Lubricating oil state estimation device, lubricating oil state estimation system, and lubricating oil state estimation method

Abstract

This lubricating oil state estimation device estimates oil state information indicating a state of lubricating oil supplied to a sliding portion that has a fixed component and a movable component that slides relative to the fixed component. The lubricating oil state estimation device is provided with: a sound wave prediction unit that obtains predicted sound wave information indicating, by a probability distribution, a sound wave predicted to be generated in the sliding portion, on the basis of a prior distribution of the oil state information based on a state equation that indicates, by a probability distribution, a temporal change in the oil state information, and design information of the sliding portion; and an estimation unit that calculates an estimated value of the oil state information on the basis of an observation equation indicating a correlation between the predicted sound wave information as a state variable and an observation variable indicating, by a probability distribution including an error in a measured value of the sound wave, the measured value of the sound wave, and the prior distribution of the oil state information. Thus, the lubricating oil state estimation device can accurately estimate the state of the lubricating oil supplied to the sliding portion and contribute to maintenance of equipment that has the sliding portion.

Description

Lubricant state estimation device, lubricant state estimation system, and lubricant state estimation method

[0001] This disclosure relates to a lubricant state estimation device, a lubricant state estimation system, and a lubricant state estimation method for estimating the state of lubricant in a sliding part.

[0002] Conventionally, it is known that acoustic emission (AE) signals, which are one type of sound waves generated in sliding parts, are used as a method for detecting abnormalities in sliding parts. Sliding parts consist of, for example, a rotating shaft and a sliding bearing in a compressor. Patent Document 1 discloses a device for detecting plastic deformation of a bearing and the occurrence of microcracks inside a bearing from AE signals.

[0003] Japanese Unexamined Patent Publication No. 62-282258

[0004] Generally, deterioration of the lubricating oil supplied to the sliding parts increases the risk of abnormalities occurring in the sliding parts. The device described in Patent Document 1 detects events occurring in the structural parts of the sliding parts, but it was not possible to accurately grasp the condition of the lubricating oil.

[0005] This disclosure was made to solve the problems described above, and aims to provide a lubricant state estimation device, a lubricant state estimation system, and a lubricant state estimation method that can accurately estimate the state of lubricant supplied to a sliding part and contribute to the maintenance of equipment having a sliding part.

[0006] The lubricating oil state estimation device according to this disclosure is a lubricating oil state estimation device that estimates oil state information indicating the state of lubricating oil supplied to a sliding part having a fixed part and a movable part that slides with respect to the fixed part, and comprises: a sound wave prediction unit that obtains predicted sound wave information indicating the sound waves predicted to be generated in the sliding part by a probability distribution based on a prior distribution of oil state information based on a state equation that indicates the change in oil state information by a probability distribution, and design information of the sliding part; and an estimation unit that calculates an estimated value of oil state information based on a prior distribution of oil state information, an observation equation that shows the correlation between the predicted sound wave information as a state variable and an observation variable that indicates the measured value of the sound wave by a probability distribution including the error of the measured value of the sound wave, and an observation distribution of oil state information.

[0007] The lubrication oil state estimation system according to this disclosure comprises a lubrication oil state estimation device and a measuring device that measures sound waves generated in the sliding part and transmits them to the lubrication oil state estimation device.

[0008] The lubrication oil state estimation method according to this disclosure is a lubrication oil state estimation method for estimating oil state information indicating the state of lubrication oil supplied to a sliding part having a fixed part and a movable part that slides with respect to the fixed part, comprising: a sound wave prediction step of obtaining predicted sound wave information indicating the sound waves predicted to be generated in the sliding part by a probability distribution, based on a prior distribution of oil state information based on a state equation that indicates the change in oil state information by a probability distribution, and design information of the sliding part; and an estimation step of calculating an estimated value of oil state information based on a prior distribution of oil state information, based on a prediction sound wave information as a state variable, an observation equation that shows the correlation between the measured value of the sound wave and an observation variable that indicates the measured value of the sound wave by a probability distribution including the error of the measured value of the sound wave, and a prior distribution of oil state information.

[0009] The lubricating oil state estimation device, lubricating oil state estimation system, and lubricating oil state estimation method of this disclosure calculates an estimated value of oil state information based on a correlation between predicted sound wave information as a state variable and an observed variable that shows the correlation between the measured value of the sound wave and the measured value of the sound wave using a probability distribution including the error of the measured value, and a prior distribution of oil state information. Therefore, the lubricating oil state estimation device, lubricating oil state estimation system, and lubricating oil state estimation method of this disclosure can accurately estimate the state of the lubricating oil supplied to the sliding part and contribute to the maintenance of equipment having a sliding part.

[0010] This is a functional block diagram showing the lubricating oil state estimation device according to Embodiment 1. This is a hardware configuration diagram showing the lubricating oil state estimation device according to Embodiment 1. This is a hardware configuration diagram showing the lubricating oil state estimation device according to Embodiment 1. This is a schematic diagram of a compressor to which the lubricating oil state estimation device according to Embodiment 1 is applied. This is a schematic diagram of the sliding part of a compressor to which the lubricating oil state estimation device according to Embodiment 1 is applied. This is a flowchart showing the lubricating oil state estimation method according to Embodiment 1. This is a flowchart showing the processing of the specified value acquisition unit according to Embodiment 1. This is a flowchart showing the processing of the sound wave prediction unit according to Embodiment 1. This is a flowchart showing the processing of the estimation unit according to Embodiment 1. This is a flowchart showing the processing performed by the sound wave prediction unit according to a modified example of Embodiment 1. This is a functional block diagram showing the lubricating oil state estimation device according to Embodiment 2. This is a diagram for explaining the learning method according to Embodiment 2. This is a diagram for explaining the sound wave measurement method according to Embodiment 2. This is a diagram for explaining the estimation method according to Embodiment 2. This is a functional block diagram showing the lubricating oil state estimation device according to Embodiment 3. This is a schematic diagram of a compressor to which the lubricating oil state estimation device according to Embodiment 3 is applied. This is a flowchart showing the lubricating oil state estimation method according to Embodiment 3. This is a flowchart showing the processing of the sound wave prediction unit according to Embodiment 3. This is a flowchart showing the processing of the measurement value processing unit according to Embodiment 3. This is a flowchart showing the processing of the estimation unit according to Embodiment 3. This is a flowchart showing the processing of the sound wave prediction unit according to Embodiment 3. This is a schematic diagram showing the test apparatus according to Embodiment 4. This is a flowchart showing the method for determining the sound wave generation model according to Embodiment 4. This is a flowchart showing the method for determining the sound wave generation model according to Embodiment 5. This is a flowchart showing the processing of the sound wave prediction unit according to Embodiment 6. This is a flowchart showing the processing of the sound wave prediction unit according to Embodiment 6. This is a functional block diagram showing the lubricating oil state estimation system according to Embodiment 7.

[0011] Hereinafter, preferred embodiments of the lubricating oil state estimation device, lubricating oil state estimation system, and lubricating oil state estimation method for detecting abnormalities in sliding parts according to this disclosure will be described with reference to the drawings. In each drawing, the same or equivalent components are denoted by the same reference numerals, and their detailed descriptions are omitted. Note that the relative dimensions or shapes of the components in each drawing may differ from those of the actual components.

[0012] Embodiment 1. Figure 1 is a functional block diagram showing a lubricating oil state estimation device 1 according to Embodiment 1. As shown in Figure 1, the lubricating oil state estimation device 1 has a specified value acquisition unit 2, a measured value acquisition unit 3, a calculation unit 4, an output unit 5, and a storage unit 6. The lubricating oil state estimation device 1 is a device for estimating the state of lubricating oil supplied to the sliding part 50 of equipment having a sliding part 50, such as a compressor 100.

[0013] The specified value acquisition unit 2 acquires a specified value entered by the user via an input device 11 (e.g., a keyboard, touch panel, or microphone) connected to the lubricating oil state estimation device 1. The measured value acquisition unit 3 acquires the measured value of sound waves measured by a measuring device 12 (e.g., a sound wave sensor 20 and a sound wave processing device 30 that processes signals received from the sound wave sensor 20) connected to the lubricating oil estimation device. The measuring device 12 measures sound waves generated in the sliding part 50 of the equipment to which the lubricating oil state estimation device 1 is applied. The specified value and measured value will be explained later.

[0014] The calculation unit 4 estimates oil state information, which indicates the physical properties of the lubricating oil supplied to the sliding part 50 and the oil film thickness, based on the information acquired by the specified value acquisition unit 2 and the measured value acquisition unit 3. The calculation unit 4 includes a sound wave prediction unit 7, a measured value processing unit 8, and an estimation unit 9. The sound wave prediction unit 7 generates a prior distribution of oil state information and predicted sound wave information, which indicates the prior distribution of sound waves, based on the specified value acquired by the specified value acquisition unit 2 and a state equation that shows the time change of oil state information using a probability distribution. The measured value processing unit 8 generates a likelihood distribution of the measured values ​​of sound waves acquired by the measured value acquisition unit 3. The estimation unit 9 estimates the posterior distribution and estimated value of oil state information based on an observation equation that shows the correlation between the predicted sound wave information (state variable) and the likelihood distribution of the measured values ​​of sound waves (observed variable), and the prior distribution of oil state information.

