaust. The vacuum exhaust stage is the stage in which the inside of the washing chamber 1 is evacuated. The oil and grease evaporation stage is the stage in which oils and greases evaporate as the inside of the washing chamber 1 is evacuated. The steam exhaust stage is the stage in which steam and other substances generated inside the washing chamber 1 are exhausted. The vacuum drying process is executed according to pre-set operations and times.

[0053] The calculation unit 75 then calculates the pressure fluctuation Δp corresponding to the pressure p shown in Figures 6 and 7. The pressure fluctuation Δp shows different trends depending on whether the drying is good (Figure 6) or poor (Figure 7). Specifically, when the drying is good (Figure 6), the inflection point of pressure p (corresponding to point H in the pressure fluctuation Δp in Figure 6) clearly appears at time t1 during the steam exhaust stage. On the other hand, when the drying is not good or is poor (Figure 7), even if an inflection point appears during the steam exhaust stage, it is not as clear as in the case of Figure 6, or no inflection point appears at all. Thus, it is possible to determine whether the drying is good or bad based on whether or not an inflection point is clearly present. Note that the worse the drying, the less likely an inflection point is to appear (or be clearly present) during the steam exhaust stage.

[0054] Furthermore, when the pressure fluctuation Δp converges to a constant value, the vacuum drying process (steam exhaust stage) is completed. Figures 6 and 7 show examples where the pressure fluctuation Δp converges to a value Δpc and the vacuum drying process is completed. Figure 6 shows an example where the vacuum drying process is completed at time t2, and Figure 7 shows an example where the vacuum drying process is completed at time t3. Note that time t2 and time t3 may be the same or different times.

[0055] Returning to Figure 3, the pressure fluctuation Δp calculated by the calculation unit 75 is output to the learning unit 76.

[0056] The learning unit 76 performs learning (machine learning) on ​​the first model M1 and the second model M2. Specifically, the learning unit 76 teaches the first model M1 the drying result corresponding to the vacuum drying process based on the inflection point of pressure p in the vacuum drying process. When there is an inflection point during the steam exhaust stage in the vacuum drying process, the learning unit 76 teaches the first model M1 an estimated value of the drying result (probability of drying completion) obtained from the pressure fluctuation Δp. For example, when there is an inflection point during the steam exhaust stage, the learning unit 76 learns the output of the first model M1 to be the value obtained by dividing the pressure fluctuation Δp at the inflection point (for example, time t1 in Figure 6) by a threshold (reference value). The threshold is set as a fixed value within a range that does not exceed the pressure fluctuation Δp at the inflection point, but is not extremely low, based on previously acquired test data. For example, the threshold is set as the lower limit (expected lower limit) of the pressure fluctuation Δp corresponding to the inflection point in the steam exhaust stage. In other words, the better the drying is and the clearer the inflection point appears during the steam exhaust stage, the closer the pressure fluctuation Δp corresponding to the inflection point will be to the threshold value. Therefore, the value obtained by dividing the pressure fluctuation Δp (e.g., absolute value) at the inflection point by the threshold value (e.g., absolute value) (the value obtained by dividing the pressure fluctuation Δp by the threshold) becomes an estimate of the probability that drying will be completed. For example, the closer the pressure fluctuation Δp at the inflection point is to the threshold, the closer it is to 1 (i.e., 100%), indicating that drying is easy (good drying). Also, for example, the further the pressure fluctuation Δp at the inflection point is from the threshold, the closer it is to 0 (i.e., 0%), indicating that drying is difficult (poor drying). Furthermore, the learning unit 76 trains the first model M1 to assume that drying is poor if there is no inflection point during the steam exhaust stage. For example, the learning unit 76 trains the first model M1 so that the output is 0 (i.e., the probability of drying being completed is 0%) if there is no inflection point during the steam exhaust stage.

[0057] Furthermore, the learning unit 76 teaches the second model M2 that the time required for drying is the time it takes for the pressure fluctuation Δp to converge to a constant value Δpc after the start of the vacuum drying process. For example, the learning unit 76 determines that the pressure fluctuation Δp has converged to a constant value when the fluctuation range of the pressure fluctuation Δp falls below a set value.

[0058] <<Processing Flow>> Figure 8 is a flowchart showing an example of the drying result estimation process according to this embodiment. Each of the following steps is started when the vacuum degreasing and cleaning apparatus 100 starts operating. The drying result estimation process uses the learned first model M1 and second model M2. The order and content of each of the following steps can be changed as appropriate.

