JP7103878B2 - Temperature estimation device and life evaluation device - Google Patents

Description

The present invention relates to a temperature estimation device and a life evaluation device.

Conventionally, the friction coefficient of the transmission mechanism is identified from the motor current and the motor speed, the temperature of the lubricant is estimated from the relationship between the friction coefficient obtained in advance and the temperature of the lubricant (grease), and the estimated lubricant is used. There is known a life evaluation device that corrects the input speed of the transmission mechanism in consideration of the temperature and evaluates the life of the lubricant from the corrected input speed (see, for example, Patent Document 1). The life evaluation device described in Patent Document 1 has an inconvenience that the temperature of the lubricant is obtained only from the friction coefficient of the transmission mechanism, so that an error is large and the life cannot be evaluated correctly.

Therefore, in order to eliminate such inconvenience, the amount of heat generated by the motor calculated based on the current value of the motor, the rotation speed of the motor, and the friction coefficient of the transmission mechanism for transmitting the power of the motor to the moving portion are used. A method has been proposed in which the temperature of the lubricant used in the transmission mechanism is estimated using the calculated frictional calorific value in the transmission mechanism, and the life of the lubricant is estimated based on the estimated temperature of the lubricant. There is.

Japanese Patent No. 4021354

In the above method, for example, the following equation (1) is used to estimate the temperature of the lubricant. In Equation 1, T is the estimated temperature of the lubricant, T ₀ is the room temperature, subscript i is the axis number that affects the lubricant temperature of the estimated target axis including the estimated target axis, and D ₁ to D ₆ are experiments in advance. E _i is the calorific value of the motor W _1i coefficient, F _i is friction calorific value W _2i coefficient, W ₁ is motor calorific value, W ₂ is friction calorific value, W ₃ is air-cooled heat dissipation amount, W ₄ is calorific value of other heat generation sources. There is. However, when estimating the temperature of the lubricant using such an equation, it is necessary to identify D ₁ to D ₆ , E _i , and _Fi in advance.

Therefore, an object of the present invention is a temperature estimation device that enables an operator to estimate the temperature of the lubricant from the motor current and the motor speed without manually identifying the coefficient of the temperature estimation formula of the lubricant in advance. It is to provide a life evaluation device.

One aspect of the present invention is a life evaluation device for evaluating the life of the lubricating material in a machine including a motor and a transmission mechanism lubricated by the lubricating material and transmitting the power of the motor to a movable portion. Friction calorific value calculation that calculates the friction calorific value in the transmission mechanism based on the motor calorific value calculation unit that calculates the motor calorific value based on the current value, the rotation speed of the motor, and the friction coefficient of the transmission mechanism. The temperature of the lubricant is calculated from the friction calorific value and the motor calorific value by machine learning using the unit, the friction calorific value calculated by the friction calorific value calculation unit, and the motor calorific value calculated by the motor calorific value calculation unit. It is a temperature estimation device including a lubricant temperature learning unit that generates a trained model for estimation.

Another aspect of the present invention is a temperature estimation device for estimating the temperature of the lubricating material in a machine including a motor and a transmission mechanism lubricated by the lubricating material and transmitting the power of the motor to a movable portion. A frictional calorific value that calculates the frictional calorific value in the transmission mechanism based on the motor calorific value calculation unit that calculates the motor calorific value based on the current value, the rotation speed of the motor, and the friction coefficient of the transmission mechanism. A learning model storage unit that stores a learning model generated by machine learning using the calculation unit, the friction calorific value calculated by the friction calorific value calculation unit, and the motor calorific value calculated by the motor calorific value calculation unit. And, based on the friction calorific value calculated by the friction calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit, the temperature of the lubricant of the motor is estimated using the trained model. It is a temperature estimation device including a lubricant temperature estimation unit.

