Thermal displacement model learning device, thermal displacement estimation device, processing system, and processing method

Abstract

A thermal displacement model learning device (2) is characterized by being provided with: a learning data acquisition unit (21) that acquires learning data that associates, by time, input data that includes at least temperature data indicating a time-series temperature of a machine tool, and thermal displacement data that indicates a time-series temperature of the machine tool; the thermal displacement data represents the thermal displacement amount of a time sequence of the machine tool; a data set generation unit (22) that generates a data set obtained by extracting a portion of the learning data from the learning data; and a learning unit (23) that generates, using the data set, a learned thermal displacement model for estimating a thermal displacement amount from input data of the machine tool, and the data set generation unit (22) generates the data set such that the lengths of time of the extracted time intervals are different from each other in at least two of the plurality of data sets.

Description

Technical Field

[0001] The present disclosure relates to a thermal displacement model learning device, a thermal displacement estimation device, a machining system and a machining method for correcting the thermal displacement of a machine tool. Background Art

[0002] A machine tool is a processing device that drives a feed axis to change the relative position between the object to be processed and the tool, thereby removing the object to be processed, called a workpiece, and processing it into a desired shape. Machine tools represented by milling machines and lathes install the tool or the object to be processed on the spindle, and perform processing while rotating the spindle. The machine tool is equipped with an actuator for driving the feed axis and the spindle, and a servo amplifier that generates power for driving the actuator. A machine tool equipped with an actuator measures the position of the feed axis through a position detector, and performs positioning by controlling so that the actual position matches the commanded position.

[0003] When a machine tool processes a workpiece, heat is generated in the machine tool. For example, actuators and servo amplifiers generate heat when converting electricity into power. In addition, in processing such as cutting and grinding, heat is generated along with deformation of the material, friction, etc. In addition, the friction generated when the feed axis is moved also generates heat. When this heat is transferred to the structure of the machine tool represented by the column and the bed, the temperature of the mechanical structure of the machine tool rises and thermal expansion occurs. This deformation of the structure caused by thermal expansion is called thermal displacement. In addition, there are also cases where the temperature of the structure changes unevenly. In this case, different thermal displacements are generated depending on the position in the structure, which sometimes causes displacements other than simple expansion and contraction of the structure, such as tilting, skewing, and twisting. It is known that this thermal displacement will reduce the positioning accuracy of the machine tool, so sometimes there will be errors in the relative position of the tool and the workpiece, and the processing accuracy will decrease.

[0004] Patent document 1 discloses a technology that can estimate the thermal displacement amount from the measured data by machine learning using the measured data and the measured values ​​of the thermal displacement amount obtained during the machining of the machine tool. Specifically, the machine learning device disclosed in Patent document 1 includes: a measured data acquisition unit that acquires a measured data group; a thermal displacement amount acquisition unit that acquires the measured values ​​of the thermal displacement amount; a storage unit that associates the measured data group as input data and the measured values ​​of the thermal displacement amount as labels and stores them as training data; and a calculation formula learning unit that performs machine learning using the training data and sets a thermal displacement amount prediction calculation formula that calculates the thermal displacement amount based on the measured data group. The storage unit stores the measured data group as training data for a specified period of time.

[0005] Prior art literature

[0006] Patent Literature

[0007] Patent Document 1: Japanese Patent Application Publication No. 2019-111648 Summary of the invention

[0008] Problem that the invention aims to solve

[0009] However, according to the above-mentioned prior art, there is the following problem: when the period for storing training data in the storage unit is inappropriate for the thermal displacement generated in the machine tool, the estimation accuracy of the thermal displacement amount generated in the machine tool may be reduced. For example, the time before the heat generated by a certain heat source causes thermal displacement is affected by the positional relationship with other heat sources, the surrounding environment, etc. Therefore, when the storage period is too short, during learning, sometimes the measurement data that affects the thermal displacement is not included in the stored training data, and the estimation accuracy of the thermal displacement amount is reduced. In addition, when the storage period is too long, during learning, due to erroneous learning of the measurement data included in the data and which contributes little to the thermal displacement, the estimation accuracy of the thermal displacement amount is sometimes reduced.

[0010] The present disclosure has been made in view of the above circumstances, and an object of the present disclosure is to obtain a thermal displacement model learning device capable of improving the estimation accuracy of the thermal displacement amount.

[0011] Means used to solve problems

[0012] In order to solve the above-mentioned problems and achieve the purpose, the thermal displacement model learning device disclosed in the present invention is characterized in that it comprises: a learning data acquisition unit, which acquires learning data, wherein the learning data establishes a correspondence between input data and thermal displacement data in time, wherein the input data at least includes temperature data representing the temperature of a time series of a machine tool, and the thermal displacement data represents the thermal displacement amount of the time series of the machine tool; a data set generation unit, which generates a data set obtained by extracting a part of the learning data from the learning data; and a learning unit, which uses the data set to generate a learned thermal displacement model for estimating the thermal displacement amount based on the input data of the machine tool, wherein the data set generation unit generates the data set so that in at least two of the multiple data sets, the time lengths of the extracted time intervals are different from each other.

[0013] Effects of the Invention

[0014] According to the present disclosure, there is an effect that the estimation accuracy of the thermal displacement amount can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a diagram showing the structure of the processing system according to the first embodiment.

[0016] Figure 2 This is a diagram showing the functional configuration of the machine tool according to the first embodiment.

[0017] Figure 3This is a diagram showing an example of installation positions of temperature sensors in the machine tool according to the first embodiment.

[0018] Figure 4 This is an explanatory diagram of an example of changes in the operating state of a machine tool.

[0019] Figure 5 This is a diagram showing the functional structure of the thermal displacement model learning device according to the first embodiment.

[0020] Figure 6 This is a diagram for explaining a first example of a method for generating a data set.

[0021] Figure 7 This is a diagram for explaining a second example of the method for generating a data set.

[0022] Figure 8 This is a diagram for explaining the third example of the method for generating a data set.

[0023] Fig. 9 This is a diagram for explaining the learning method of the thermal displacement model.

[0024] Fig.10 This is an illustration of a neural network.

[0025] Fig.11 It is used to illustrate Figure 5 FIG. 1 is a flowchart showing the operation of the thermal displacement model learning device.

[0026] Fig.12 This is a diagram showing the structure of a processing system according to the second embodiment.

[0027] Fig.13 This is a diagram showing the functional configuration of a machine tool according to the second embodiment.

[0028] Fig.14 It is a diagram showing the functional structure of the thermal displacement estimation device according to the second embodiment.

[0029] Fig.15 It is used to illustrate Fig.12 Flow chart of the operation of the machining system shown.

[0030] Fig.16 This is a diagram showing an example of a hardware configuration.

[0031] Fig.17 This is a diagram showing the functional structure of a thermal displacement model learning device according to a modified example of the second embodiment.

[0032] Fig.18 This is a diagram showing the structure of a processing system according to a third embodiment.

[0033] Fig.19 This is a diagram showing the structure of a processing system according to a fourth embodiment. DETAILED DESCRIPTION

[0034] Hereinafter, a thermal displacement model learning device, a thermal displacement estimation device, a machining system, and a machining method according to an embodiment of the present disclosure will be described in detail based on the drawings.

[0035] Implementation method 1.