[0015] The output unit 5 outputs the estimated value of the lubricating oil information estimated by the estimation unit 9 to the output device 13 (for example, a display or speaker) connected to the lubricating oil state estimation device 1.

[0016] The memory unit 6 stores various parameters and functions (models) necessary for the operation of the lubricating oil state estimation device 1. As will be described in detail later, the memory unit 6 stores, for example, a function representing the oil film model, a function representing the sound wave generation model, a state equation, and an observation equation.

[0017] Figures 2 and 3 are hardware configuration diagrams showing a lubricating oil state estimation device 1 according to Embodiment 1. The lubricating oil state estimation device 1 is, for example, a server device and, as shown in Figure 2, is composed of a processing circuit 41 such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array). Furthermore, if the functions of the lubricating oil state estimation device 1 are executed by program, the lubricating oil state estimation device 1 may be composed of a processor 42 such as a CPU and a memory 43, as shown in Figure 3. Figure 3 shows that the processor 42 and the memory 43 are connected to each other so as to be able to communicate with each other via a bus 44. The lubricating oil state estimation method performed by the lubricating oil state estimation device 1 is realized by the processor 42 reading and executing a program stored in the memory 43. The memory 43 may be a non-volatile or volatile semiconductor memory such as RAM, ROM, flash memory, EPROM, and EEPROM, or a removable recording medium such as a magnetic disk, flexible disk, optical disk, compact disk, minidisc, or DVD. The functions of each part of the lubricating oil state estimation device 1 may be partially implemented by dedicated hardware and partially implemented by a program.

[0018] Here, the compressor 100 to which the lubricating oil state estimation device 1 is applied, its sliding part 50, and the method for measuring the measured value will be explained using Figures 4 and 5. First, the configuration of the compressor 100 to which the lubricating oil state estimation device 1 is applied will be explained using Figure 4. Figure 4 is a schematic diagram of the compressor 100 to which the lubricating oil state estimation device 1 according to Embodiment 1 is applied. Figure 4 is a diagram showing a longitudinal cross-section of the compressor 100. In Figure 4, a sealed rotary refrigerant compressor is exemplified as the compressor 100 to which the lubricating oil state estimation device 1 is applied. The compressor 100 is installed in a refrigeration cycle device having a refrigerant circuit. The refrigeration cycle device is, for example, an air conditioner. As shown in Figure 4, the compressor 100 has a housing 110, an intake pipe 120, a discharge pipe 130, a motor 140, a rotating shaft 150, and a compression mechanism 160.

[0019] The housing 110 houses the motor 140, the rotating shaft 150, and the compression mechanism 160. In the case of a sealed refrigerant compressor, the housing 110 is, for example, a pressure vessel formed in a cylindrical shape. The inside of the housing 110 is filled with compressed high-temperature, high-pressure refrigerant gas. Lubricating oil for lubricating the sliding parts 50 is stored at the bottom of the housing 110.

[0020] The suction pipe 120 is a pipe through which refrigerant flows into the compressor 100 from other equipment in the refrigerant circuit where the compressor 100 is located. The suction pipe 120 is located on the side of the housing 110 and is connected to the cylinder chamber 161a of the cylinder 161A, which will be described later. The inlet of the suction pipe 120 is covered by an intake muffler 120a that separates the liquid refrigerant from the refrigerant gas. The discharge pipe 130 is a pipe through which refrigerant flows out of the compressor 100 to other equipment in the refrigerant circuit where the compressor 100 is located. The discharge pipe 130 is located on the top of the housing 110.

[0021] The motor 140 is positioned above the compression mechanism 160 within the housing 110. The motor 140's rotor rotates using power supplied from a commercial power source or the like, driving the compression mechanism 160.

[0022] The rotating shaft 150 transmits the rotational force of the motor 140 to the compression mechanism 160. The rotating shaft 150 has a main shaft 150a portion 21a, a crank 150b, and a sub-shaft 150c portion 21c. The main shaft 150a portion 21a, the crank 150b, and the sub-shaft 150c portion 21c are arranged in this order from top to bottom. The rotor of the motor 140 is fixed to the main shaft 150a portion of the rotating shaft 150 by shrink fitting or press fitting. The crank 150b is eccentric with respect to the main shaft 150a portion and the sub-shaft 150c portion of the rotating shaft 150. An oil pump (not shown) is also provided at the bottom of the rotating shaft 150. The oil pump pumps up lubricating oil from the bottom of the housing 110 as the rotating shaft 150 rotates and supplies it to the compression mechanism 160. This ensures the mechanical lubrication of the compression mechanism 160.

[0023] The compression mechanism 160 is located in the lower part of the housing 110. The compression mechanism 160 is connected to the motor 140 by a rotating shaft 150, and compresses the refrigerant gas by the rotational force transmitted from the motor 140 and discharges it into the housing 110. The compression mechanism 160 has a cylinder 161A, a rolling piston 162, an upper bearing 163, a lower bearing 164, and vanes (not shown). The cylinder 161A is formed in a hollow cylindrical shape. The cylinder 161A is fixed to the inner circumferential surface of the housing 110. Inside the cylinder 161A, a cylindrical space, i.e., a cylinder chamber 161a, is formed with both ends open in the vertical direction. The cylinder 161A is also provided with an intake port (not shown) for drawing refrigerant gas into the cylinder chamber 161a from the intake pipe 120.

[0024] The cylinder chamber 161a houses the crank 150b of the rotating shaft 150, the rolling piston 162, and the vanes. The rolling piston 162 is slidably fitted to the outer circumference of the crank 150b of the rotating shaft 150. As the rotating shaft 150 rotates, the crank 150b of the rotating shaft 150 performs eccentric rotational motion within the cylinder chamber 161a, causing the volume of the working chamber, which is partitioned by the inner surface of the cylinder 161A, the outer surface of the rolling piston 162, and the vanes 26, to increase or decrease. This compresses the refrigerant drawn in from the intake port.

[0025] The upper bearing 163 is slidably fitted onto the main shaft 150a portion 21a of the rotating shaft 150. The lower bearing 164 is slidably fitted onto the sub-shaft 150c portion 21c of the rotating shaft 150. As a result, the rotating shaft 150 is rotatably supported by the upper bearing 163 and the lower bearing 164. The upper bearing 163 and the lower bearing 164 are sliding bearings that support the rotating shaft 150.

[0026] The upper bearing 163 also serves as an end plate that closes the upper opening of the cylinder chamber 161a. The lower bearing 164 also serves as an end plate that closes the lower opening of the cylinder chamber 161a. The upward-facing end face of the cylinder 161A is in contact with the upper bearing 163. The downward-facing end face of the cylinder 161A is in contact with the lower bearing 164.

[0027] The upper bearing 163 is provided with a discharge port (not shown) for discharging compressed refrigerant gas to the outside of the cylinder chamber 161a. In addition, to reduce noise such as pulsation noise from the refrigerant gas intermittently discharged from the discharge port, a discharge muffler 160a is attached to the upper part of the upper bearing 163 so as to cover the upper bearing 163. The discharge muffler 160a is provided with a discharge hole (not shown). The discharge hole connects the space formed by the discharge muffler 160a and the upper bearing 163 to the space inside the housing 110. The refrigerant gas compressed in the cylinder chamber 161a passes through the discharge port and is discharged into the space formed by the discharge muffler 160a and the upper bearing 163, and then discharged into the housing 110 from the discharge hole. The high-temperature, high-pressure refrigerant gas discharged from the cylinder chamber 161a into the housing 110 via the discharge muffler 160a rises inside the housing 110 and is discharged to the outside of the housing 110 via the discharge pipe 130. The refrigerant discharged from the compressor 100 circulates through the refrigerant circuit and returns to the intake muffler 120a.

[0028] In the compressor 100 described above, there are multiple sliding parts 50 inside the housing 110. Figure 5 is a schematic diagram of the sliding parts 50 of the compressor 100 to which the lubrication oil state estimation device 1 according to Embodiment 1 is applied. As shown in Figure 5, the sliding parts 50 are composed of fixed parts 51 and movable parts 52, and are a general term for structures in which the movable parts 52 slide against the fixed parts 51. The fixed parts 51 are parts whose position is fixed with respect to the housing 110. The movable parts 52 are parts that can slide against the fixed parts 51. The lubrication oil O is present between the fixed parts 51 and the movable parts 52, and the thickness δ of the lubrication oil between the fixed parts 51 and the movable parts 52 will be referred to as the oil film thickness δ in the following explanation. If abnormalities such as deformation of the members, wear of the members, or contact between members occur in these sliding parts 50, the rotational movement of the rotating shaft 150 or the compression movement of the compression mechanism 160 will be hindered, leading to a decrease in the performance and failure of the compressor 100.