[0059] (Step SP10) The acquisition unit 71 acquires the measurement result of the mass W of the workpiece 14 by the load cell 61 and the image captured by the camera 62. For example, before the workpiece 14 is brought into the washing chamber 1, the measurement by the load cell 61 and the image captured by the camera 62 are performed, and the acquisition unit 71 acquires the information. Then the process moves on to step SP11.

[0060] (Step SP11) The image analysis unit 72 analyzes the images captured by the camera 62 to determine the volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of the workpiece 14. Then, the process proceeds to step SP12.

[0061] (Step SP12) The acquisition unit 71 acquires the measurement result (temperature change ΔT) from the temperature sensor 64 and the measurement result (immersion time t) from the timer 65. For example, the temperature sensor 64 and timer 65 perform measurements in accordance with the vacuum cleaning process (especially the immersion process), and the acquisition unit 71 acquires the information. Steps SP11 and SP12 may be executed in parallel. Then, the process moves on to step SP13.

[0062] (Step SP13) The estimation unit 73 uses the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number of workpieces N, the temperature change ΔT, and the immersion time t to estimate the probability that drying will be completed according to the first model M1. Then, the process proceeds to step SP14.

[0063] (Step SP14) The estimation unit 73 uses the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number of workpieces N, the temperature change ΔT, and the immersion time t to estimate the drying time using the second model M2. Steps SP13 and SP14 may be executed in parallel. Then the process proceeds to step SP15.

[0064] (Step SP15) The notification unit 74 notifies the estimation result from the estimation unit 73. Then the process ends.

[0065] As described above, the drying result estimation process is performed. It is preferable that the estimation result be notified before the start of the vacuum drying process, or as soon as possible after the start of the vacuum drying process.

[0066] Figure 9 is a flowchart showing an example of the learning process flow according to this embodiment. Each of the following steps is started, for example, when an execution instruction is given by an operator. The vacuum degreasing and cleaning device 100 is also operated in parallel with the learning process. The order and content of each of the following steps can be changed as appropriate.

[0067] (Step SP20) The acquisition unit 71 acquires the measurement result of the mass W of the workpiece 14 by the load cell 61 and the image captured by the camera 62. For example, before the workpiece 14 is brought into the washing chamber 1, the measurement by the load cell 61 and the image captured by the camera 62 are performed, and the acquisition unit 71 acquires the information. Then the process moves on to step SP21.

[0068] (Step SP21) The image analysis unit 72 analyzes the images captured by the camera 62 to determine the volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of the workpiece 14. Then, the process proceeds to step SP22.

[0069] (Step SP22) The acquisition unit 71 acquires the measurement result (temperature change amount ΔT) from the temperature sensor 64 and the measurement result (immersion time t) from the timer 65. For example, the temperature sensor 64 and timer 65 perform measurements in accordance with the vacuum cleaning process (especially the immersion process), and the acquisition unit 71 acquires the information. Steps SP21 and SP22 may be executed in parallel. Then, the process moves on to step SP23.

[0070] (Step SP23) The learning unit 76 takes the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number of workpieces N, the temperature change ΔT, and the immersion time t as input to the first model M1, which is the learning target, and outputs (infers) the probability that drying will be completed. Then the process moves on to step SP24.

[0071] (Step SP24) The learning unit 76 receives the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, the number of workpieces N, the temperature change ΔT, and the immersion time t as input to the second model M2, which is the learning target, and outputs (infers) the time required for drying. Steps SP23 and SP24 may be executed in parallel. Then the process moves on to step SP25.

[0072] (Step SP25) The acquisition unit 71 acquires the measurement result of the pressure p in the cleaning chamber 1 from the pressure sensor 63 when the vacuum drying process in the vacuum degreasing cleaning apparatus 100 begins. Then the process proceeds to step SP26.

[0073] (Step SP26) The calculation unit 75 calculates the pressure fluctuation Δp in accordance with the time-series data of the pressure p inside the washing chamber 1 that was acquired. Then, the process proceeds to step SP27.