[0036] Figure 1 1 is a diagram showing a structure of a machining system 100-1 according to Embodiment 1. The machining system 100-1 includes a machine tool 1 and a thermal displacement model learning device 2. The machine tool 1 is a machining device that performs removal machining on a machining object using a tool. The thermal displacement model learning device 2 learns a thermal displacement model for estimating a thermal displacement amount from input data based on thermal displacement data, temperature data, and machine status data acquired from the machine tool 1.

[0037] Figure 2 1 is a diagram showing the functional structure of the machine tool 1 according to the first embodiment. The machine tool 1 includes: a feed shaft 11 provided with a tool T for processing a workpiece W, so that the relative position between the tool T and the workpiece W is changed; a feed shaft motor 12 as an actuator for driving the feed shaft 11; a spindle 13 for rotating the tool T or the workpiece W; a spindle motor 14 as an actuator for driving the spindle 13; a control device 15 for controlling the machine tool 1; a peripheral device 16 as a cooling device for suppressing heat generation when the actuator is driven; a temperature sensor TS for measuring the temperature of the machine tool 1; and a displacement sensor DS for measuring the relative position between the tool T and the workpiece W.

[0038] The displacement sensor DS is provided on the machine tool 1 so as to detect the relative position between the tool T and the workpiece W. The displacement sensor DS measures the relative position and outputs the measured relative position as thermal displacement data to the thermal displacement model learning device 2. The displacement sensor DS may be a sensor capable of detecting displacement in at least one axial direction, for example, a laser sensor using a triangulation method or a laser sensor using a TOF (Time Of Flight) method. In addition, a structure in which a plurality of orthogonal displacement sensors are provided to detect displacement in a plurality of axial directions may be adopted.

[0039] The temperature sensor TS is provided in the machine tool 1. The temperature sensor TS detects temperature and outputs the detected temperature as temperature data to the thermal displacement model learning device 2. Figure 2 Although one temperature sensor TS is shown in the figure, it is preferred that the machining system 100-1 has a plurality of temperature sensors TS. It is preferred that the temperature sensor TS is installed at a location where a large temperature change is estimated or a location where a large influence on thermal displacement is estimated. In addition, it is preferred that a temperature sensor TS is also installed outside the machine tool 1 so that the temperature sensor TS can measure the temperature of the space where the machine tool 1 is installed.

[0040] Figure 3 This is a diagram showing an example of the installation position of the temperature sensor TS in the machine tool 1 according to the first embodiment. Figure 3 The machine tool 1 shown is a milling machine type processing device, and a Y-axis feed axis motor 12 and a table TA are provided on a bed 17 , and a workpiece W is placed on the table TA. In addition, the machine tool 1 is provided with X-axis and Z-axis feed axis motors 12 , a spindle 13 , and a spindle motor 14 on a column 18 . Figure 3 The black circle in FIG. 1 shows an example of the location where the temperature sensor TS is installed. As described above, the location where the temperature sensor TS is installed can be a location where the temperature change is estimated to be large, and a location where the influence of thermal displacement is estimated to be large. As an example of a location where the temperature change is estimated to be large, a drive mechanism such as the feed axis motor 12 for driving the X, Y, and Z axes and the spindle motor 14 for rotating the spindle 13 can be cited. As an example of a location where the influence of thermal displacement is estimated to be large, a structure of the machine tool 1 such as the bed 17, the column 18, and the feed screw 19 can be cited. The temperature sensor TS can be installed at the vertex of the structure, a point between multiple vertices, and the inside, etc.

[0041] return Figure 2 The peripheral device 16 is a cooling device, which is provided to cool the machine tool 1 and prevent thermal displacement. The peripheral device 16 performs heat exchange between the structure of the machine tool 1, the workpiece W, and the tool T and the external air outside the machine tool 1. For example, the peripheral device 16 includes a cooling device that injects cutting oil for cooling the workpiece W, a pump that circulates a refrigerant that cools a heat exchange pipe provided in the machine tool 1, and the like. The peripheral device 16 outputs the operation information of the peripheral device 16 to the control device 15.

[0042] The operation information of the peripheral device 16 at least includes an operation status indicating whether the peripheral device 16 is in operation. In addition to the operation status, the operation information of the peripheral device 16 may also include power consumption, power of the peripheral device 16, etc. Here, the operation status included in the operation information of the peripheral device 16 can be represented by a numerical value. For example, the operation status "1" indicates "the state in which the peripheral device 16 is in operation", and the operation status "0" indicates "the state in which the peripheral device 16 is stopped".

[0043] The machining program is a series of command statements issued to the control device 15 in order to make the feed axis 11, the spindle 13 and the peripheral equipment 16 of the machine tool 1 perform the desired actions. Specifically, the machining program can include information indicating the position command of the feed axis 11, the speed command of the feed axis 11, the rotation speed command of the spindle 13, the operation and stop command of the peripheral equipment 16, etc.

[0044] The control device 15 reads the machining program and analyzes the instructions written in the machining program. In addition, the control device 15 generates a command position and a command speed for the feed axis 11 so that the feed axis 11 performs a desired action, and gives an operation amount to the feed axis motor 12. In addition, the control device 15 generates a command speed for the spindle 13 so that the spindle 13 performs a desired action, and gives an operation amount to the spindle motor 14.

[0045] Furthermore, the control device 15 outputs machine state data indicating the operating state of the machine tool 1 to the thermal displacement model learning device 2. The machine state data is data indicating the operating state of the machine tool 1, for example, indicating the working state of the peripheral device 16 for adjusting the temperature of the machine tool 1 and the operating state of the actuator provided in the machine tool 1. Specifically, the machine state data includes at least one or more of the command position for the feed axis 11, the command speed for the feed axis 11, the command speed for the spindle 13, and the operating information of the peripheral device 16.

[0046] In addition, when a position detector is provided on the machine tool 1, the control device 15 can perform feedback control so that the difference between the actual position detected by the position detector and the command position, or the difference between the actual speed and the command speed is 0. In this case, the machine state data can include information of one or more of the actual position of the feed axis 11, the actual speed of the feed axis 11, and the actual speed of the spindle 13.

[0047] Here, the heat generated inside and outside the machine tool 1 and the heat exchange are explained. Part of the electric energy provided by the control device 15 when driving the feed shaft motor 12 and the spindle motor 14 becomes loss heat. In addition, when the feed shaft 11 and the spindle 13 are in motion, loss heat is generated due to mechanical friction. This loss heat has the effect of increasing the temperature of the structure of the machine tool 1, but is also cooled by the peripheral equipment 16. In addition, processing heat is generated when processing between the tool T and the workpiece W. In addition, when there is a temperature difference between the main body of the machine tool 1 and the surrounding environment, the temperature of the structure of the machine tool 1 changes over time according to the temperature difference. The heat generation, heat dissipation, and forced cooling generated inside and outside the above-mentioned machine tool 1 cause the temperature change of the structure, so that the structure produces thermal expansion and thermal strain. As a result, the thermal change of this structure becomes the processing error between the tool T and the workpiece W. In the present disclosure, the processing error between the tool T and the workpiece W caused by the thermal change of the machine tool 1 is called thermal displacement.

[0048] Next, the operating status of the machine tool will be described. Figure 4 This is an explanatory diagram of an example of changes in the operating state of machine tool 1 . Figure 4 The diagram shows the change in the operating state of the machine tool 1 from when the power is turned on and the machine tool 1 is operated according to the machining program to when the power is turned off.