[0029] Specifically, in the compressor 100 shown in Figure 4, the outer circumferential surface of the main shaft 150a portion 21a of the rotating shaft 150 and the inner circumferential surface of the upper bearing 163 correspond to the sliding portion 50. When the outer circumferential surface of the main shaft 150a portion 21a of the rotating shaft 150 and the inner circumferential surface of the upper bearing 163 constitute the sliding portion 50, the upper bearing 163 is a fixed component 51, and the main shaft 150a portion 21a of the rotating shaft 150 is a movable component 52. Also, the outer circumferential surface of the sub-shaft 150c portion 21c of the rotating shaft 150 and the inner circumferential surface of the lower bearing 164 correspond to the sliding portion 50. When the outer circumferential surface of the sub-shaft 150c portion 21c of the rotating shaft 150 and the inner circumferential surface of the lower bearing 164 constitute the sliding portion 50, the lower bearing 164 is a fixed component 51, and the sub-shaft 150c portion 21c of the rotating shaft 150 is a movable component 52. Furthermore, the outer circumferential surface of the rolling piston 162 and the inner circumferential surface of the cylinder 161A correspond to the sliding portion 50. In the case of the sliding portion 50 between the outer circumferential surface of the rolling piston 162 and the inner circumferential surface of the cylinder 161A, the cylinder 161A is the fixed part 51 and the rolling piston 162 is the movable part 52.

[0030] Next, we will return to Figure 4 and explain the method for measuring the measured values. An acoustic wave sensor 20 is provided on the outside of the housing 110. The acoustic wave sensor 20 is a sensor that detects sound waves that change based on the state of the sliding part 50. The acoustic wave sensor 20 can be selected based on the sound waves generated in the sliding part 50 of the equipment to which the lubricating oil state estimation device 1 is applied, but for example, it is an AE sensor that detects AE waves. The AE sensor detects stress vibration waves (AE waves) generated in the sliding part 50 and propagating through the cylinder 161A inside the housing 110 and the housing 110 as an AE signal. The acoustic wave sensor 20 may be detachably attached to the housing 110. An acoustic wave processing device 30 is connected to the acoustic wave sensor 20. The acoustic wave processing device 30 has, for example, an amplifier and a data logger, and measures sound waves (AE waves) by amplifying the sound wave signal (AE signal) transmitted from the acoustic wave sensor 20 by passing it through a predetermined filter, and records the measured sound waves (AE waves). The sound wave processing device 30, for example, measures and records sound waves at a predetermined measurement cycle, and outputs the measured value at that time to the lubricating oil state estimation device 1 each time the measurement cycle has elapsed.

[0031] Next, the method for measuring the lubricating oil state performed by the lubricating oil state estimation device 1 will be explained, using the case where the lubricating oil state estimation device 1 is applied to the compressor 100 shown in Figure 4 as an example. Figure 6 is a flowchart of the lubricating oil state estimation method according to Embodiment 1. First, the specified value acquisition unit 2 acquires a specified value input by the user (step S100).

[0032] Here, the process of step S100 will be specifically explained using Figure 7. Figure 7 is a flowchart of the process of the specified value acquisition unit 2. As shown in Figure 7, the specified value acquisition unit 2 acquires the initial values ​​and variations of the oil state information entered by the user as specified values ​​(step S110). Specifically, the specified value acquisition unit 2 acquires the initial values ​​of the oil's physical properties α and β, and the initial value of the oil film thickness δ. The specified value acquisition unit 2 also acquires the variations of the oil's physical properties α and β, and the variations of the initial value of the oil film thickness δ. The oil's physical properties α and β both indicate the degree to which the oil's physical properties change. As an example, physical property α represents the variation in oil film thickness δ with respect to the load corresponding to the rotational speed of the compressor 100. Physical property β represents the variation in oil film thickness δ with respect to the variation in load W. Furthermore, in parallel with step S110, the specified value acquisition unit 2 acquires the design information entered by the user as specified values ​​(step S120). Specifically, the specified value acquisition unit 2 acquires the design value of the bearing load W(t) and the shape S of the sliding part 50. Here, when the lubrication oil state estimation device 1 is applied to the compressor 100, the bearing load changes over time because the volume of the working chamber changes when the refrigerant is compressed. For this reason, the design value of the bearing load is expressed as a function of time t. The shape S of the sliding part 50 corresponds to, for example, one or more combinations of the inner diameter, outer diameter, and vertical length of the upper bearing 163 and lower bearing 164, and the diameter of the rotating shaft 150. Step S120 may be performed in parallel with step S110 or after step S110.

[0033] Returning to Figure 6, following step S100, the sound wave prediction unit 7 generates a prior distribution of oil state information (step S200). Specifically, the prior distribution x ｔ As shown in equation (1) below, it is generated by the equation of state, and the physical property value α of the oil at time tｔ and β ｔ , and the initial oil film thickness δ ０ｔ as components. The equation of state defines the prior distribution x ｔ by adding the variation v of oil state information, which is represented by a normal distribution with a mean of 0 and a variance v ｔ－１ to the prior distribution x at time t-1, which is one time point earlier ｔ of the oil state information represented by the normal distribution ｔ . The variation v ｔ is represented by a vector having, as components, the variation v in the physical property values of the oil at time t α，ｔ and v β，ｔ , and the variation in the initial value of the oil film thickness v δ０，ｔ as components. For the initial prior distribution x ｔ－１ , the initial values of the physical property values α and β of the oil acquired by the specified value acquiring unit 2, and the initial value of the oil film thickness δ are set. Further, for the variation v ｔ , the variations in the physical property values α and β of the oil acquired by the specified value acquiring unit 2, and the variation in the initial value of the oil film thickness δ are set. Note that step S200 only needs to be performed at least after step S110 is completed, and may be performed before step S120. Further, the variation may be represented by a different probability distribution such as Weibull distribution or lognormal distribution, instead of the normal distribution.

[0034]

[0035] Subsequent to step S200, the sound wave prediction unit 7 calculates predicted sound wave information that indicates, by a probability distribution, a sound wave predicted to be generated at the sliding portion 50 based on the prior distribution and design information (step S300).

[0036] Here, the processing of step S300 will be specifically described with reference to Fig. 8. Fig. 8 is a flowchart showing the processing of the sound wave prediction unit 7 according to the first embodiment. First, as shown in the following formula (2), the oil film thickness can be obtained from the oil state information and the design information using a function f() representing a physical model (referred to as an oil film model) representing the relationship between the oil state information, the design information, and the oil film thickness. Specifically, the function f() representing the oil film model can be expressed as the following formula (3).

[0037]

[0038]

[0039] The oil film thickness δ(t) at time t is given by equation (4) below, obtained by rearranging equation (2). The sound wave prediction unit 7 calculates the oil film thickness δ(t) at time t by inputting the prior distribution of oil state information and design information into the function f() that represents the oil film model, as shown in equation (4) (step S310). For example, the oil film thickness δ can be calculated based on the change in load W and fluid lubrication theory.

[0040]

[0041] Next, the sound wave prediction unit 7 uses the oil film thickness δ(t) obtained in step S310 as shown in equation (5) below, and a function g that represents a physical model (referred to as the sound wave generation model) that shows the relationship between the oil film thickness and the sound waves generated at the sliding part 50. ｉ By entering the information in ( ), predictive sound wave information A ｐｒｅｄｉｃ The following is calculated (step S320). Specifically, as an example, there is a method that models the change in oil film stiffness that occurs with changes in oil film thickness and the sound waves that are generated. The change in oil film stiffness that occurs with changes in oil film thickness can be calculated based on fluid lubrication theory. In other words, the sound wave generation model models the relationship between the oil film thickness, which changes due to the load applied from the movable part to the fixed part in the sliding part 50, and the sound waves that are generated. Predicted sound wave information A ｐｒｅｄｉｃ This is the prior distribution of sound waves predicted by inputting the prior distribution of oil state information into a physical model.

[0042]

[0043] Returning to Figure 6, the measurement value processing unit 8 acquires the measured values ​​of the sound waves measured by the measuring device (step S400). When the oil state estimation device is applied to the compressor 100 in Figure 3, the sound waves are, for example, AE waves. Subsequently, the measurement value processing unit 8 calculates the likelihood distribution A of the measured values ​​of the sound waves as shown in equation (6) below. ｍ (t) is generated (step S500). Equation (6) shows the likelihood distribution A of the sound wave. ｍ (t) with mean 0 and variance ε ｔ Measurement error ε shown in the normal distribution ｔIt is represented by a probability distribution that includes A. ｇ (t) is the measurement error ε from the measured sound wave. ｔ This is the value after subtracting [a certain factor].

[0044]

[0045] Then, the estimation unit 9 generates predicted sound wave information A ｐｒｅｄｉｃ and the likelihood distribution A of sound waves ｍ Based on (t), an estimated value of the oil condition information is calculated (step S600).