Another aspect of the present invention is a temperature estimation device for estimating the temperature of the lubricating material in a machine including a motor and a transmission mechanism lubricated by the lubricating material and transmitting the power of the motor to a movable portion. A frictional calorific value that calculates the frictional calorific value in the transmission mechanism based on the motor calorific value calculation unit that calculates the motor calorific value based on the current value, the rotation speed of the motor, and the friction coefficient of the transmission mechanism. A learning model storage unit that stores a learning model generated by machine learning using the calculation unit, the friction calorific value calculated by the friction calorific value calculation unit, and the motor calorific value calculated by the motor calorific value calculation unit. And, based on the friction calorific value calculated by the friction calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit, the temperature of the lubricant of the motor is estimated using the trained model. It is a life evaluation device including a friction material temperature estimation unit and a life estimation unit that estimates the life of the lubricant based on the temperature of the lubricant estimated by the lubricant temperature estimation unit.

According to the present invention, it is possible to estimate the temperature of the lubricant from the motor current and the motor speed without the operator manually identifying the coefficient of the temperature estimation formula of the lubricant in advance.

It is a schematic hardware block diagram of the life evaluation apparatus by one Embodiment. It is a schematic functional block diagram of the life evaluation apparatus by one Embodiment. It is a figure which illustrates the relationship between the temperature and the life of a lubricant.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a schematic hardware configuration diagram showing a main part of the life evaluation device according to the first embodiment. The life evaluation device 1 is an embodiment of the temperature estimation device 1'according to the present invention, and estimates the life of the lubricant based on the temperature of the lubricant estimated by using the temperature estimation method of the lubricant of the present invention. Is what you do. The life evaluation device 1 can be mounted on, for example, a control device that controls a robot. Further, the life evaluation device 1 is a computer such as a robot, a personal computer attached to the robot control device, a cell computer, a host computer, an edge server, a cloud server, etc. connected to the robot control device by a wired / wireless network. Can be implemented as. In this embodiment, an example in which the life evaluation device 1 is mounted on a control device that controls a robot is shown.

The life evaluation device 1 of the present embodiment has a function used for controlling a general robot, and is connected to the robot 2 illustrated in FIG. 1 via an interface 19 to control the robot 2. The robot 2 has one or more links (movable parts) and one or more joints, and examples thereof include a 6-axis articulated robot. The joint portion of the robot 2 is equipped with a speed reducer (transmission mechanism) and a motor that generates power to be input to the speed reducer, and the inside of the speed reducer is filled with a lubricant (grease). Lubricates the gears and bearings that make up the reducer. Then, when the motor included in the robot 2 is operated, the rotation of the motor is decelerated and transmitted by the speed reducer, and the joint portion rotates.

The robot 2 controlled by the life evaluation device (control device) 1 is responsible for, for example, work for transporting workpieces, assembling parts, and other operations for manufacturing. The life evaluation device 1 interfaces with the temperature of the installation location of the robot 2 and the state quantity of the motor driving each joint of the robot 2 from the set value input by the operator and the value acquired from each part of the robot 2. It can be obtained via. Since the detailed configuration of the robot 2 is already known, detailed description in the present specification will be omitted.

The CPU 11 included in the life evaluation device 1 according to the present embodiment is a processor that controls the life evaluation device 1 as a whole. The CPU 11 reads the system program stored in the ROM 12 via the bus 20, and controls the entire life evaluation device 1 according to the system program. Temporary calculation data, display data, various data input by the operator via the input device 71, and the like are temporarily stored in the RAM 13.

The non-volatile memory 14 is configured as a memory in which the storage state is maintained even when the power of the life evaluation device 1 is turned off, for example, by backing up with a battery (not shown). The non-volatile memory 14 includes a setting area for storing setting information related to the operation of the life evaluation device 1, a control program for the robot 2 input from the input device 71, and a robot 2 read from an external storage device (not shown). The control program of the above, the state amount of the motor acquired from the robot 2 via the interface 19, and the like are stored. The program and various data stored in the non-volatile memory 14 may be expanded in the RAM 13 at the time of execution / use. Further, various system programs such as a known analysis program (including a system program for controlling interaction with the machine learning device 100, which will be described later) are written in the ROM 12 in advance.

On the display device 70, each data read into the memory, data obtained as a result of executing a program or the like, data output from the machine learning device 100 described later, and the like are output and displayed via the interface 17. Will be done. Further, the input device 71 composed of a keyboard, a pointing device, and the like receives commands, data, and the like based on operations by the operator and passes them to the CPU 11 via the interface 18.