[0049] Figure 4 An example of the temporal change of the operating state and machining program execution state of each of the main spindle 13 , the feed axis 11 , and the peripheral device 16 , which are components of the machine tool 1 , is shown. Figure 4 The horizontal axis represents time, and t0 to t9 represent moments. The details from moment t0 to moment t9 will be described later. The machining program execution state shows whether the machine tool 1 is performing a program operation or is stopped. Program operation means that the machine tool 1 moves according to the instructions recorded in the machining program. The machine tool 1 performs a program operation for a certain machining, and when the program operation is completed, the operator performs operations such as replacing the workpiece W. After that, the operation is performed according to the process of performing a program operation for the next machining.

[0050] exist Figure 4 In the example shown, the power of the machine tool 1 is turned on at time t0, and the power of the machine tool 1 is turned off at time t9. The period from time t0 to time t1 represents the standby state before the operation based on the machining program is started. In addition, the period from time t8 to time t9 represents the period from the end of the operation based on the machining program to the power of the machine tool 1 being turned off.

[0051] In addition, as shown in the machining program execution status, the program operation based on machining program #1 is performed between time t1 and time t5, and the program operation based on machining program #2 is performed between time t6 and time t8. When these machining programs are executed, the spindle speed as the rotation speed of the spindle 13, the feed speed as the speed of the feed axis 11, and the operating status of the peripheral device 16 change with time.

[0052] Figure 4 The "spindle speed" in FIG. 1 represents the rotation speed of the spindle 13. Here, Figure 4 In the embodiment, the state in which the main shaft 13 rotates at 1000 revolutions per minute is represented as "S1000". Figure 4 As shown in the spindle speed, the spindle motor 14 rotates at 1000 rpm between time t1 and time t4, rotates at 3000 rpm between time t4 and time t5, and rotates at 3000 rpm between time t6 and time t8. When the spindle 13 rotates, heat is generated from the spindle motor 14 and the structure of the spindle 13. The higher the spindle speed, the greater the heat generated by the spindle 13. Since the spindle speed varies according to the command of the machining program, the heat generated by the spindle 13 also varies according to the machining program.

[0053] Figure 4 The "feed speed" in FIG. 1 shows the moving speed of the feed shaft 11. Here, Figure 4 In the example, the state where the feed axis 11 moves 100 mm per minute is represented as "F100". Figure 4As shown in the feed speed, the feed shaft motor 12 rotates at a speed of 100 mm per minute between time t1 and time t3, rotates at a speed of 300 mm per minute between time t3 and time t5, and rotates at a speed of 300 mm per minute between time t6 and time t8. When the feed shaft 11 moves, the feed shaft motor 12 and the structure of the feed shaft 11 generate heat. The faster the feed speed, the greater the heat generation. Since the feed speed changes according to the command of the machining program, the heat generation of the feed shaft 11 also changes according to the machining program.

[0054] Figure 4 The "operating state of peripheral equipment" indicates whether the peripheral equipment 16 for cooling the machine tool 1 is in operation or in a stopped state. The peripheral equipment 16 does not always operate during machining, but operates or stops according to the commands recorded in the machining program. Figure 4 In the example of the operating state of the peripheral device 16, the peripheral device 16 operates between time t2 and time t5, stops between time t5 and time t7, and operates between time t7 and time t8. When the peripheral device 16 operates, heat moves from the machine tool 1 to the external environment of the machine tool 1, and the machine tool 1 is cooled. Since the operating state of the peripheral device 16 changes according to the command of the machining program, the state based on the cooling of the peripheral device 16 also changes according to the machining program.

[0055] If used Figure 4 As described above, since the heat generated in the feed shaft 11 and the main shaft 13 and the cooling state of the peripheral device 16 are not constant, the thermal displacement amount of the machine tool 1 also changes with time according to the machining program.

[0056] In addition, there is a time when the operation based on the machining program is not performed, such as between time t5 and time t6. Between time t5 and time t6, for example, when the operator is disassembling or installing the workpiece W or cleaning the inside of the machine tool 1, the machine tool 1 may stop for a production change adjustment operation. When the machine tool 1 stops in this way, the operation conditions of the spindle 13, the feed axis 11, and the peripheral equipment 16 are different from those during the program operation, so the generated thermal displacement also shows a tendency different from that during the program operation.

[0057] As described above, the speed of the feed axis 11 and the spindle 13 and the operating state of the peripheral device 16 affect the generation of thermal displacement. In this embodiment, each section with different operating states of the machine tool 1, such as the section from time t0 to time t1 and the section from time t1 to time t2, is defined as an operation unit. The time length of the operation unit varies according to the timing of the command recorded in the machining program, and therefore is a variable length.

[0058] In addition, although Figure 4 2 shows an example in which the machine tool 1 executes two different machining programs, but it is sufficient that the machining program is executed at least once, and the type of machining program and the number of executions are not limited to this example.

[0059] In addition, the machine tool 1 can also be powered on again after power is cut off to run the machining program. In this case, the time interval from power cut off to power on again is defined as one operation action unit.

[0060] Next, the functional structure of the thermal displacement model learning device 2 will be described. Figure 5 1 is a diagram showing a functional configuration of thermal displacement model learning device 2 according to Embodiment 1. Thermal displacement model learning device 2 uses data output by machine tool 1 during operation to learn a thermal displacement model for estimating a thermal displacement amount from input data including at least temperature data.

[0061] The thermal displacement model learning device 2 includes a learning data acquisition unit 21 , a data set generation unit 22 , a learning unit 23 , and a model storage unit 24 .

[0062] The learning data acquisition unit 21 acquires the thermal displacement data, temperature data and machine status data output by the machine tool 1, and generates learning data that is synchronized in time with the acquired data. Here, the temperature data and the machine status data are examples of information included in the input data. The input data at least includes the temperature data. The learning data is time series data in which the thermal displacement data, temperature data and machine status data are arranged in a constant time period. Here, the constant time period represents a sampling period pre-set in the learning data acquisition unit 21. The learning data acquisition unit 21 receives data of a time interval corresponding to an operating action unit of at least one interval or more from the machine tool 1, and generates learning data. The learning data acquisition unit 21 outputs the generated learning data to the data set generation unit 22.

[0063] The data set generation unit 22 detects a specific time based on the operating state of the machine tool 1, extracts thermal displacement data, temperature data, and machine state data from the learning data based on the detected time, and aggregates the extracted data into a set of data sets, and outputs the multiple sets of data sets to the learning unit 23. In the following, the specific time detected by the data set generation unit 22 is referred to as the extraction reference time. Figure 4 The extraction reference time is described. The data set generation unit 22 detects each of the times t1 to t8 as the time when the operating state of the machine tool 1 has changed. Figure 4In the example, the extraction reference time is set to the time when one of the spindle speed, feed speed, working state of the peripheral device 16 and the start and end of the machining program changes, but the extraction reference time is not limited to these examples. For example, the timing of calling and ending a subroutine, the timing of changing a tool T, etc. can also be regarded as a change in the operating state of the machine tool 1. In addition, when the machining program indicates the timing of the change of the operating state in advance, the data set generation unit 22 can read this information. In addition, the data set generation unit 22 can add and set a new extraction reference time by inserting it between two extraction reference times in time.

[0064] The data set generating unit 22 generates a data set by extracting a part of the learning data from the learning data based on the extraction reference time.