[0046] Here, the process of step S600 will be specifically explained using Figure 9. Figure 9 is a flowchart showing the process of the estimation unit 9 according to Embodiment 1. First, the estimation unit 9 processes the predicted sound wave information A ｐｒｅｄｉｃ Obtain the likelihood distribution A of the sound wave (step S610), ｍ (t) is obtained (step S620). Next, the estimation unit 9 adds the predicted sound wave information A to the sound wave observation equation shown by the following equation (7). ｐｒｅｄｉｃ and the likelihood distribution A of sound waves ｍ Enter (t) (step S630).

[0047]

[0048] Next, the estimation unit 9 obtains the posterior distribution of oil state information from the observation equation and the prior distribution of oil state information (step S640). As a method for obtaining the posterior distribution of oil state information, any of the following can be used: a particle filter, an ensemble Kalman filter, an extended Kalman filter, and an unsciented Kalman filter. Then, the estimation unit 9 determines the value of oil state information that has the highest probability in the posterior distribution of oil state information, and sets this value to the estimated value of oil state information x. ｔ＿ｐｒｅ The output is as follows (step S650). For example, when a particle filter is used, multiple candidate particles for the predicted state are created based on the prior distribution, and among these particles, it is decided based on the likelihood distribution which ones to keep, which ones to increase in number, and which ones to remove. Then, the distribution of the remaining particles is taken as the posterior distribution, and an estimated value x is calculated based on the posterior distribution. ｔ＿ｐｒｅ This determines the accuracy of the estimation.

[0049] Returning to Figure 6, the estimation unit 9 determines whether or not the oil state information has been estimated over a predetermined measurement time (step S700). If the time from the start of the oil state information estimation to the present has reached the measurement time (step S700: YES), the output unit 5 outputs the estimated value of the oil state information (step S800). If the time from the start of the oil state information estimation to the present has not reached the measurement time (step S700: NO), the sound wave prediction unit 7 updates the prior distribution with the estimated value of the oil state information (step S900). Then, the processes in steps S300, S400, S500, and S600 are repeatedly executed until the condition in step S700 is met. In other words, the updating of the predicted sound wave information, acquisition of measured values ​​of sound waves and generation of the likelihood distribution, and calculation of the estimated value of the oil state information are repeatedly performed.

[0050] Here, the process of step S900 will be specifically explained using Figure 10. Figure 10 is a flowchart showing the process of the sound wave prediction unit 7 according to Embodiment 1. As shown in Figure 10, first the sound wave prediction unit 7 calculates the estimated value x ｔ＿ｐｒｅ The value of x in equation (1), which is the state equation. ｔ－１ Substitute the values ​​into the equation (step S910). Next, the sound wave prediction unit 7 updates the prior distribution by advancing t by one time point, using the state equation of equation (1) above (step S920).

[0051] As described above, the lubricating oil state estimation device 1 and lubricating oil state estimation method of Embodiment 1 calculate estimated values ​​of oil state information based on predicted sound wave information as a state variable, an observation equation showing the correlation between the measured values ​​of sound waves and an observed variable which is represented by a probability distribution including the error of the measured values, and a prior distribution of oil state information. In this way, the lubricating oil state estimation device 1 and lubricating oil state estimation method of Embodiment 1 calculate estimated values ​​of oil state information based on measured values ​​of sound waves, so it is possible to estimate the state of the lubricating oil supplied to the sliding parts with greater accuracy than simply estimating the state of the lubricating oil from the usage period of the lubricating oil, and contribute to the maintenance of equipment having sliding parts.

[0052] Furthermore, users can determine whether the lubricant has deteriorated by comparing the estimated state of the lubricant with the state of a normal lubricant. In addition, by modeling the changes in the physical properties of the lubricant over time, it is possible to predict the progression of the lubricant's deterioration based on the estimated state and the model, and to determine the limit of the lubricant's use and the timing for replacement.

[0053] Furthermore, according to Embodiment 1, the sound wave prediction unit obtains predicted sound wave information by inputting oil state information and design information into a sound wave generation model, which is a physical model that models the generation of sound waves in the sliding part. Therefore, it does not require data that shows sound waves during abnormal conditions.

[0054] Furthermore, according to Embodiment 1, an AE sensor is used as the measuring device. By using a high-precision acoustic wave sensor, the accuracy of estimating the state of the lubricating oil can be improved.

[0055] Furthermore, according to Embodiment 1, the estimation unit calculates an estimated value of oil state information using a particle filter. Since the particle filter is a filter that can handle nonlinear behavior and non-Gaussian distribution variability, using a particle filter makes it possible to create a general-purpose lubricating oil state estimation device 1 that does not limit the probability distribution showing the variability. Specifically, even with filters other than particle filters, such as an extended Kalman filter, if the variability shows a normal distribution as described above, the estimated value of oil state information can be mathematically predicted based on the normal distribution. In this respect, since the particle filter does not formulate the probability distribution mathematically when estimating, it can estimate the estimated value of oil state information even when assuming variability of any probability distribution other than a normal distribution.

[0056] Furthermore, according to Embodiment 1, the sliding part is the bearing of the compressor 100, and the measuring device measures sound waves on the surface of the compressor housing. In this way, by applying the lubricating oil state estimation device 1 of Embodiment 1 to the compressor, it is possible to determine the usage limit of the lubricating oil supplied to the compressor and use this to determine the timing of repairs and replacements of the sliding part 50 of the compressor 100.

[0057] (Modified Version of Embodiment 1) When applying the lubricating oil state estimation device 1 to the compressor 100 shown in Figure 4, there is a distance between the sound wave sensor 20 that measures sound wave signals and the sliding part 50 which is the source of sound waves, and the sound wave sensor 20 may measure sound waves transmitted from the source of sound waves. When generating predicted sound wave information, the sound wave prediction unit 7 may take into account the influence of sound wave transmission and perform the processing of step S300 described in Embodiment 1 as follows. Figure 11 is a flowchart showing the processing performed by the sound wave prediction unit 7 according to the modified version of Embodiment 1. In the lubricating oil state estimation method of the modified version of Embodiment 1, as shown in Figure 11, when step S310 is performed and the oil film thickness δ(t) is calculated, the sound wave prediction unit 7 uses the oil film thickness δ(t) obtained in step S310 as shown in the following equation (8) and a function g that represents a sound wave generation model that shows the relationship between the oil film thickness and the sound waves generated at the sliding part 50. ｉ By entering the information in ( ), predictive sound wave information A ｊ＿ｐｒｅｄｉｃ The result is calculated (step S320). In order to distinguish it from the output by equation (5) described in Embodiment 1, the output by equation (8) is called the predicted sound wave information A. ｊ＿ｐｒｅｄｉｃ However, Predicted sound wave information A ｊ＿ｐｒｅｄｉｃ This is the predicted sound wave information A in equation (5). ｐｒｅｄｉｃ It is identical to that.

[0058]

[0059] Next, the sound wave prediction unit 7 calculates the predicted sound wave information A obtained in step S320 as shown in equation (9) below. ｊ＿ｐｒｅｄｉｃ This is a physical model (referred to as a sound wave transmission model) that represents the relationship between sound waves generated at the sliding part 50 and sound waves transmitted to the sound wave sensor 20, and the function p １，１ By entering the information in ( ), predictive sound wave information A ｐｒｅｄｉｃ The function p is calculated (step S330). The sound wave transmission model models the transmission of sound waves from the sliding part to the sound wave measurement position. １，１ ( ) is stored in memory unit 6.

[0060]

[0061] Furthermore, in Embodiment 1, the sound wave observation equation shown in equation (7) is also modified as shown in equation (10) below.

[0062]

[0063] According to a modified embodiment of the first embodiment, a sound wave generation model that models the generation of sound waves at the sliding part and a sound wave transmission model that models the transmission of sound waves from the sliding part to the sound wave measurement position are combined to obtain predicted sound wave information. By dividing the model for obtaining predicted sound wave information into a sound wave generation model and a sound wave transmission model, the accuracy of each model can be verified individually, thereby improving the accuracy of the model.

[0064] Embodiment 2. Figure 12 is a functional block diagram showing the lubricating oil state estimation device 1A according to Embodiment 2. The lubricating oil state estimation device 1A of Embodiment 2 shown in Figure 12 differs from the lubricating oil state estimation device 1 of Embodiment 1 in that a trained model LM for estimating the sound wave transmission model is stored in the storage unit 6. The differences from Embodiment 1 will be explained below in detail.