The interface 21 is an interface for connecting the life evaluation device 1 and the machine learning device 100. The machine learning device 100 stores a processor 101 that controls the entire machine learning device 100, a ROM 102 that stores a system program and the like, a RAM 103 that temporarily stores each process related to machine learning, a learning model, and the like. The non-volatile memory 104 used for the above is provided. The machine learning device 100 obtains each information that can be acquired by the life evaluation device 1 via the interface 21 (for example, the room temperature set via the input device 71, the state amount of each motor acquired from the robot 2, and the like). It can be observed. Further, the life evaluation device 1 acquires the processing result from the machine learning device 100 via the interface 21.

FIG. 2 is a schematic functional block diagram of the life evaluation device 1 and the machine learning device 100 according to the embodiment. In each functional block shown in FIG. 2, the CPU 11 included in the life evaluation device 1 shown in FIG. 1 and the processor 101 of the machine learning device 100 execute their respective system programs, and the life evaluation device 1 and the machine learning device are executed. It is realized by controlling the operation of each part of 100.

The life evaluation device 1 includes a control unit 30 that controls the robot 2, a motor heat generation calculation unit 32 that calculates the heat generation amount of the motor included in the robot 2, and a friction heat generation that calculates the frictional heat generation amount in the speed reducer provided in the robot 2. The amount calculation unit 34, the air-cooled heat radiation amount calculation unit 36, and the lubricant temperature learning unit 38 that learns the temperature of the lubricant of the robot 2 based on the heat amount calculated by these calculation units 32, 34, 36, respectively. A lubricant temperature estimation unit 40 that estimates the temperature of the lubricant of the robot 2 based on the amount of heat calculated by the calculation units 32, 34, 36, and a life estimation unit that calculates the life of the lubricant based on the estimated temperature. It has 42 and.

The control unit 30 controls the robot 2 based on a control operation of an operation panel (not shown) by an operator or a control program stored in a non-volatile memory 14 or the like. When the control program commands the movement of each axis (joint) included in the robot 2, the control unit 30 outputs command data as a change amount of the axis angle to the motor driving the axis for each control cycle. , Etc., which are provided with functions for general control required for controlling each part of the robot 2. Further, the control unit 30 acquires the motor state amount (motor current value, motor position, speed, acceleration, motor torque, etc.) of each motor included in the robot 2 and transfers them to the calculation units 32, 34, and 36. Output.

The motor calorific value calculation unit 32 uses the following equation 2 based on the motor state amount input from the control unit 30, that is, the motor current value (motor current value) and the motor rotation speed (motor rotation speed). Use to calculate the amount of heat generated by the motor. In Equation 2, _{W 1} is the amount of heat generated by the motor, A ₁ , A ₂ , and A ₃ are the coefficients identified in advance by experiments, etc., IM is the motor current value, and _SM is the motor rotation speed. The first term shows the copper loss in the motor, and the second term shows the iron loss.

The friction calorific value calculation unit 34 calculates the friction calorific value using the following equations 3 based on the motor state amount input from the control unit 30, that is, the friction torque and the motor rotation speed. In Equation 3, _{W 2} is the amount of heat generated by friction, T is the friction torque, B ₁ , B ₂ , and B ₃ are the coefficient of friction identified in advance by experiments, and TM is the link of the robot when there is no friction. It is the torque required to drive the. The torque _TM is generally calculated by a known calculation method such as the Lagrange method or the Newton-Euler method using the position, acceleration, mass information, etc. of the link of the robot 2.

The air-cooled heat radiation amount calculation unit 36 calculates the air-cooled heat radiation amount generated because the speed reducer itself moves in the air by driving the robot 2 and a relative speed is generated with the surrounding air. Based on the rotation speed of the motor input from the control unit 30, the moving speed of the robot 2 at the position of the reducer is calculated, and based on the calculated moving speed, the amount of air-cooled heat radiation is calculated using the following equation 4 calculate. In Equation 4, W ₃ is the air-cooled heat dissipation amount, C ₁ is the value calculated by the sum of the motor heat generation amount W ₁ and the friction heat generation amount W ₂ , and S _R is the moving speed of the robot 2 at the position of the reducer. be. The air-cooled heat dissipation amount calculation unit 36 may be omitted if it does not have a large effect on the temperature change of the lubricant as compared with the motor heat generation amount and the friction heat generation amount.