[0065] Figure 6 This is a diagram illustrating the first example of a method for generating a data set. Figure 6 , a first example of a method in which the dataset generating unit 22 generates a dataset based on learning data is described. Figure 6 The data sets #a1 to #an extracted from the learning data are shown when the extraction reference time t1 is the initial reference time of data extraction. The input data #a1 to #an are temperature data and machine status data with the extraction reference time t1 as the initial time. When the i-th input data is set as input data #ai, the time interval of input data #ai is extended backward in time by a predetermined time extension width T1 from the end of the time interval of input data #ai-1. The time length of input data #a2 is longer than the time length of input data #a1 by the time extension width T1. In addition, the time length of input data #a3 is longer than the time length of input data #a2 by the time extension width T1. That is, when the time width of input data #a1 is set to T, the time width of input data #ai is "T+(i-1)T1". Input data #a1 to #an are data of time intervals of each time width "T+(i-1)t1" starting from the extraction reference time t1. The training data #a1~#an are the thermal displacement data at the end time of each of the input data #a1~#an. Figure 6 As shown, the data set generation unit 22 generates a plurality of data sets #a1 to #an from one extraction reference time t1. Through the above processing, the data set generation unit 22 can generate data sets so that the time lengths of the extracted time intervals of the plurality of data sets #a1 to #an generated from one extraction reference time t1 are different. The data set generation unit 22 performs the above processing for each of the plurality of extraction reference times included in the learning data. In addition, as in the case of data set #an, the end time may exceed t2 which is the next extraction reference time of the extraction reference time t1.

[0066] Figure 7 This is a diagram illustrating a second example of a method for generating a data set. Figure 7 A second example of the method in which the data set generating unit 22 generates a data set from the learning data will be described. Figure 7 The data set #b1 to #bn extracted from the learning data is shown when the extraction reference time t2 is the end reference time of data extraction. The input data #b1 to #bn are temperature data and machine status data with the extraction reference time t2 as the end time. When the i-th input data is set as input data #bi, the time interval of the input data #bi is extended backward in time by a predetermined time extension width T1 from the beginning of the time interval of the input data #bi-1. Similar to the first example, when the time width of the input data #b1 is set to T, the time width of the input data #bi is "T+(i-1)T1". The input data #b1 to #bn are data of time intervals of each time width "T+(i-1)T1" with the extraction reference time t2 as the end. The training data #b1 to #bn are the thermal displacement data at the end time of each input data #b1 to #bn. As shown Figure 7 As shown, the data set generation unit 22 generates a plurality of data sets #b1 to #bn from one extraction reference time t2. Through the above processing, the data set generation unit 22 can generate data sets so that the time lengths of the extracted time intervals of the plurality of data sets #b1 to #bn generated from one extraction reference time t2 are different. The data set generation unit 22 performs the above processing on the plurality of extraction reference times included in the learning data. In addition, as shown in data set #bn, the initial time can be traced back to a time earlier than t1, which is the extraction reference time before the extraction reference time t2.

[0067] Figure 8 : is an explanatory diagram of a third example of a method for generating a data set. In the third example, the data set generating unit 22 generates a data set so that the set of data included in all data sets uniformly includes the entirety of the acquired data. Specifically, the data set generating unit 22 generates a data set by dividing the period from the extraction reference time t1 to the extraction reference time t2 next to the extraction reference time t1 into n equal intervals. n is a natural number. Figure 8The data sets #c1 to #cn extracted from the learning data between the extraction reference time t1 and the final extraction reference time t2 are shown when the extraction reference time t1 is the initial reference time of data extraction. The input data #c1 is the temperature data and the machine status data with the extraction reference time t1 as the initial time and the time t1+(t2-t1) / n as the final time. In addition, the input data #c2 is the temperature data and the machine status data with the time t1+(t2-t1) / n as the initial time and the time t1+2×(t2-t1) / n as the final time. The input data #ci is the temperature data and the machine status data with the time t1+(i-1)×(t2-t1) / n as the initial time and the time t1+i×(t2-t1) / n as the final time. The training data #c1 to #cn are the thermal displacement data at the final time of each of the input data #c1 to #cn. Through the above processing, the dataset generation unit 22 can generate a dataset in a manner that there is no duplication between the extraction reference time t1 and the time t2 and that the dataset is uniformly included in the input data. The dataset generation unit 22 performs the above processing on each of the multiple extraction reference times included in the learning data. In addition, in the third example, for the datasets #c1 to #cn generated between the extraction reference times t1 and t2, the time length of the extracted time interval is uniform. However, if Figure 4 As shown in FIG. 1 , the time lengths between adjacent extraction reference times are not uniform, and the time length of the time interval extracted when the data set generation unit 22 generates a data set is determined by the time length between two adjacent extraction reference times. Therefore, by performing the above-mentioned processing on all extraction reference times, as a whole, a plurality of data sets with different time lengths of the extracted time intervals are generated.

[0068] In the first to third examples of the data set generation method, the training data #a1 to #an, #b1 to #bn, #c1 to #cn are set as thermal displacement data at a time corresponding to the end time of the input data #a1 to #an, #b1 to #bn, #c1 to #cn, but the time length of the training data is not limited to a time, and it can also be time series data. For example, the training data #a1 to #an, #b1 to #bn, #c1 to #cn can be time series data having the same time length as the input data #a1 to #an, #b1 to #bn, #c1 to #cn.

[0069] In addition, in the first example of the above-mentioned data set generation method, from input data #a1 to input data #an, time interval #a1 corresponding to input data #a1 is selected n times and adopted as input data. In contrast, the number of times extended interval #a2 is adopted as input data is (n-1) times, and the number of times extended interval #an is adopted as input data is only 1 time. Similarly, in the second example, time interval #b1 corresponding to input data #b1 is selected n times from input data #b1 to input data #bn and adopted as input data. In contrast, the number of times extended interval #b2 is adopted as input data is (n-1) times, and the number of times extended interval #bn is adopted as input data is only 1 time. That is, among the multiple input data generated with one extraction reference time as the time reference, there are intervals with a large number of extraction times and intervals with a small number of extraction times. When generating a data set from each extraction reference time, the data set generation unit 22 evens out the number of extraction times of each time interval and extended interval over the entire learning data. Specifically, when combining the groups of all data sets generated based on each extraction reference time, when the deviation of the number of extractions in the time interval is above the allowable value, the data set generation unit 22 performs histogram equalization processing. Here, the histogram equalization processing can adopt a well-known algorithm. Through this processing, the data set generation unit 22 can equalize the deviation of the number of data extractions throughout the entire learning data. As a result, the entire group of data sets generated by the data set generation unit 22 will not extract only specific intervals from the learning data, and multiple data sets that evenly reflect the operating status contained in the learning data can be generated. In addition, although the equalization of the number of extractions of the input data is described here, the number of extractions can also be equalized with a focus on the training data.

[0070] As described above, in the first to third examples of the data set generation method, based on the extraction reference time when the operating state of the machine tool 1 changes, multiple data sets with different time lengths of the extracted time interval can be generated. In addition, in the third example, the data set is generated so that the set of data included in all data sets uniformly includes all acquired data. As a result, when the learning unit 23 described below performs learning of the thermal displacement model, the entirety of the learning data can be uniformly set as the object of learning, thereby achieving the following effects: suppressing over-learning in which the model is consistent with only a part of the data, and improving the generalization performance of the thermal displacement model.

[0071] In addition, in the third example of the method for generating the above-mentioned data set, an example is given of dividing the time interval between two adjacent extraction reference times into n equal intervals, but dividing the time interval between two adjacent extraction reference times according to non-constant time lengths in a manner that the input data is not repeated also has the same effect as the third example.