[0065] The trained model LM is a model that has learned the effects of sound wave transmission, and based on the representative parameters described later, it is similar to the sound wave transmission model p explained in the modified example of Embodiment 1. １，１() is output. The learning method of the trained model LM will be explained using Figure 13. Figure 13 is a diagram for explaining the learning method according to Embodiment 2. As shown in Figure 13, first, in the device to which the lubricating oil state estimation device 1A is applied, a plurality of sound wave sensors 20 are attached, and sound waves are measured for each of the plurality of transmission paths from the sliding part 50 where sound waves are generated to the plurality of sound wave sensors 20 (step S1000). Figure 14 is a diagram for explaining the sound wave measurement method according to Embodiment 2. As shown in Figure 14, when the device to which the lubricating oil state estimation device 1A is applied is a compressor 100, n sound wave sensors 20_1, 20_2, ... 20_n are installed at multiple locations on the side of the housing 110 of the compressor 100 where the vertical heights are different. At this time, the transmission paths of each sound wave from the sliding part 50 to each sound wave sensor 20_1, 20_2, ... 20_n are defined as transmission paths Pt1 to Ptn, with subscripts 1 to n. The sound wave processing device 30 independently measures the sound wave signals detected by the n sound wave sensors 20_1, 20_2, ... 20_n. The sound waves measured from the signals detected by each sound wave sensor 20_1, 20_2, ... 20_n are assigned subscripts 1 to n corresponding to the transmission path, and A １，ｍ ~A ｎ，ｍ This will be assumed. Furthermore, sound wave measurements will be performed when the lubricating oil has not deteriorated and is in a normal state of oil properties.

[0066] Next, representative parameters are determined (step S1001). Here, for example, the length of the transmission path from the sliding part 50 where sound waves are generated to the multiple sound wave sensors 20 is used as a representative parameter. The length of the transmission path to the multiple sound wave sensors 20_1, 20_2, ... 20_n is given by adding subscripts 1 to n corresponding to the transmission path, L Ｐ，１ ~L Ｐ，ｎ Let's assume that.

[0067] Then, the measured sound wave A １，ｍ ~A ｎ，ｍ and the corresponding representative parameter L Ｐ，１ ~L Ｐ，ＮThe data is input to the learning device as learning data. The learning device then learns the relationship between the measured sound waves and the representative parameters based on the learning data (step S1002). When the representative parameters are given as input as a learning result, the learning device creates a sound wave transmission model p from the sliding part 50 to the measurement position. １，ｉ The trained model LM outputs ( ). The training method of the learning device is not particularly limited. The learning device is a computer having a processor such as a CPU or GPU and memory, and is configured to perform so-called machine learning.

[0068] Next, using Figure 15, we will examine the sound wave transmission model p using the trained model LM. １，ｉ The estimation method for ( ) will be explained. Figure 15 is a diagram illustrating the estimation method according to Embodiment 2. First, when the lubricating oil state estimation method described in the modified example of Embodiment 1 is performed, representative parameters of the sound wave sensor 20 that is actually attached to the side of the housing 110 of the compressor 100 are obtained (step S1101). Next, the representative parameters are input to the trained model (S1102). As a result, the trained model LM learns the sound wave transmission model p corresponding to the representative parameters １，ｉ Output ( ).

[0069] According to Embodiment 2, the sound wave transmission model p １，ｉ The ( ) is output by inputting representative parameters into a trained model LM that outputs a sound wave transmission model when different representative parameters are given as input for each of the multiple measurement positions. The trained model LM is constructed by learning the relationship between sound waves measured at multiple measurement positions and the different representative parameters for each of the multiple measurement positions. The sound wave transmission model, which is affected by the transmission path, can be optimized to match the actual location where the sound wave sensor 20 is installed, and predicted sound wave information can be obtained with high accuracy.

[0070] Embodiment 3. Figure 16 is a functional block diagram showing the lubricating oil state estimation device 1B according to Embodiment 3. The lubricating oil state estimation device 1B of Embodiment 3 shown in Figure 16 differs from the lubricating oil state estimation device 1 of Embodiment 1 in that the measurement value acquisition unit 3B acquires the torque of the motor 140 as a measurement value, and the calculation unit 4B utilizes the torque of the motor 140 to estimate the state of the lubricating oil. The following will mainly explain the differences from Embodiment 1. As a method for utilizing the torque value of the motor 140 to estimate the state of the lubricating oil, the following will describe a method in which the torque value of the motor 140 is added to the likelihood distribution, the bearing load is added to the prior distribution, and the bearing load of the prior distribution is updated with the torque value of the motor 140. This method can improve the accuracy of estimating the state of the lubricating oil.

[0071] Figure 17 is a schematic diagram of a compressor 100 to which the lubricating oil state estimation device 1 according to Embodiment 3 is applied. As shown in Figure 17, in the compressor 100 to which the lubricating oil state estimation device 1 according to Embodiment 3 is applied, a current-voltage sensor 62 is installed between the motor 140 and the power supply 61 that supplies power to the motor 140. The current-voltage sensor 62 transmits the measured current and voltage to the measurement value acquisition unit 3B. The measurement value acquisition unit 3B acquires the current and voltage transmitted from the current-voltage sensor 62 and converts them into the torque value of the motor 140 by a well-known method.

[0072] The following describes the lubrication oil state estimation method according to Embodiment 3 using Figures 18 to 22. The lubrication oil state estimation method according to Embodiment 3 shares the same overall flow as the lubrication oil state estimation method of Embodiment 1. Therefore, in Figures 18 to 22, common steps are given to processes that are common to the lubrication oil state estimation method of Embodiment 1, and their explanations are omitted, while the differing processes are explained in detail. Figure 18 is a flowchart of the lubrication oil state estimation method according to Embodiment 3. As shown in Figure 18, the specified value acquisition unit 2, instead of step S100 described in Embodiment 1, acquires the variation in the design value of the bearing load as one of the design information, in addition to the design information described in Embodiment 1, as a specified value entered by the user (step S101).

[0073] Following step S101, the sound wave prediction unit 7B generates prior distributions of oil state information and bearing loads, instead of step S200 as described in Embodiment 1 (step S201). Specifically, the prior distribution x' ｔ As shown in equation (11) below, it is generated by the equation of state, and the physical property value α of the oil at time t ｔ and β ｔ Initial value of oil film thickness δ ０ｔ It is represented by a vector whose component is the bearing load W(t). The equation of state is given by the prior distribution x' ｔ The prior distribution x' of time t-1, one time point prior. ｔ－１ The mean is 0, and the variance is v'. ｔ The variation of oil state information shown by a normal distribution v' ｔ This is expressed by adding the variation v'. ｔ v is the variation in the physical properties of the oil at time t. α，ｔ and v β，ｔ , variation in the initial value of oil film thickness v δ０，ｔ , and variation in bearing load v ｗ，ｔ It is represented by a vector with components x'. The initial prior distribution x' ｔ－１ The initial values ​​of the oil's physical properties α and β, the initial value of the oil film thickness δ, and the design value of the bearing load W(t) are set in the specified value acquisition unit 2. ｔ This setting includes the variation in the physical properties α and β of the oil acquired by the specified value acquisition unit 2, the variation in the initial value of the oil film thickness δ, and the variation in the design value of the bearing load W(t).

[0074]

[0075] Following step S201, the sound wave prediction unit 7B calculates predicted sound wave information that shows the sound waves predicted to be generated in the sliding unit 50 as a probability distribution, based on the prior distribution and design information (step S301).

[0076] Here, the process of step S301 will be specifically explained using Figure 19. Figure 19 is a flowchart showing the process of the sound wave prediction unit 7B according to Embodiment 3. First, as shown in equation (12) below, the oil film thickness δ' can be determined from the oil state information and design information using a function f() that represents an oil film model representing the relationship between the oil state information and design information and the oil film thickness δ'. Here, in equation (12), the bearing load W is treated as the parameter to be estimated.

[0077]

[0078] The oil film thickness δ'(t) at time t is given by equation (13) below, obtained by rearranging equation (12). The sound wave prediction unit 7B calculates the oil film thickness δ'(t) at time t by inputting the prior distribution of oil state information and design information into the function f() that represents the oil film model, as shown in equation (13) below (step S311).

[0079]

[0080] Next, the sound wave prediction unit 7B uses the oil film thickness δ'(t) obtained in step S311 as shown in equation (14) below, and a function g that represents the sound wave generation model showing the relationship between the oil film thickness and the sound waves generated at the sliding part 50. ｉ By inputting into (), predictive sound wave information A' ｐｒｅｄｉｃ Calculates the predicted sound wave information A'. ｐｒｅｄｉｃ This is the prior distribution of sound waves predicted by inputting the prior distribution of oil state information into a physical model.

[0081]

[0082] Returning to Figure 18, the measurement value processing unit 8B acquires the measured values ​​of sound waves and torque, instead of step S400 described in Embodiment 1 (step S401). Subsequently, the measurement value processing unit 8B generates likelihood distributions for both measured values, instead of step S500 described in Embodiment 1 (step S501).

[0083] Here, the process of step S501 will be specifically explained using Figure 20. Figure 20 is a flowchart showing the process of the measurement value processing unit 8B according to Embodiment 3. As shown in Figure 20, first, the measurement value processing unit 8B generates the likelihood distributions of sound waves and torque respectively, as shown in the following equations (15) and (16) (step S510). In equation (15), the likelihood distribution A of sound waves ｍ (t) with mean 0 and variance ε ｔ Measurement error ε shown in the normal distribution ｔ It is represented by a probability distribution that includes A. ｇ (t) is the measurement error ε from the measured sound wave. ｔ This is the value obtained by subtracting . In equation (16), the likelihood distribution Tm(t) of torque is given by a mean of 0 and a variance of ε. Ｔｒ，ｔ Measurement error ε shown in the normal distribution Ｔｒ，ｔ It is represented by a probability distribution that includes T. ｇ (t) is the measurement error ε from the measured sound wave. Ｔｒ，ｔ This is the value after subtracting [a certain factor].