The lubricant temperature learning unit 38 performs supervised learning based on the calorific value calculated by each calculation unit 32, 34, 36 and the temperature of the lubricant, and creates a learned model used for estimating the lubricant temperature. It is a functional means (learning). The lubricant temperature learning unit 38 of the present embodiment performs regression analysis using, for example, the multiple regression equation of the above equation 1 as a learning model, and the coefficients D ₁ to D ₆ and E _i used in the equation 1 are used. , Create a trained model by _estimating all or part of Fi.

The learning by the lubricant temperature learning unit 38 is performed when the life evaluation device 1 is functioning as a learning mode. At this time, the lubricant temperature learning unit 38 creates teacher data using the calculation units 32, 34, and 36 as input data and the temperature of the lubricant as a label (output data), and uses the created teacher data as the input data. Perform supervised learning using. The temperature of the lubricant as a label (output data) may be set manually by, for example, the temperature of the lubricant measured by the operator via the input device 71, or may be installed inside or near the speed reducer. The temperature value detected by the temperature sensor (not shown) may be automatically acquired. In this way, based on the plurality of teacher data obtained in each pattern by executing the operations of various patterns, the lubricant temperature learning unit 38 uses the amount of heat calculated by the calculation units 32, 34, and 36. Generate a trained model for estimating the temperature of the lubricant. Then, the learned model generated by the lubricant temperature learning unit 38 is stored in the learning model storage unit 44 provided on the non-volatile memory 104, and is used for estimating the lubricant temperature by the lubricant temperature estimation unit 40.

The room temperature T ₀ in the equation 1 may be set manually by the operator via, for example, the input device 71, or the temperature value detected by a temperature sensor (not shown) for measuring the room temperature may be set. It may be acquired automatically. For the other heat generation source W ₄ of the equation 1, the amount of heat generated from a nearby device is set in advance. Further, when estimating a part of the coefficient, a predetermined value is set in advance for the other coefficient of the coefficient to be estimated.

The lubricant temperature learning unit 38 is not always necessary for the life evaluation device 1 after the trained model that has been sufficiently learned is generated. For example, the manufacturer of the life evaluation device 1 ships the life evaluation device 1 after the lubricant temperature learning unit 38 is disabled in a state where the fully learned trained model is stored in the learning model storage unit 44. You may do so.

The lubricant temperature estimation unit 40 estimates the temperature of the lubricant using the learned model stored in the learning model storage unit 44 based on the amount of heat calculated by the calculation units 32, 34, and 36. The lubricant temperature estimation unit 40 of the present embodiment uses a trained model as a trained model in which each coefficient is estimated by supervised learning by the lubricant temperature learning unit 38, and is calculated by the calculation units 32, 34, and 36. The temperature of the lubricant is estimated (calculated) by substituting the calculated amount of heat.

The lubricant temperature estimation unit 40 estimates the lubricant temperature when the life evaluation device 1 is functioning as the estimation mode. The lubricant temperature estimation unit 40 is not always necessary for the life evaluation device 1 at the stage of learning by the lubricant temperature learning unit 38. For example, while the manufacturer of the life evaluation device 1 is generating the trained model, the trained model may be generated after the lubricant temperature estimation unit 40 is disabled.

As an example of a learned model (equation 1 set) created by the lubricant temperature learning unit 38 and used by the lubricant temperature estimation unit 40 in the present embodiment, each axis of the robot 2 which is a 6-axis articulated robot is J. When ₁ axis, J ₂ axis, J ₃ axis, J ₄ axis, J ₅ axis, and J ₆ axis are used, the J ₄ axis, J ₅ axis, and J ₆ axis motors and reducers are densely packed and affect each other. Is given, the estimated temperature T of the lubricant with the J ₆ axis as the estimation target axis is estimated by the equation 5. In Equation 5, E ₄ is the coefficient of the motor calorific value W ₁₄ of the J ₄ -axis, E ₅ is the coefficient of the motor calorific value W ₁₅ of the J ₅ -axis, and E ₆ is the coefficient of the motor calorific value W ₁₆ of the J ₆ axis. The coefficient, F ₄ is the coefficient of the friction calorific value W ₂₄ on the J ₄ axis, F ₅ is the coefficient of the friction calorific value W ₂₅ on the J ₅ axis, and F ₆ is the coefficient of the friction calorific value W ₂₆ on the J ₆ axis.