[0072] Next, the learning unit 23 will be described. Fig. 9 : is an explanatory diagram of a learning method of a thermal displacement model. The learning unit 23 has the following function: using the input data and training data contained in the data set generated by the data set generation unit 22, the learning unit 23 has a thermal displacement model for estimating the thermal displacement amount based on the input data. The learning unit 23 has a thermal displacement model inside, and when the input data contained in the data set is input in a time series, the thermal displacement model outputs the thermal displacement amount in a time series. The learning unit 23 optimizes the internal parameters of the thermal displacement model so that the difference between the estimated value of the thermal displacement amount as the output of the thermal displacement model and the training data contained in the data set is minimized. Thus, the learning unit 23 determines the thermal displacement model that can estimate the thermal displacement amount based on the input data, and outputs it as a learned thermal displacement model.

[0073] In the thermal displacement model learning device 2, it is characterized in that the data set output by the data set generation unit 22 has a variable length. That is, in at least two of the multiple data sets output by the data set generation unit 22, the time lengths of the extracted time intervals are different from each other. Therefore, the learning unit 23 uses a thermal displacement model corresponding to the input data of variable length. The learning unit 23 learns the thermal displacement amount through so-called supervised learning, for example, according to a neural network. Here, supervised learning refers to a method of learning features in the learning data by providing a group of input and result data, i.e., learning data, to a learning device and inferring the result based on the input. Here, an example of the learning unit 23 using RNN (Recurrent Neural Network) to learn a thermal displacement model is described.

[0074] RNN is a neural network having a structure in which the output of an intermediate layer is input to a layer that recursively outputs the result. The learning unit 23 can use, for example, Fig.10 The structure of the neural network is shown. Fig.10 This is an illustration of a neural network. Fig.10 The neural network shown is an RNN whose input layer consists of n inputs from x1 to xn, whose middle layer consists of m nodes from h1 to hm, and whose output layer consists of one output from y1. Here, Fig.10 The upper figure shows the state of the neural network at time t-1, Fig.10 The lower figure shows the state of the neural network at time t. The intermediate layer at time t outputs a weighted sum to the output layer, which is a weighted sum of the value calculated based on the input of the input layer at time t and the value calculated based on the input from the intermediate layer at time t-1. Here, if the vector formed by combining the input layers x1 to xn at time t is set to x t , let the vector formed by combining the intermediate layers h1 to hm at time t be h t , then h tIt can be expressed by the following mathematical formula (1). In addition, in mathematical formula (1), if the vector x t The dimension is set to n and the vector h t The dimension of is set to m, then W becomes an n×m dimensional linear transformation matrix. In addition, R represents an M×M dimensional linear transformation matrix, b represents a bias vector, and g represents an activation function.

[0075] h t =g(Wx t +Rh t-1 +b)…(1)

[0076] The learning unit 23 uses a backpropagation through time method to adjust the weight of the neural network so that the thermal displacement amount as an output matches the thermal displacement data.

[0077] In addition, although Fig.10 , an example of a neural network with an input layer of n dimensions, an intermediate layer of m dimensions, and an output layer of 1 dimension is shown, but the dimensions of each layer are not limited to this example and can be any dimension. Fig.10 An example of a neural network in which the intermediate layer is one layer is shown, but the number of the intermediate layers may be plural.

[0078] Furthermore, in order to improve the generalization performance of the heat displacement model, the learning unit 23 may use methods such as “dropout” for randomly excluding neurons during learning and “early stopping” for stopping learning as early as possible by monitoring errors.

[0079] In addition, although the example of using the Elman network in the RNN is described above, the thermal displacement model can be constructed and learned using LSTM (Long Short Term Memory) or GRU (Gated Recurrent Unit), which are other well-known RNNs. As another method, the learning unit 23 can learn by regression analysis using an autoregressive model.

[0080] Fig.11 It is used to illustrate Figure 5 The thermal displacement model learning device 2 performs the following operations: Fig.11 The processing shown can generate a learned thermal displacement model.

[0081] The learning data acquisition unit 21 of the thermal displacement model learning device 2 acquires thermal displacement data, temperature data, and machine state data from the machine tool 1 (step S101). The learning data acquisition unit 21 associates the acquired thermal displacement data, temperature data, and machine state data in time and generates learning data (step S102).

[0082] The data set generation unit 22 generates a plurality of data sets based on the learning data (step S103). The learning unit 23 uses the data sets to optimize the internal parameters of the thermal displacement model so that when the input data included in the data set is input, the output thermal displacement amount is consistent with the training data included in the data set, thereby learning the thermal displacement model (step S104).

[0083] As described above, according to the first embodiment, the data set generation unit 22 generates a plurality of data sets of variable lengths, and in at least two of the plurality of data sets, the data sets of which the time lengths of the extracted time intervals are different from each other are used to learn the thermal displacement model. Thus, the thermal displacement model learning device 2 can use the data set generated using the time intervals of appropriate time lengths for the thermal displacement generated in the machine tool 1 to learn the thermal displacement model, and can improve the estimation accuracy of the thermal displacement amount. Here, the thermal displacement model learning device 2 does not depend on the operation duration of the machine tool 1, and can improve the estimation accuracy of the thermal displacement amount.

[0084] The machine tool 1 operates according to machining programs with different operation modes based on the variable length of the machining program time according to the finished shapes of various workpieces W. In order to generate a thermal displacement model capable of accurately estimating the thermal displacement amount, it is necessary to obtain a data set for learning based on the machining program and the time length is variable.

[0085] In the first embodiment, the data set used by the learning unit 23 to learn the thermal displacement model is generated based on the time when the operating state of the machine tool 1 running based on the machining program changes. Therefore, the thermal displacement model learning device 2 in the first embodiment can perform the learning of the thermal displacement model using the data set with the change of the operating state of the machine tool 1 in actual machining as the time reference. Therefore, the thermal displacement model can learn the input-output relationship reflecting the operating state of the machine tool.

[0086] When using the first and second examples of the dataset generation method, the dataset generation unit 22 can generate a dataset by extracting data having a time width that is a multiple of a predetermined time extension width T1 from a time defined as an extraction reference time from the learning data.

[0087] Furthermore, since the thermal displacement model is a temporally recursive model, in addition to the input data input at the time when the estimated value of the thermal displacement amount is calculated, learning can be performed in a manner that calculates the thermal displacement amount affected by the information included in the past input data at that time. Therefore, the thermal displacement model learning device 2 can perform learning by including not only the temperature data and the machine state data at the moment when the thermal displacement is calculated, but also the influence of the temporal changes of these data.

[0088] Implementation method 2.

[0089] Fig.12 1 is a diagram showing the structure of a machining system 100-2 according to Embodiment 2. The machining system 100-2 according to Embodiment 2 includes a machine tool 1A, a thermal displacement model learning device 2, and a thermal displacement estimation device 3. In addition to the structure of the machining system 100-1 according to Embodiment 1, the machining system 100-2 further includes the thermal displacement estimation device 3, and includes the machine tool 1A instead of the machine tool 1 of the machining system 100-1.

[0090] The processing performed by the thermal displacement model learning device 2 is the same as that in the first embodiment except that the learned thermal displacement model is output to the thermal displacement estimation device 3, and therefore the description thereof is omitted here. The thermal displacement estimation device 3 obtains the learned thermal displacement model generated by the thermal displacement model learning device 2 and input data for inference including at least temperature data indicating the time series temperature of the machine tool 1A to be inferred. The thermal displacement estimation device 3 generates an estimated value of the thermal displacement amount of the machine tool 1A to be inferred based on the input data using the thermal displacement model. The thermal displacement estimation device 3 outputs the estimated value of the thermal displacement amount to the machine tool 1A.