[0084]

[0085]

[0086] Next, the measurement value processing unit 8B calculates the likelihood distribution A of the measurement values, which includes both the sound wave and torque measurements, as shown in equation (17) below. Ｔｏｔａｌ (t) is generated (step S520).

[0087]

[0088] Returning to Figure 18, the estimation unit 9B, instead of step S600 in Embodiment 1, predicts the sound wave information A'. ｐｒｅｄｉｃ and the likelihood distribution A of the measured values Ｔｏｔａｌ Based on (t), the oil condition information and estimated bearing load values ​​are calculated (step S601).

[0089] Here, the process of step S601 will be specifically explained using Figure 21. Figure 21 is a flowchart showing the process of the estimation unit 9B according to Embodiment 3. First, the estimation unit 9B processes the predicted sound wave information A'. ｐｒｅｄｉｃ Obtain the likelihood distribution A of the measured value (step S611), Ｔｏｔａｌ(t) is acquired (step S621). Next, the estimation unit 9B substitutes the predicted sound wave information A' as a state variable into the observation equation for sound waves represented by the following formula (18) ｐｒｅｄｉｃ and inputs the likelihood distribution A of sound waves as an observation variable Ｔｏｔａｌ (t) (step S631).

[0090]

[0091] Subsequently, the estimation unit 9B obtains the posterior distribution of the oil state information from the observation equation and the prior distribution of the oil state information (step S641). As a method for obtaining the posterior distribution of the oil state information, any one of a particle filter, an ensemble Kalman filter, an extended Kalman filter, and an Unscented Kalman filter can be used. Then, the estimation unit 9B takes the value with the maximum probability in the posterior distribution of the oil state information as the estimated value x' of the oil state information and the bearing load ｔ＿ｐｒｅ and outputs the same (step S651).

[0092] Returning to FIG. 18, the estimation unit 9B determines whether or not the estimation of the oil state information has been performed over a predetermined measurement time (step S700). If the time from the start of the estimation of the oil state information to the present has reached the measurement time (step S700: YES), the output unit 5 outputs the estimated values of the oil state information and the bearing load (step S801). If the time from the start of the estimation of the oil state information to the present has not reached the measurement time (step S700: NO), the sound wave prediction unit 7B updates the prior distribution with the estimated values of the oil state information and the bearing load (step S901). Then, the processes of step S301, step S401, step S501, and step S601 are repeatedly executed until the condition of step S700 is satisfied. That is, the update of predicted sound wave information, the acquisition of sound wave measurement values and the generation of likelihood distribution, and the calculation of estimated values of oil state information and bearing load are repeatedly performed.

[0093] Here, the process of step S901 will be specifically described with reference to FIG. 22. FIG. 22 is a flowchart showing the process of the sound wave prediction unit 7B according to the third embodiment. As shown in FIG. 22, first, the sound wave prediction unit 7B substitutes the value of the estimated value x' ｔ＿ｐｒｅ into x' of formula (11), which is a state equationｔ－１ Substitute the values ​​into the equation (step S911). Next, the sound wave prediction unit 7B updates the prior distribution by advancing t by one time, using the state equation of equation (11) above (step S921).

[0094] According to Embodiment 3, the observed variables in the observation equation are represented by probability distributions of the measured value of the sound wave and the error in the measured value of the sound wave, and the measured value of the torque input to the movable part of the sliding section and the error in the measured value of the torque. As a result, the bearing load input as design information can also be updated based on the measured value of the torque, thereby improving the accuracy of estimating the state of the lubricating oil.

[0095] Furthermore, the measured torque may not be used directly to predict the bearing load, but rather by other methods. For example, slight fluctuations in the rotational motion of the motor 140 will affect the torque. Therefore, the rotation period of the motor 140 and the timing of sound wave acquisition by the measurement value acquisition unit can be synchronized based on the torque measurement result. This makes it possible to keep the influence of the bearing load on the estimated value of the oil state information output constant.

[0096] Embodiment 4. In Embodiment 4, the sound wave generation model described in Embodiment 1 is determined by performing a test using the test device 15 before performing the lubrication oil state estimation method described in Embodiment 1. In other words, the lubrication oil state estimation device 1 of Embodiment 4 uses a sound wave generation model that has been calibrated in advance by a test with the test device 15 in the lubrication oil state estimation method. Figure 23 is a schematic diagram showing the test device 15 according to Embodiment 4. As shown in Figure 23, the test device 15 has a motor 63, a rotating shaft 150A, an upper bearing 163A, a lower bearing 164A, a support part 71, a lubrication pump 72, a torque meter 74, sound wave sensors 20A and 20B, and a sound wave processing device 30A. The test device 15 has a configuration corresponding to the sliding part 50 of the compressor 100 and is a device for schematically reproducing the sound waves generated in the sliding part 50 of the compressor 100.

[0097] Motor 63 rotates using power supplied from a commercial power source (not shown) and drives the rotating shaft 150A. The rotating shaft 150A is inserted into an annularly formed upper bearing 163A and lower bearing 164A. The rotating shaft 150A is rotated by motor 63. The upper bearing 163A and lower bearing 164A are slidably fitted onto the rotating shaft 150A. As a result, the rotating shaft 150A is rotatably supported by the upper bearing 163A and lower bearing 164A. A sliding portion 50 is formed by the outer circumferential surface of the rotating shaft 150A and the respective inner circumferential surfaces of the upper bearing 163A and lower bearing 164A. The upper bearing 163A and lower bearing 164A are sliding bearings that support the rotating shaft 150A. The support portion 71 supports the upper bearing 163A and lower bearing 164A. The support portion 71 has a pressurizing portion 71a that applies a load to the sliding portion 50. The pressurizing portion 71a, for example, presses the rotating shaft 150A in the direction of the upper bearing 163A and the lower bearing 164A, thereby replicating the load applied to the fixed part 51 of the sliding portion 50 of the compressor 100.

[0098] The lubrication pump 72 supplies lubricating oil to the rotating shaft 150A, the upper bearing 163A, and the lower bearing 164A through the supply passage 73. In Embodiment 4, normal lubricating oil that has not deteriorated is supplied. The torque meter 74 is connected to the rotating shaft 150A and measures the torque applied to the rotating shaft 150A. The sound wave sensors 20A and 20B detect sound wave signals. The sound wave sensors 20A and 20B are provided on the upper bearing 163A and the lower bearing 164A, respectively. The sound wave processing device 30A measures sound waves from the sound wave signals detected by the sound wave sensors 20A and 20B.

[0099] Using Figure 24, a method for determining the sound wave generation model by testing with the test apparatus 15 will be explained. Figure 24 is a flowchart of the method for determining the sound wave generation model according to Embodiment 4. As shown in Figure 24, first, the physical properties of normal oil to be introduced into the test apparatus 15 are measured (step S1301). Next, normal oil is put into the oil supply pump 72 shown in Figure 23 and supplied to the upper bearing 163A and lower bearing 164A through the supply passage 73. This introduces normal oil into the test apparatus 15 (step S1302). Subsequently, a sound wave sensor 20 is installed near the sound wave source (step S1303), and the sound wave sensor 20 is connected to the sound wave processing device 30A. In the case of the test apparatus 15 in Figure 23, the sound wave sensor 20 is installed at a location close to the rotating shaft 150A in the upper bearing 163A and lower bearing 164A. Then, the test apparatus 15 is operated under the same operating conditions as the equipment whose lubrication oil state is to be estimated (for example, the compressor 100) (step S1304). Specifically, the motor 63 and the pressurizing section 71a of the support section 71 are operated under conditions such that the same load applied to the sliding section 50 of the test device 15 is the same load applied to the fixed part 51 of the sliding section 50 during the operation of the equipment whose lubrication state is to be estimated. Then, when the operation of the test device 15 is stable, sound waves are measured by the sound wave sensors 20A and 20B and the sound wave processing device 30A (step S1305). At this time, the load applied by the pressurizing section 71a and the output of the torque meter 74 are also measured. Finally, based on the physical properties of the oil measured in step S1301 and the sound waves and load measured in step S1305, a sound wave generation model showing the relationship between the oil film thickness and the generated sound waves is determined (step S1306).

[0100] As described above, the sound wave generation model used in the lubricating oil state estimation device 1 of Embodiment 4 is calibrated based on the measurement results of sound waves generated in the test device 15 when normal lubricating oil is introduced into the test device 15 which reproduces the sliding part 50. Unlike the actual machine in which the lubricating oil state estimation device 1 is actually applied, the accuracy of estimating the state of the lubricating oil can be improved by calibrating the sound wave generation model in an environment where measurement is easy.