In addition, the equation ₅ exemplifies the case where the J ₁ axis, the J ₂ axis and the J ₃ axis do not affect the J ₄ axis and the J ₅ axis affect the J ₆ axis. _If the _2nd axis and the _J3 axis also affect the _J6 axis, _W11 , _W12 , _W13 , _W21 , _W22 , _W23 , _E1 , E2 _, E3, _F1 , _F2 , F ₃ may be added to Equation 5 to calculate the estimated temperature T of the lubricant.
Further, by setting the coefficients D ₅ and D ₆ to 1 in the equation 5, the temperature of the lubricating oil can be estimated using the equation 5 as a simple linear model. By doing so, it is possible to reduce the load on the calculation at the time of learning and at the time of estimation.

The life estimation unit 42 calculates the life usage amount of the lubricant using the following equation 6 based on the estimated temperature of the lubricant. In general, the temperature and life of a lubricant have a relationship as shown in FIG. That is, when the temperature of the lubricant is T _th or less, the life of the lubricant is a constant rated life Sg ₀ , whereas when the temperature is higher than T _th , the life of the lubricant decreases exponentially. Based on the estimated temperature and time, the life usage can be calculated using the equation 6. In Equation 6, Sg is the amount of the lubricant used for the life, A is a constant identified in advance by experiments, etc., k is (T-T _th ) / Δ when T> T _th , and 0 when T ≤ T _th . Δ is the difference from T _th of the temperature at which the amount of life used is A times.

Then, the life of the lubricant calculated by the life estimation unit 42 using the equation 6 is displayed and output to the display device 70. The life of the lubricant calculated by the life estimation unit 42 may be transmitted and output to a host computer, a cloud server, or the like via a network (not shown), for example. Further, the life estimation unit 42 may calculate the remaining life of the lubricant by using the following equation (7) and output the calculated remaining life.

The life evaluation device 1 of the present embodiment configured as described above performs learning based on a plurality of teacher data obtained by causing the robot 2 to perform various patterns of movements, and a sufficient learning result is obtained. The temperature of the lubricant can be estimated with high accuracy using the trained model. Then, by estimating the life of the lubricant based on the estimated temperature of the lubricant, it becomes possible to accurately determine the replacement time of the lubricant.

As a modification of the life evaluation device 1 of the present embodiment, the lubricant temperature learning unit 38 uses, for example, a neural network as a learning model, and each calculation unit 32, 34, 36 is used as input data, and the temperature of the lubricant. It may be configured to perform supervised learning based on the teacher data having the label (output data). In this case, a neural network having three layers of an input layer, an intermediate layer, and an output layer may be used as the learning model, but so-called deep learning using a neural network having three or more layers. It is also possible to configure it to perform more effective learning and inference by using the method of.

In this modification, the lubricant temperature learning unit 38 is based on a plurality of teacher data obtained in each pattern by executing various patterns of operations, as in the case of using the lubricant temperature estimation formula as a learning model. Generates a trained model for estimating the temperature of the lubricant from the amount of heat calculated by each of the calculation units 32, 34, and 36. Then, the learned model generated by the lubricant temperature learning unit 38 is stored in the learning model storage unit 44 provided on the non-volatile memory 104, and is used for estimating the lubricant temperature by the lubricant temperature estimation unit 40.

Then, the lubricant temperature estimation unit 40 estimates the temperature of the lubricant using the learned model stored in the learning model storage unit 44 based on the amount of heat calculated by the calculation units 32, 34, and 36. .. In the lubricant temperature estimation unit 40 of this modification, the neural network generated (parameters are determined) by supervised learning by the lubricant temperature learning unit 38 is used as a trained model, and the calculation units 32, 34, respectively. By inputting the amount of heat calculated by 36, the temperature of the lubricant is directly estimated (calculated).