[0074] (Step SP27) The learning unit 76 determines whether the pressure fluctuation Δp has converged to a constant value since the start of the vacuum drying process. If the pressure fluctuation Δp has not converged to a constant value, the process returns to step SP25 and is executed again. If the pressure fluctuation Δp has converged to a constant value, the process proceeds to step SP28. If the pressure fluctuation Δp has converged to a constant value, the vacuum drying process in the vacuum degreasing and cleaning apparatus 100 is completed.

[0075] (Step SP28) The learning unit 76 determines whether or not there is an inflection point during the steam exhaust stage in the vacuum drying process. If an inflection point is present, the process proceeds to step SP29. If there is no inflection point, the process proceeds to step SP30.

[0076] (Step SP29) The learning unit 76 learns the first model M1 by backpropagation, using the output of the first model M1 as training data, which is the value obtained by dividing the pressure fluctuation Δp at the inflection point by a threshold value. For example, in step SP23, the learning unit 76 performs backpropagation using the error between the probability that drying is completed output by the first model M1 and the training data (i.e., 1). Then, the process moves on to step SP31.

[0077] (Step SP30) The learning unit 76 learns the first model M1 by backpropagation, using the training data of the output of the first model M1 as 0 (i.e., a probability of drying being completed of 0%). For example, in step SP23, the learning unit 76 performs backpropagation based on the error between the probability of drying being completed output by the first model M1 and the training data (i.e., 0). Then, the process moves on to step SP31.

[0078] (Step SP31) The learning unit 76 uses backpropagation to learn the second model M2, using the time from the start of the vacuum drying process to the end of the vacuum drying process as the drying time. The drying time is the time (elapsed time) from the start of the vacuum drying process until the pressure fluctuation Δp converges to a constant value, and serves as the training data for learning. In step SP24, the learning unit 76 performs backpropagation based on the error between the drying time output by the second model M2 and the training data (actual time taken). Steps SP28 to SP30 and step SP31 may be executed in parallel. Then the process ends.

[0079] In this way, the first model M1 and the second model M2 are trained in accordance with the operation of the actual vacuum degreasing and cleaning apparatus 100. In particular, training data for the output of the first model M1 is generated based on the presence or absence of inflection points during the steam exhaust stage and is used for training. Also, training data for the second model M2 is generated based on the convergence of the pressure fluctuation Δp of the pressure p and is used for training.

[0080] <Effects> In this embodiment, the drying result of the workpiece 14 in the vacuum drying process performed after the vacuum cleaning process is estimated using the cleaning information. Therefore, for example, an operator can recognize the drying result in the vacuum drying process, making it possible to consider shortening the time of the vacuum drying process. This makes it possible to optimize the vacuum drying process. In other words, the time of the vacuum drying process can be optimized, and energy consumption can be reduced. Furthermore, in the overall cycle including the vacuum cleaning process and the vacuum drying process, the vacuum drying process may take, for example, about 40 percent of the cycle time, but by optimizing the vacuum drying process, it is possible to effectively reduce the energy consumption of the entire cycle.

[0081] Furthermore, by estimating the probability of successful drying and the time required for drying, it is possible to appropriately set parameters such as time in the vacuum drying process. This optimizes energy consumption.

[0082] Furthermore, by estimating the drying results using the temperature change ΔT and immersion time t, the drying results in the vacuum drying process can be accurately estimated from the information obtained before the vacuum drying process.

[0083] Furthermore, by estimating the drying result using workpiece information such as the mass W of the workpiece 14, the volume V of the workpiece 14, the surface area A of the workpiece 14, and the number N of workpieces 14, the drying result can be estimated with greater accuracy by taking into account the state of the workpiece 14 being dried.

[0084] Furthermore, by referring to the inflection point of the pressure p in the space where the workpiece 14 is placed (washing chamber 1), it is possible to efficiently determine whether the drying is good or bad in the vacuum drying process. Therefore, the first model M1 can be appropriately trained using the inflection point of pressure p.

[0085] Furthermore, if there is an inflection point during the steam exhaust stage, the probability of drying being completed is estimated, and if there is no inflection point during the steam exhaust stage, it is considered a drying failure, allowing the first model M1 to be properly trained. A drying failure occurs, for example, when there is residual liquid on the workpiece 14.

[0086] ≪Variations≫ It should be noted that the present invention is not limited to the embodiments described above. That is, any design modifications made to the above-described examples by those skilled in the art are also included within the scope of the present invention, as long as they retain the features of the present invention. Furthermore, the elements of the above embodiments and the following modifications can be combined to the extent that it is technically possible, and any combination thereof is also included within the scope of the present invention, as long as it retains the features of the present invention.