[0101] Furthermore, the load and torque applied by the pressurizing section 71a, measured when the test device 15 is operated under the same operating conditions as the equipment whose lubrication state is to be estimated, can be used to calibrate the data used to predict the bearing load from the torque measurement values ​​described in Embodiment 3.

[0102] Embodiment 5. In Embodiment 5, the sound wave generation model described in Embodiment 1 is determined by performing a test using the test apparatus 15 before performing the lubricating oil state estimation method described in Embodiment 1. In other words, the lubricating oil state estimation apparatus 1 of Embodiment 5 uses the sound wave generation model, which has been calibrated in advance by a test with the test apparatus 15, in the lubricating oil state estimation method. Embodiment 5 differs from Embodiment 4 in that deteriorated lubricating oil is used in the test. The test apparatus 15 is the same as the test apparatus 15 described in Embodiment 4. Therefore, the description of the test apparatus 15 is omitted.

[0103] Using Figure 25, a method for determining the sound wave generation model by testing with the test apparatus 15 will be explained. Figure 25 is a flowchart of the method for determining the sound wave generation model according to Embodiment 5. As shown in Figure 25, first, the physical properties of the deteriorated oil to be introduced into the test apparatus 15 are measured (step S1401). Next, the deteriorated oil is put into the oil supply pump 72 shown in Figure 23 and supplied to the upper bearing 163A and the lower bearing 164A through the supply passage 73. This introduces the deteriorated oil into the test apparatus 15 (step S1402). Subsequently, a sound wave sensor 20 is installed near the sound wave source (step S1403), and the sound wave sensor 20 is connected to the sound wave processing device 30A. In the case of the test apparatus 15 in Figure 23, the sound wave sensor 20 is installed at a location close to the rotation shaft 150A in the upper bearing 163A and the lower bearing 164A. Then, the test apparatus 15 is operated under the same operating conditions as the equipment whose lubrication oil condition is to be estimated (for example, the compressor 100) (step S1404). Specifically, the motor 63 and the pressurizing section 71a of the support section 71 are operated under conditions such that the same load applied to the sliding section 50 of the test apparatus 15 as the load applied to the fixing part 51 of the sliding section 50 during the operation of the equipment whose lubrication oil condition is to be estimated is applied. After that, when the operation of the test apparatus 15 is stable, sound waves are measured by the sound wave sensors 20A and 20B and the sound wave processing device 30A (step S1405). At this time, the load applied by the pressurizing section 71a and the output of the torque meter 74 are also measured. Finally, based on the physical properties of the oil measured in step S1401 and the sound waves and load measured in step S1405, a sound wave generation model showing the relationship between the oil film thickness and the generated sound waves is determined (step S1406).

[0104] As described above, the sound wave generation model used in the lubricating oil state estimation device 1 of Embodiment 5 is calibrated based on the measurement results of sound waves generated in the test device 15 when deteriorated lubricating oil is introduced into the test device 15, which reproduces the sliding part 50. Unlike the actual machine in which the lubricating oil state estimation device 1 is actually applied, the accuracy of estimating the state of the lubricating oil can be improved by calibrating the sound wave generation model in an environment where measurement is easy.

[0105] Embodiment 6. Embodiment 6 differs from Embodiment 1 in that the state equation used to generate the prior distribution of oil state information incorporates the change in the state of the lubricating oil over time (aging deterioration). The following will focus on explaining the differences from Embodiment 1.

[0106] Figure 26 is a flowchart of the lubricating oil state estimation method according to Embodiment 6. The lubricating oil state estimation method according to Embodiment 6 shares the same overall flow as the lubricating oil state estimation method of Embodiment 1. Therefore, in Figure 26, common steps are given the same step numbers and explanations are omitted for processes that are common to both Embodiment 1 and Embodiment 1, and the explanation will focus on the differing processes. Following step S100, the sound wave prediction unit 7 generates a prior distribution of oil state information (step S202). Here, the prior distribution x'' ｔ As shown in equation (19) below, it is generated by the equation of state, but unlike Embodiment 1, the physical properties of the oil at time t α(t) and β(t), and the initial value δ of the oil film thickness are used. ０ It is represented by a vector with (t) as its component: α(t) and β(t), and δ ０ (t) is a function that models the aging degradation of the lubricating oil, with time t as the argument, from the start of estimating the state of the lubricating oil to the present. Here, it is assumed that the measurement interval of sound waves, shown by t and t-1, is long and the physical properties of the oil have changed. Therefore, α(t) and β(t), and δ ０ (t) indicates that as oil deteriorates over time, even when the same load is applied, the oil film thickness will not be the same as before deterioration; in other words, the rate of change in oil film thickness differs in response to changes in load.

[0107]

[0108] Following step S202, the sound wave prediction unit 7 calculates predicted sound wave information that shows the sound waves predicted to be generated in the sliding part 50 as a probability distribution, based on the prior distribution and design information (step S302).

[0109] Here, the processing of step S302 will be specifically described with reference to Fig. 27. Fig. 27 is a flowchart showing the processing of the sound wave prediction unit 7 according to the sixth embodiment. First, the oil film thickness δ´´(t) at time t is expressed by the following formula (20), which is a modification of formula (2) described in the first embodiment. As shown in formula (20), the sound wave prediction unit 7 inputs the prior distribution of oil state information and design information into the function f() representing the oil film model, thereby obtaining the oil film thickness δ´´(t) at time t (step S312).

[0110]

[0111] Subsequently, the sound wave prediction unit 7 applies the oil film thickness δ´´(t) obtained in step S312 as shown in the following formula (21) to the function g representing a sound wave generation model that represents the relationship between the oil film thickness and the sound wave generated at the sliding portion 50 ｉ () to calculate predicted sound wave information A´´ ｐｒｅｄｉｃ (step S322). The predicted sound wave information A´´ ｐｒｅｄｉｃ is a prior distribution of sound waves predicted by inputting the prior distribution of oil state information into a physical model. For the subsequent processing, description is omitted because it is common except that the state equation of formula (19) and the predicted sound wave information of formula (20) are used instead of the state equation of formula (1) and the predicted sound wave information of formula (5) described in the first embodiment.

[0112]

[0113] According to the sixth embodiment, the sound wave prediction unit 7 obtains predicted sound wave information based on a state equation incorporating the state change of lubricating oil. Thereby, the estimation accuracy of the lubricating oil state can be improved.

[0114] Embodiment 7. Figure 28 is a functional block diagram showing the lubricating oil state estimation system 200 according to Embodiment 7. As shown in Figure 28, the lubricating oil state estimation system 200 consists of the lubricating oil state estimation device 1, input device 11, measurement device 12, and output device 13 of Embodiment 1. The lubricating oil state estimation device 1 is, for example, a PC, smartphone, tablet, or server device. The lubricating oil state estimation device 1 is not the one described in Embodiment 1, but may be one described in any of Embodiments 2 to 6. The lubricating oil state estimation device 1 may also be implemented by combining multiple devices to realize the functions of the lubricating oil state estimation device 1 described in Embodiments 1 to 6.

[0115] The input device 11 is a device for the user to input specified values ​​that the lubricating oil state estimation device 1 uses to estimate oil state information. The input device 11 is a device connected to the lubricating oil state estimation device 1 by wireless or wired connection. The input device 11 is, for example, a keyboard or microphone connected by wire to a server device. Alternatively, the input device 11 may be a keyboard or microphone connected by wire to a PC connected to the server device via a network. Furthermore, the input device 11 may be a touch panel of a smartphone or tablet connected to the server device via a network. For example, in the case of the lubricating oil state estimation device 1 of Embodiment 1, the input device 11 accepts input of the initial values ​​and variations of the oil properties of the lubricating oil supplied to the sliding part 50, the initial values ​​and variations of the oil film thickness, the design value of the bearing load, and the bearing shape. Furthermore, the input device 11 connected to the lubricating oil state estimation device 1 of Embodiment 2 accepts input of representative parameters of the sound wave sensor 20 (for example, the measurement position or transmission distance of the sound wave) in addition to the information input from the input device 11 connected to the lubricating oil state estimation device 1 of Embodiment 1.

[0116] The measuring device 12 is the sound wave sensor 20 and sound wave processing device 30 described in Embodiments 1 to 6, or the current and voltage sensor 62. The measuring device 12 is connected to the lubricating oil state estimation device 1 by wire or wireless connection and transmits the measured sound waves, or the measured current and voltage, to the lubricating oil state estimation device 1.

[0117] The output device 13 is a device for visually or audibly displaying to the user the estimated value of the oil state information output by the output unit 5 of the lubricating oil state estimation device 1. The output device 13 is connected to the lubricating oil state estimation device 1 by wireless or wired connection. The output device 13 is, for example, a display or speaker wired to a server device. Alternatively, the output device 13 may be a display or speaker of a PC connected to the server device via a network. Furthermore, the output device 13 may be a touch panel of a smartphone or tablet connected to the server device via a network.