[0091] In addition to the functions of the machine tool 1 , the machine tool 1A has a function of outputting temperature data and machine state data to the thermal displacement estimation device 3 and a function of correcting thermal displacement based on an estimated value of the thermal displacement amount output by the thermal displacement estimation device 3 .

[0092] Fig.13 1 is a diagram showing a functional structure of a machine tool 1A according to Embodiment 2. The structure of the machine tool 1A is the same as that of the machine tool 1, except that a control device 15A is provided instead of the control device 15 of the machine tool 1. The control device 15A corrects the operation amount of at least one of the spindle motor 14 driving the spindle 13 and the feed axis motor 12 driving the feed axis 11 based on the estimated value of the thermal displacement amount output by the thermal displacement estimation device 3, thereby correcting the thermal displacement. The machine tool 1A performs the removal process using the corrected operation amount.

[0093] Fig.14 1 is a diagram showing a functional configuration of a thermal displacement estimation device 3 according to Embodiment 2. The thermal displacement estimation device 3 includes an inference data acquisition unit 31 , a thermal displacement model acquisition unit 32 , and an inference unit 33 .

[0094] The inference data acquisition unit 31 acquires temperature data and machine state data from the machine tool 1A to be inferred, and outputs the acquired temperature data and machine state data synchronized in time as inference data to the inference unit 33. The inference data includes at least temperature data, and is generated as a set of data that is the same as the input data used when learning the learned thermal displacement model. In addition, the inference data is sequentially output at the same time period as the learning data generated by the thermal displacement model learning device 2.

[0095] The thermal displacement model acquisition unit 32 acquires the learned thermal displacement model used for inference from the model storage unit 24 of the thermal displacement model learning device 2. The thermal displacement model acquisition unit 32 outputs the acquired thermal displacement model to the inference unit 33.

[0096] The inference unit 33 uses the learned thermal displacement model to output the estimated value of the thermal displacement amount from the input data included in the inference data. The inference unit 33 outputs the estimated value of the thermal displacement amount to the machine tool 1A. The inference unit 33 inputs the machine state data included in the inference data into the learned thermal displacement model in the same data structure and time period as when learning, so that the estimated value of the thermal displacement amount can be calculated in the same time period as the input. The inference unit 33 outputs the calculated estimated value of the thermal displacement amount to the control device 15A of the machine tool 1A in the same period as the input interval of the inference data.

[0097] The control device 15A of the machine tool 1A receives the estimated value of the thermal displacement amount output by the inference unit 33, and corrects, for example, the position command of the feed axis 11 so as to cancel the thermal displacement indicated by the estimated value of the thermal displacement amount. Then, the feed axis motor 12 as the actuator of the feed axis 11 is operated using the corrected position command. The correction of the position command is updated at each input interval of the inference data.

[0098] Fig.15 It is used to illustrate Fig.12 The process of the machining system 100-2 is shown in FIG. 1 . The thermal displacement model learning device 2 of the machining system 100-2 performs thermal displacement model generation processing (step S121). In addition, the process of step S121 is equivalent to Fig.11 The processes of steps S101 to S104 are shown. These processes are performed before the processing for correcting thermal displacement is performed.

[0099] Machine tool 1A starts running the program for machining the object to be corrected for thermal displacement (step S122 ). The processing after step S123 is performed while machining the object to be corrected for thermal displacement is being performed.

[0100] The inference data acquisition unit 31 receives the temperature data and machine state data sequentially output during the machining of the machine tool 1A, and generates inference data based on the received data (step S123). The inference data has the same data structure as the input data of the thermal displacement model generated in step S121. The generated inference data is sequentially output to the inference unit 33 at the same time period as the learning data generated in the thermal displacement model learning device 2.

[0101] Inference unit 33 estimates the thermal displacement amount from the inference data using the thermal displacement model (step S124). The estimated value of the thermal displacement amount is calculated in the same time cycle as the inference data generated in step S123 and output to control device 15A of machine tool 1A.

[0102] The control device 15A of the machine tool 1A uses the estimated value of the thermal displacement to correct the operation amount to the actuator, thereby correcting the position command and performing the machining (step S125). The machine tool 1A determines whether the program operation of the machining to be corrected for the thermal displacement has been completed (step S126). If the program operation has not been completed (step S126: No), the process returns to step S123, and the machining system 100-2 repeats the processing from step S123 to step S126. If the program operation has been completed (step S126: Yes), the machining system 100-2 ends the processing.

[0103] As described above, machining system 100 - 2 of Embodiment 2 includes thermal displacement estimation device 3 that estimates the thermal displacement amount using the learned thermal displacement model. Machine tool 1A corrects the operation amount of the actuator of machine tool 1A based on the estimated value of the thermal displacement amount so as to cancel the thermal displacement indicated by the estimated value.

[0104] In addition, the inference unit 33 can successively calculate the estimated value of the thermal displacement amount for the inference data successively input and output it to the machine tool 1A. Therefore, the thermal displacement amount can be estimated regardless of the time required for the machining program executed by the machine tool 1A. In addition, since the estimated value of the thermal displacement amount is calculated based on the thermal displacement model as a time recursive model, the thermal displacement amount can be estimated including not only the instantaneous temperature data and machine state data for calculating the estimated value of the thermal displacement amount, but also the influence of the change of these data over time. Therefore, the machine tool 1A can correct the thermal displacement with good accuracy without depending on the elapsed time from the start of operation.

[0105] In addition, as another method, in Fig.15 In step S126, it is determined whether the power of the machine tool 1A is turned off. The machining system 100-2 may return to step S123 and continue processing before the power of the machine tool 1A is turned off. In this case, even if the program is executed multiple times in succession, the thermal displacement can be continuously corrected.

[0106] Here, the hardware structure of the thermal displacement model learning device 2 is described. The thermal displacement model learning device 2 of Embodiment 1 and Embodiment 2 is implemented, for example, by a computer system. The thermal displacement model learning device 2 can be implemented by one computer system or by multiple computer systems. For example, the thermal displacement model learning device 2 can be implemented by a cloud system. In a cloud system, the division between the hardware of the computer system and devices such as servers of various functions can be arbitrarily set. For example, one computer system can have the functions of multiple devices, or multiple computer systems can have the functions of one device.

[0107] A configuration example of a computer system that realizes the thermal displacement model learning device 2 will be described. Fig.16 FIG. 1 is a diagram showing an example of a hardware configuration. Fig.16 As shown, the computer system includes a control unit 101 , an input unit 102 , a storage unit 103 , a display unit 104 , a communication unit 105 , and an output unit 106 , which are connected via a system bus 107 .

[0108] exist Fig.16 In the embodiment, the control unit 101 is, for example, a CPU (Central Processing Unit). The control unit 101 executes a thermal displacement model learning program that describes the various processes implemented by the thermal displacement model learning device 2 of this embodiment. The input unit 102 is composed of, for example, a keyboard, a mouse, etc., and is used by the user of the computer system to input various information. The storage unit 103 includes various memories such as RAM (Random Access Memory), ROM (Read Only Memory), and storage devices such as a hard disk, and stores the program executed by the above-mentioned control unit 101, the necessary data obtained during the processing, etc. In addition, the storage unit 103 is also used as a temporary storage area for the program. The display unit 104 is composed of an LCD (Liquid Crystal Display) and the like, and displays various screens to the user of the computer system. The communication unit 105 is a communication circuit that implements communication processing. The communication unit 105 can also be composed of a plurality of communication circuits corresponding to a plurality of communication methods. The output unit 106 is an output interface that outputs data to an external device such as a printer and an external storage device.