[0087] For example, in the above embodiment, the image captured by the camera 62 was used as one example of information regarding the volume V, surface area A, and number N of the workpiece 14, but it is not limited to the use of captured images. For example, the information may be input by an operator or the like, or the information may be obtained from the specifications of the workpiece 14. Furthermore, the information regarding the mass W of the workpiece 14 is not limited to the measurement result by the load cell 61, and the information may be input by an operator or the like, or the information may be obtained from the specifications of the workpiece 14.

[0088] Furthermore, in the above embodiment, the temperature change amount ΔT was given as an example of the temperature change of the cleaning agent 17 supplied to the workpiece 14 in the immersion process. However, the temperature change amount ΔT is not limited to the above, as long as it is a temperature change of the cleaning agent 17 supplied to the workpiece 14 (information indicating heat exchange between the workpiece 14 and the cleaning agent 17). For example, the temperature change amount ΔT may be the temperature change between the temperature of the cleaning agent 17 before it is sprayed onto the workpiece 14 and the temperature of the cleaning agent 17 after it is sprayed onto the workpiece 14 in the secondary shower process.

[0089] Furthermore, in the above embodiment, the estimation unit 73 estimated the drying result of the workpiece 14 based on workpiece information and cleaning information. However, if the mass W, volume V, surface area A, and number N of the target workpiece 14 are fixed, the drying result may be estimated based solely on the cleaning information.

[0090] In the above embodiment, one example was given where various information, such as the measurement result of the mass W of the workpiece 14, is acquired in conjunction with the operation of the vacuum degreasing and cleaning apparatus 100. However, learning may also be performed using information related to past operations already performed by the vacuum degreasing and cleaning apparatus 100. That is, the necessary data may be extracted from the log of past operations of the vacuum degreasing and cleaning apparatus 100, and the first model M1 and the second model M2 may be learned from this data. [Explanation of symbols]

[0091] 14: Work 17: Cleaning agent 66:Drying estimation device 71: Acquisition part 73:Estimation part 75: Calculation section 76: Learning Department 100: Vacuum degreasing and cleaning device A:Surface area M1: First Model (Model) N: Quantity V: Volume W: mass p: pressure t: Soaking time ΔT: Temperature change Δp: pressure fluctuation

Claims

1. A drying estimation device applied to a vacuum degreasing and cleaning apparatus that performs a vacuum cleaning process and a vacuum drying process on a workpiece, An acquisition unit that acquires cleaning information, which is information relating to the cleaning state of the workpiece in the vacuum cleaning process, An estimation unit that estimates the drying result of the workpiece in the vacuum drying process performed after the vacuum cleaning process based on the cleaning information, A drying estimation device characterized by comprising the following features.

2. The drying estimation apparatus according to claim 1, characterized in that the estimation unit estimates the probability that drying will be completed and the time required for drying as the drying result.

3. The drying estimation apparatus according to claim 1 or 2, characterized in that the cleaning information is information relating to at least one of the following indicators: the amount of temperature change of the cleaning agent supplied to the workpiece in the vacuum cleaning step, and the immersion time for which the workpiece is immersed in the cleaning agent in the vacuum cleaning step.

4. The acquisition unit acquires work information, which is information relating to at least one of the following indicators: the mass of the work, the volume of the work, the surface area of ​​the work, and the number of work pieces. The drying estimation apparatus according to claim 1 or 2, characterized in that the estimation unit estimates the drying result based on the work information and the cleaning information.

5. A learning unit that uses the inflection point of the pressure in the space where the workpiece is placed during the vacuum drying process to train a model of the drying result of the workpiece corresponding to the vacuum drying process, Furthermore, The drying estimation apparatus according to claim 1 or 2, characterized in that the estimation unit estimates the drying result of the workpiece using the learned model.

6. The drying estimation device according to claim 5, characterized in that the learning unit, when there is an inflection point during the steam exhaust stage in the vacuum drying process, uses the drying result as an estimated value of the probability of drying being completed obtained from the pressure fluctuation of the pressure, and when there is no inflection point during the steam exhaust stage, uses the drying result as a drying failure and trains the model accordingly.

Images