As another modification of the life evaluation device 1 of the present embodiment, in the learning by the lubricant temperature learning unit 38, a known regularization term (for example, L2 regularization term, sparse regularization term, etc.) is used instead of the usual regression analysis. Regression analysis may be performed in consideration of. By adopting this method, for example, even when the amount of teacher data that can be used for learning is small, overfitting is less likely to occur and the generalization performance of the trained model can be improved.

Although the embodiments of the present invention have been described above, the present invention is not limited to the examples of the above-described embodiments, and can be implemented in various embodiments by making appropriate changes.

For example, the learning algorithm and the calculation algorithm executed by the machine learning device 100, the control algorithm executed by the life evaluation device 1, and the like are not limited to those described above, and various algorithms can be adopted.

Further, in the above-described embodiment, the life evaluation device 1 and the machine learning device 100 are described as devices having different CPUs, but the machine learning device 100 is a system stored in the CPU 11 included in the life evaluation device 1 and the ROM 12. It may be realized by a program.

In the above embodiment, an example of estimating the life of the lubricant at the joint portion of the robot 2 by the life evaluation device 1 is shown, but the target for estimating the life of the lubricant is, for example, a machine tool, an electric vehicle, a railroad, or a power generator. It can also be applied to machines and the like.

In the above embodiment, the life evaluation device 1 estimates the temperature of the lubricant and estimates the life of the lubricant based on the estimation result. For example, the life estimation unit 42 is removed from the life evaluation device 1 to lubricate the lubricant. It may be configured as a temperature estimation device 1'for estimating the temperature of the material, and may be configured to output (display output, transmission output, etc.) the temperature of the lubricating material estimated by the temperature estimation device 1'. With this configuration, the layout for estimating the temperature of the lubricant and the life estimation unit 42 are mounted on different devices, and the arrangement of each functional means can be determined by design. good.

In the above-described embodiment, the life evaluation device 1 is shown as an example of the temperature estimation device 1'of the present invention, but the temperature of the lubricant estimated by the temperature estimation device 1'is other than the life evaluation of the lubricant. It may be used for the purpose.
For example, the temperature of the lubricant estimated by the lubricant temperature estimation unit 40 is output to the cooling unit for cooling the lubricant (bearing, speed reducer) provided in the robot 2 to control the cooling function. You may do so. With this configuration, the lubricant of the transmission mechanism is cooled by the cooling unit that operates according to the temperature of the lubricant estimated based on the motor state amount, and the temperature of the lubricant is kept below a predetermined temperature. , It becomes possible to extend the life of the lubricant and the members around it.
More specifically, when the robot 2 is provided with a cooling fan that is driven by a cooling unit and blows air to the bearing of the transmission mechanism, the cooling unit is a transmission mechanism estimated by the lubricant temperature estimation unit 40. The fan is driven when the estimated temperature T of the lubricant of the bearing of the bearing becomes equal to or higher than a preset threshold value. On the other hand, the cooling unit does not drive the fan when the estimated temperature T of the lubricant of the bearing of the transmission mechanism is less than the threshold value. As a result, the temperature of the bearing is cooled when it changes from below the threshold value to above the threshold value.

Further, the drive control of the motor may be performed according to the temperature of the lubricant estimated by the lubricant temperature estimation unit 40. With this configuration, the drive of the motor, which is a heat generating source, is controlled according to the temperature of the lubricant of the transmission mechanism, the temperature of the lubricant is kept below a predetermined temperature, and the life of the lubricant and its surrounding members is extended. Will be able to.
More specifically, in the control unit 30 that controls the motor of the robot 2, when the estimated temperature T of the lubricant of the bearing of the transmission mechanism estimated by the lubricant temperature estimation unit 40 becomes equal to or higher than a preset threshold value. In addition, the drive of the motor is stopped or decelerated. On the other hand, when the estimated temperature T of the lubricant of the bearing of the transmission mechanism is less than the threshold value, the driving of the motor is continued. As a result, the temperature of the bearing is cooled when it changes from below the threshold value to above the threshold value.