[0109] in addition, Fig.16 is an example, and the structure of the computer system is not limited to Fig.16 For example, the computer system may not have the output unit 106. In addition, when the thermal displacement model learning device 2 is implemented by multiple computer systems, these computer systems may not all be Fig.16For example, a portion of the computer system may not have Fig.16 At least one of the display portion 104 , the output portion 106 , and the input portion 102 shown.

[0110] Here, an operation example of the computer system before the state of the thermal displacement model learning program in which the processing of the thermal displacement model learning device 2 of the first and second embodiments is executed is described. The computer system of the above configuration installs the thermal displacement model learning program in the storage unit 103 from, for example, a CD-ROM or DVD-ROM set in a CD (Compact Disc)-ROM drive or a DVD (Digital Versatile Disc)-ROM drive (not shown). Then, when the thermal displacement model learning program is executed, the thermal displacement model learning program read from the storage unit 103 is stored in an area of ​​the main storage device of the storage unit 103. In this state, the control unit 101 executes the processing of the thermal displacement model learning device 2 of the first and second embodiments according to the thermal displacement model learning program stored in the storage unit 103.

[0111] In addition, in the above description, a CD-ROM or DVD-ROM is used as a recording medium to provide a program that records the processing in the thermal displacement model learning device 2, but it is not limited to this. Depending on the structure of the computer system, the capacity of the program provided, etc., for example, a program provided by a transmission medium such as the Internet via the communication unit 105 can be used.

[0112] The thermal displacement model learning program of the first embodiment and the second embodiment causes a computer to execute the following steps: a step of acquiring learning data that associates input data including at least temperature data indicating the time series temperature of the machine tool 1 with thermal displacement data indicating the time series thermal displacement amount of the machine tool 1 in time; a step of generating a data set by extracting a part of the learning data from the learning data; and a step of generating a learned thermal displacement model for estimating the thermal displacement amount based on the input data of the machine tool 1A using the data set. Here, in the step of generating the data set, the data set is generated so that the time lengths of the time intervals extracted in at least two of the plurality of data sets are different from each other.

[0113] Figure 5 The learning data acquisition unit 21 shown in FIG. Fig.16 The communication unit 105 shown implements, Figure 5 The model storage unit 24 shown is Fig.16 A portion of the storage unit 103 is shown. Figure 5 The data set generation unit 22 and the learning unit 23 shown are composed of Fig.16 The control unit 101 shown is implemented.

[0114] Similar to the thermal displacement model learning device 2, the thermal displacement estimation device 3 is also implemented by one or more computer systems. The program for causing the thermal displacement estimation device 3 to execute the operations described in the second embodiment is provided by a storage medium, a transmission medium, etc. and installed in the computer system, similar to the thermal displacement model learning program described above. Thus, the above operations in the thermal displacement estimation device 3 are implemented.

[0115] Fig.14 The inference data acquisition unit 31 and the thermal displacement model acquisition unit 32 shown in FIG. Fig.16 The communication unit 105 shown implements, Fig.14 The inference unit 33 shown is composed of Fig.16 The control unit 101 shown is implemented.

[0116] in addition, Figure 5 and Fig.14 The division of functions in the devices shown is an example. If the processing systems 100-1 and 100-2 can perform the above-mentioned operations, the division of functions in the devices is not limited to Figure 5 and Fig.14 For example, as shown in the following modified example, the thermal displacement model learning device 2 and the thermal displacement estimation device 3 may be integrated so that the thermal displacement model learning device 2 also functions as the thermal displacement estimation device 3 .

[0117] Fig.17 1 is a diagram showing the functional structure of a thermal displacement model learning device 2A according to a modified example of the second embodiment. The thermal displacement model learning device 2A has the function of a thermal displacement estimation device 3 in addition to the function of the thermal displacement model learning device 2. The functions of each component have been described in detail in Fig.14 This is explained in , so it is omitted here. Fig.17 In the illustrated configuration, when the function of the thermal displacement model learning device 2A is realized by one computer system, the function of the thermal displacement model acquisition unit 32 is realized by exchanging data within the computer system.

[0118] Implementation method 3.

[0119] Fig.18 1 is a diagram showing the structure of a machining system 100-3 according to Embodiment 3. The machining system 100-3 includes a plurality of machine tools 1 and a thermal displacement model learning device 2B provided on a server. The thermal displacement model learning device 2B learns a thermal displacement model based on thermal displacement data, temperature data, and machine status data received from the plurality of machine tools 1. The function of the machine tool 1 is the same as that of Embodiment 1, and the function of the thermal displacement model learning device 2B is the same as that of Embodiment 1 except that the thermal displacement model of the plurality of machine tools 1 is learned for each machine tool 1.

[0120] Implementation method 4.

[0121] Fig.19 1 is a diagram showing a configuration of a machining system 100 - 4 according to Embodiment 4. The machining system 100 - 4 includes a plurality of machine tools 1 and 1A, a thermal displacement model learning device 2B provided on a server, and a thermal displacement estimation device 3B.

[0122] The machine tool 1 is a processing device for learning, and is the same as the machine tool 1 of the first embodiment except that the feature information is output to the thermal displacement model learning device 2B, wherein the feature information is information indicating the features of the machine tool 1 and includes at least the model of the machine tool 1. The machine tool 1A is a processing device for estimation, and is the same as the machine tool 1A of the second embodiment except that the feature information is output to the thermal displacement estimation device 3B, wherein the feature information is information indicating the features of the machine tool 1A and includes at least the model of the machine tool 1A.

[0123] The thermal displacement model learning device 2B is similar to the thermal displacement model learning device 2B in the third embodiment except that it receives characteristic information from the machine tool 1, and includes: a learning data acquisition unit 21B that generates learning data based on data received from a plurality of machine tools 1; and a model storage unit 24B that stores a plurality of thermal displacement models in association with characteristic information of the machine tool 1 that is a source of learning data. The thermal displacement estimation device 3B includes a thermal displacement model acquisition unit 32B instead of the thermal displacement model acquisition unit 32 of the thermal displacement estimation device 3. The thermal displacement model acquisition unit 32B includes a model selection unit 321 that selects a thermal displacement model to be used by the inference unit 33 from a plurality of thermal displacement models stored in the model storage unit 24B based on the characteristic information received from the machine tool 1A that is the inference target.

[0124] Here, the characteristic information includes at least the model number, and in addition to the model number, may include information indicating the state of the machine tools 1 and 1A, such as the environment and installation conditions of the factory where the machine tools 1 and 1A are installed. The model selection unit 321 may select a thermal displacement model generated based on data obtained from a machine tool 1 of the same model as the machine tool 1A to be inferred, and the state of the machine tool 1 is similar to that of the machine tool 1A to be inferred. Thus, even if there is no thermal displacement model generated based on the data of the machine tool 1A to be inferred, it is possible to obtain an estimated value of the thermal displacement amount by using a thermal displacement model generated based on the data of the machine tool 1 of the same model as the machine tool 1A, that is, the machine tool 1 in a similar environment to that of the machine tool 1A to be inferred.