[0118] As described above, in the lubricating oil state estimation system of Embodiment 7, similar to Embodiment 1, an estimated value of oil state information is calculated based on a predictive sound wave information as a state variable, an observation equation showing the correlation between the measured sound wave value and an observed variable that represents the measured value of the sound wave using a probability distribution including the error of the measured value, and a prior distribution of oil state information. In this way, the lubricating oil state estimation system of Embodiment 7 calculates an estimated value of oil state information based on the measured value of the sound wave, so it can estimate the state of the lubricating oil supplied to the sliding part with greater accuracy than simply estimating the state of the lubricating oil from the length of use of the lubricating oil, and can contribute to the maintenance of equipment having sliding parts.

[0119] Furthermore, users can determine whether the lubricant has deteriorated by comparing the estimated state of the lubricant with the state of a normal lubricant. In addition, by modeling the changes in the physical properties of the lubricant over time, it is possible to predict the progression of the lubricant's deterioration based on the estimated state and the model, and to determine the limit of the lubricant's use and the timing for replacement.

[0120] The above describes the embodiments of the present disclosure, but the present disclosure is not limited to the configurations of the embodiments described above, and various modifications are possible within the scope of its technical idea. Furthermore, the present disclosure includes all possible combinations of the configurations shown in each of the embodiments described above. For example, in Embodiment 1, in equations (2) to (5), the function f() represents the oil film model and the function g() represents the sound wave generation model. ｉAs shown in parentheses, the relationship between oil film thickness and sound waves is modeled. However, in addition to oil film thickness, the spring stiffness of the oil film or the bearing load may also be used as parameters to model the relationship with sound waves. When using the spring stiffness of the oil film or the bearing load as parameters to model the relationship with sound waves, a function f() representing the oil film model and a function g() representing the sound wave generation model are used based on the physical models of these parameters. ｉ Change the expression that uses parentheses.

[0121] Furthermore, in each embodiment, the estimation unit 9 may determine whether the estimated value of the oil state information is the value of normal lubricating oil. The oil state information of normal lubricating oil is assumed to be stored in advance in the storage unit 6. In addition, if the estimated value of the oil state information is not the value of normal lubricating oil, the output unit 5 may output a control signal to the control device of the equipment to which the lubricating oil state estimation device 1 is applied, provided that the equipment to which the lubricating oil state estimation device 1 is applied is a device that can control the load applied to the sliding part 50 or the torque that generates the load. The control signal is a signal to instruct the control device of the equipment to which the lubricating oil state estimation device 1 is applied to control the load applied to the sliding part 50 or the torque that generates the load so that the load applied to the sliding part 50 does not exceed a predetermined threshold. Note that the output unit 5 may also directly control the equipment to which the lubricating oil state estimation device 1 is applied. Furthermore, if the equipment to which the lubricating oil state estimation device 1 is applied is equipment capable of controlling the amount of lubricating oil supplied to the sliding part 50, the device may output a control signal to the control device of the equipment to which the lubricating oil state estimation device 1 is applied, which controls the amount of oil supplied to the sliding part 50. In this case as well, the output unit 5 may directly control the equipment to which the lubricating oil state estimation device 1 is applied.

[0122] 1, 1A, 1B Lubrication oil state estimation device, 2 Specified value acquisition unit, 3, 3B Measured value acquisition unit, 4, 4A, 4B Calculation unit, 5 Output unit, 6 Memory unit, 7, 7B Sound wave prediction unit, 8, 8B Measured value processing unit, 9, 9B Estimation unit, 11 Input device, 12 Measurement device, 13 Output device, 15 Test device, 20, 20A, 20B, 20_1, 20_2, 20_n Sound wave sensor, 30, 30A Sound wave processing device, 41 Processing circuit, 42 Processor, 43 Memory, 44 Bus, 50 Sliding part, 51 Fixed part, 52 Movable part, 61 Power supply, 62 Current voltage sensor, 63 Motor, 71 Support part, 71a Pressurizing part, 72 Pump, 73 Supply path, 74 Torque meter, 100 Compressor, 110 Housing, 120 Intake pipe, 120a Intake muffler, 130 Discharge pipe, 140 Motor, 150, 150A Rotating shaft, 150a Main shaft, 150b Crank, 150c Sub-shaft, 160 Compression mechanism, 160a Discharge muffler, 161 Cylinder, 161a Cylinder chamber, 161A Cylinder, 162 Rolling piston, 163, 163A Upper bearing, 164, 164A Lower bearing, 200 Lubrication oil state estimation system.

Claims

1. A lubricating oil state estimation device for estimating oil state information indicating the state of lubricating oil supplied to a sliding part having a fixed part and a movable part that slides with respect to the fixed part, comprising: a sound wave prediction unit that obtains predicted sound wave information indicating the sound waves predicted to be generated in the sliding part by a probability distribution, based on a prior distribution of the oil state information based on a state equation that indicates the time change of the oil state information by a probability distribution, and design information of the sliding part; and an estimation unit that calculates an estimated value of the oil state information based on an observation equation that shows the correlation between the predicted sound wave information as a state variable and an observation variable that indicates the measured value of the sound wave by a probability distribution including the error of the measured value of the sound wave, and the prior distribution of the oil state information.

2. The lubricating oil state estimation device according to claim 1, wherein the sound wave prediction unit inputs the oil state information and the design information into a sound wave generation model that models the relationship between the oil film thickness, which changes due to the load applied from the movable part to the fixed part in the sliding part, and the sound waves generated, in order to obtain the predicted sound wave information.

3. The lubricating oil state estimation device according to claim 2, wherein the sound wave prediction unit obtains the predicted sound wave information by combining the sound wave generation model and a sound wave transmission model that models the transmission of sound waves from the sliding part to the sound wave measurement position.

4. The lubricating oil state estimation device according to claim 3, wherein the sound wave transmission model is output by inputting the representative parameters to a trained model that outputs the sound wave transmission model when different representative parameters are given as input at each of the multiple measurement positions, and the trained model is constructed by learning the relationship between sound waves measured at the multiple measurement positions and the different representative parameters at each of the multiple measurement positions.

5. The lubricant state estimation device according to any one of claims 2 to 4, wherein the sound wave generation model is calibrated based on the measurement results of sound waves generated in a test device that reproduces the sliding part when normal lubricant is introduced into the test device.

6. The lubricant condition estimation device according to any one of claims 2 to 4, wherein the sound wave generation model is calibrated based on the measurement results of sound waves generated in a test device that reproduces the sliding part when deteriorated lubricant is introduced into the test device.

7. The lubrication oil state estimation device according to any one of claims 1 to 6, wherein the observed variables represent, by probability distribution, the measured value of the sound wave and the error of the measured value of the sound wave, and the measured value of the torque input to the movable part of the sliding part and the error of the measured value of the torque.

8. The lubricating oil state estimation device according to any one of claims 1 to 7, wherein the estimation unit calculates an estimated value of the oil state information using a particle filter.

9. The lubricating oil state estimation device according to any one of claims 1 to 8, wherein the sound wave prediction unit obtains the predicted sound wave information based on the state equation that incorporates the state change of the lubricating oil.

10. The lubricating oil state estimation device according to any one of claims 1 to 9, wherein the oil state information is the oil properties of the lubricating oil and the thickness of the oil film.

11. The lubrication oil state estimation device according to any one of claims 1 to 10, wherein the design information is the shape of the sliding part and the design value of the load on the fixed part.

12. A lubricating oil state estimation system comprising: a lubricating oil state estimation device according to any one of claims 1 to 11; and a measuring device that measures sound waves generated in the sliding part and transmits them to the lubricating oil state estimation device.

13. The lubricating oil state estimation system according to claim 12, wherein the measuring device is an AE sensor.

14. The lubrication oil state estimation system according to claim 12 or 13, wherein the sliding part is the rotating shaft and bearing of the compressor, and the measuring device measures sound waves on the surface of the housing that houses the rotating shaft and bearing of the compressor.

15. A method for estimating oil state information indicating the state of lubricating oil supplied to a sliding part having a fixed part and a movable part that slides with respect to the fixed part, comprising: a sound wave prediction step of obtaining predicted sound wave information indicating the sound waves predicted to be generated in the sliding part by a probability distribution, based on a prior distribution of the oil state information based on a state equation that indicates the time change of the oil state information by a probability distribution, and design information of the sliding part; and an estimation step of calculating an estimated value of the oil state information based on an observation equation that shows the correlation between the predicted sound wave information as a state variable and an observation variable that indicates the measured value of the sound wave by a probability distribution including the error of the measured value of the sound wave, and the prior distribution of the oil state information.

16. The lubrication oil state estimation method according to claim 15, comprising: a step of predicting sound waves generated in the sliding part using a sound wave generation model that models the generation of sound waves in the sliding part; and a step of obtaining predicted sound wave information indicating sound waves generated in the sliding part and transmitted to the measurement position using a sound wave transmission model that models the transmission of sound waves from the sliding part to the sound wave measurement position.

Images