1 Life evaluation device 1'Temperature estimation device 2 Robot 11 CPU
12 ROM
13 RAM
14 Non-volatile memory 17, 18, 19 Interface 20 Bus 21 Interface 30 Control unit 32 Motor calorific value calculation unit 34 Friction calorific value calculation unit 36 Air-cooled heat dissipation calculation unit 38 Lubricating material temperature learning unit 40 Lubricating material temperature estimation unit 42 Life estimation Unit 44 Learning model storage unit 70 Display device 71 Input device 100 Machine learning device 101 Processor 102 ROM
103 RAM
104 Non-volatile memory

Claims (10)

A temperature estimation device that estimates the temperature of the lubricant in a machine including a motor and a transmission mechanism that is lubricated by the lubricant and transmits the power of the motor to a moving portion.
A motor calorific value calculation unit that calculates the motor calorific value based on the current value of the motor, and a motor calorific value calculation unit.
A friction calorific value calculation unit that calculates the friction calorific value in the transmission mechanism based on the rotation speed of the motor and the friction coefficient of the transmission mechanism.
To estimate the temperature of the lubricant from the friction calorific value and the motor calorific value by machine learning using the friction calorific value calculated by the friction calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit. Lubricant temperature learning unit to generate a trained model of
A temperature estimation device including. A temperature estimation device that estimates the temperature of the lubricant in a machine including a motor and a transmission mechanism that is lubricated by the lubricant and transmits the power of the motor to a moving portion.
A motor calorific value calculation unit that calculates the motor calorific value based on the current value of the motor, and a motor calorific value calculation unit.
A friction calorific value calculation unit that calculates the friction calorific value in the transmission mechanism based on the rotation speed of the motor and the friction coefficient of the transmission mechanism.
A learning model storage unit that stores a learned model generated by machine learning using the friction calorific value calculated by the friction calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit.
Lubrication that estimates the temperature of the lubricant of the motor using the trained model based on the frictional calorific value calculated by the frictional calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit. Material temperature estimation unit and
A temperature estimation device including. In the machine learning, the friction calorific value calculated by the friction calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit are used as input data, and the measured value of the temperature of the lubricant is used as a label for each other. Supervised learning using associated teacher data,
The temperature estimation device according to claim 1 or 2. The machine learning uses the following equation as a learning model.
The temperature estimation device according to any one of claims 1 to 3. In Equation 1, T is the estimated temperature of the lubricant, T ₀ is the room temperature, subscript i is the axis number that affects the lubricant temperature of the estimated target axis including the estimated target axis, and D ₁ to D ₆ are in advance. The coefficient identified by executing various patterns of operations through experiments, etc., and acquiring data on the grease temperature, room temperature, calorific value, moving speed of the speed reducer, and calorific value of other heat sources in each pattern, E _i is the motor . The calorific value W _1i is the coefficient, _Fi is the frictional calorific value W _2i coefficient, W ₃ is the air-cooled heat radiation amount, and W ₄ is the calorific value of other heat sources.

The machine learning uses a linear model as a learning model.
The temperature estimation device according to any one of claims 1 to 3. The machine learning uses a neural network as a learning model.
The temperature estimation device according to claim 3. When performing the machine learning, learning that considers the regularization term is performed.
The temperature estimation device according to any one of claims 1 to 6. When performing the machine learning, learning using L2 regularization or sparse regularization is performed.
The temperature estimation device according to claim 7 .
A life evaluation device including a life estimation unit that estimates the life of the lubricant based on the temperature of the lubricant estimated by the lubricant temperature estimation unit according to any one of claims 2 to 8. A life evaluation device for evaluating the life of the lubricant in a machine including a motor and a transmission mechanism lubricated by the lubricant and transmitting the power of the motor to a moving portion.
A motor calorific value calculation unit that calculates the motor calorific value based on the current value of the motor, and a motor calorific value calculation unit.
A friction calorific value calculation unit that calculates the friction calorific value in the transmission mechanism based on the rotation speed of the motor and the friction coefficient of the transmission mechanism.
A learning model storage unit that stores a learned model generated by machine learning using the friction calorific value calculated by the friction calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit.
Lubrication that estimates the temperature of the lubricant of the motor using the trained model based on the frictional calorific value calculated by the frictional calorific value calculation unit and the motor calorific value calculated by the motor calorific value calculation unit. Material temperature estimation unit and
A life estimation unit that estimates the life of the lubricant based on the temperature of the lubricant estimated by the lubricant temperature estimation unit, and a life estimation unit.
Life evaluation device equipped with.

Images