[0125] In addition, in the above, it is described that the thermal displacement model acquisition unit 32B has a structure with a model selection unit 321, but it can also be a structure as follows: the thermal displacement model acquisition unit 32B obtains multiple types of thermal displacement models from the model storage unit 24B, and the inference unit 33 selects an estimated value to be output based on the feature information from multiple estimated values ​​of thermal displacement calculated using multiple types of thermal displacement models.

[0126] The structures shown in the above embodiments are merely examples, and may be combined with other known technologies, the embodiments may be combined with each other, and part of the structures may be omitted or modified within a scope not departing from the gist.

[0127] For example, the functions of the thermal displacement model learning device 2 and the thermal displacement estimation device 3 may be built into the machine tool 1. In addition, when the thermal displacement model is learned using data obtained from a plurality of machine tools 1 as in Embodiments 3 and 4, the thermal displacement model learning device 2B may obtain learning data from a plurality of machine tools 1 used in the same area, or may learn the thermal displacement model using learning data collected from a plurality of machine tools 1 operating independently in different areas. In addition, the machine tool 1 for which the learning data is collected may be added to or removed from the object midway. Furthermore, the thermal displacement model learning device 2 that has already learned the thermal displacement model for a certain machine tool 1 may be applied to another machine tool 1 to relearn and update the thermal displacement model for the other machine tool 1.

[0128] Description of Reference Numerals

[0129] 1, 1A machine tool; 2, 2A, 2B thermal displacement model learning device; 3, 3B thermal displacement estimation device; 11 feed axis; 12 feed axis motor; 13 spindle; 14 spindle motor; 15, 15A control device; 16 peripheral equipment; 17 bed; 18 column; 19 feed screw; 21, 21B learning data acquisition unit; 22 data set generation unit; 23 learning unit; 24, 24B model storage unit; 31 reasoning data acquisition unit; 32, 32B thermal displacement model acquisition unit; 33 reasoning unit; 100-1, 100-2, 100-3, 100-4 machining system; 101 control unit; 102 input unit; 103 storage unit; 104 display unit; 105 communication unit; 106 output unit; 321 model selection unit; DS displacement sensor; T tool; TA worktable; TS temperature sensor; W workpiece.

Claims

1. A thermal displacement model learning device, characterized in that: have: A learning data acquisition unit acquires learning data, wherein the learning data establishes a correspondence between input data and thermal displacement data in time, wherein the input data at least includes temperature data indicating a time series temperature of a machine tool, and the thermal displacement data indicates a time series thermal displacement amount of the machine tool; a data set generating unit for generating a data set by extracting a part of the learning data from the learning data; as well as a learning unit, which generates a learned thermal displacement model using the data set, wherein the learned thermal displacement model is used to estimate the thermal displacement amount according to the input data of the machine tool, The data set generation unit generates the data sets so that the time lengths of the extracted time intervals are different from each other in at least two of the plurality of data sets.

2. The thermal displacement model learning device according to claim 1, characterized in that: In addition to the temperature data, the input data also includes machine status data representing an operating status of the machine tool.

3. The thermal displacement model learning device according to claim 1 or 2, characterized in that: The learning unit optimizes internal parameters of the thermal displacement model so as to minimize a difference between an estimated value of a thermal displacement amount which is an output of the thermal displacement model when the input data included in the data set is input and the thermal displacement data corresponding to the input data input to the thermal displacement model, thereby generating the learned thermal displacement model.

4. The thermal displacement model learning device according to any one of claims 1 to 3, characterized in that: Also available: an inference data acquisition unit that acquires inference data, the inference data being input data for inference including at least temperature data indicating a time-series temperature of the machine tool; as well as The inference unit outputs an estimated value of the thermal displacement amount of the machine tool based on the inference data input from the inference data acquisition unit using the thermal displacement model.

5. The thermal displacement model learning device according to claim 4, characterized in that: The inference data includes machine status data indicating an operating status of the machine tool in addition to the temperature data.

6. The thermal displacement model learning device according to claim 2 or 5, characterized in that: The machine state data indicates an operating state of a peripheral device that adjusts a temperature of the machine tool and an operating state of an actuator included in the machine tool.

7. The thermal displacement model learning device according to any one of claims 1 to 3, characterized in that: The learning data acquisition unit acquires the learning data from the plurality of machine tools. The learning unit generates the thermal displacement model for each of the machine tools.

8. The thermal displacement model learning device according to any one of claims 1 to 3, characterized in that: The learning data acquisition unit acquires the learning data from the plurality of machine tools. The learning unit generates the thermal displacement model for each of the machine tools, and associates and stores the generated thermal displacement model with characteristic information, wherein the characteristic information is information indicating the characteristics of the machine tool and includes at least the model of the machine tool that is the source of the learning data used to generate the thermal displacement model.

9. The thermal displacement model learning device according to claim 8, characterized in that: Also available: an inference data acquisition unit that acquires inference data, the inference data being input data for inference including at least temperature data indicating a time-series temperature of a machine tool to be inferred; and The inference unit uses the thermal displacement model of the same model as the inference target machine tool among the plurality of thermal displacement models to output an estimated value of the thermal displacement amount of the inference target machine tool based on the inference data input from the inference data acquisition unit.

10. The thermal displacement model learning device according to claim 9, characterized in that: In addition to the model, the characteristic information also includes information indicating the operating conditions of the machine tool. The inference unit outputs an estimated value of the thermal displacement amount of the machine tool to be inferred, using the thermal displacement model generated based on the learning data obtained from a machine tool of the same model as the machine tool to be inferred, that is, from a machine tool whose operating conditions are similar to those of the machine tool to be inferred.

11. A thermal displacement estimation device, characterized in that: have: a thermal displacement model acquisition unit that acquires the thermal displacement model learned by the thermal displacement model learning device according to claim 1; an inference data acquisition unit that acquires estimation data, the estimation data being input data for inference including at least temperature data indicating a time series temperature of the machine tool; as well as The inference unit uses the thermal displacement model acquired by the thermal displacement model acquisition unit to output an estimated value of the thermal displacement amount based on the input data input from the inference data acquisition unit.

12. A processing system, characterized in that: have: The thermal displacement model learning device according to any one of claims 1 to 10, or the thermal displacement estimation device according to claim 11; and The machine tool, The machine tool has: Knives; A spindle that rotates the tool or the object to be processed; and The feed axis changes the relative position between the tool and the object to be processed. The object to be processed is removed by the tool.

13. A processing method, which is a processing method using a machine tool, wherein the machine tool comprises a tool, a spindle for rotating the tool or a processing object, and a feed axis for changing the relative position between the tool and the processing object, and the processing object is subjected to removal processing by the tool, wherein the processing method comprises: A step of acquiring learning data, wherein the learning data establishes a correspondence between input data and thermal displacement data in time, wherein the input data at least includes temperature data indicating a time series temperature of the machine tool, and the thermal displacement data indicates a time series thermal displacement amount of the machine tool; a step of generating a data set by extracting a part of the learning data from the learning data; The step of using the data set to generate a learned thermal displacement model, the learned thermal displacement model is used to estimate the thermal displacement amount based on the input data of the machine tool; A step of estimating the thermal displacement amount based on input data for inference including at least the temperature data using the learned thermal displacement model; a step of correcting an operation amount of at least one of an actuator driving the main axis of the machine tool and an actuator driving the feed axis using the estimated value of the thermal displacement amount; as well as The step of performing the removal process using the corrected operation amount, In the step of generating the data set, the data set is generated such that in at least two of the plurality of data sets, time lengths of the extracted time intervals are different from each other.

Images