Thermal displacement data processing device
The thermal displacement data processing device addresses the challenge of accurately estimating thermal displacement in machine tools by reducing data dimensions and providing recommended collection conditions, thereby improving the accuracy of thermal displacement compensation and optimizing machine tool operations.
Patent Information
- Application Number
- PCT/JP2023/017907
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-12
- Publication Date
- 2024-11-21
AI Technical Summary
Accurately estimating thermal displacement in machine tools and robots is challenging due to varying relationships between temperature sensor readings and thermal displacement, making it difficult to correct for positional deviations caused by thermal expansion, and determining optimal sensor placement is also complex.
A thermal displacement data processing device that includes a data acquisition unit, a dimension reduction unit using feature selection and extraction methods, and a presentation information creation unit to process multidimensional thermal displacement data, reducing data dimensions and providing recommended collection conditions for improved estimation and compensation of thermal displacement.
Enables relatively accurate estimation of thermal displacement, facilitating better compensation for machine tool operations and optimizing machine tool settings for enhanced accuracy and performance.
Smart Images

Figure JP2023017907_21112024_PF_FP_ABST
Abstract
Description
Thermal displacement data processing device
[0001] The present invention relates to a thermal displacement data processing device.
[0002] For example, machining accuracy may be reduced due to tool positional deviation caused by thermal displacement of the drive mechanism of a machine tool, robot, etc. For this reason, it has been proposed to estimate the thermal displacement of the members of the drive mechanism based on the detection values of multiple temperature sensors arranged in the drive mechanism and correct the drive amount to compensate for the tool positional deviation caused by the thermal displacement (see, for example, Patent Document 1).
[0003] International Publication No. 2012 / 114493
[0004] The relationship between the detected values of each temperature sensor and thermal displacement varies depending on factors such as the type of heat generation, the temperature distribution of the components, and the ambient temperature. Therefore, it is difficult to accurately estimate thermal displacement, and it is not easy to determine where to place the temperature sensors in order to accurately estimate thermal displacement. Therefore, a technology that enables relatively accurate estimation of thermal displacement is desired.
[0005] A thermal displacement data processing device according to one aspect of the present disclosure is a thermal displacement data processing device that processes a thermal displacement data group including multiple thermal displacement data tables, each of which has a data series of one or more thermal displacement amounts in a drive mechanism and a data series of measurement values from multiple temperature sensors, and is equipped with a data acquisition unit that acquires the thermal displacement data group, a dimension reduction unit that reduces the dimension of the thermal displacement data group, and a presentation information creation unit that creates presentation information to be presented to a user based on the dimension-reduced thermal displacement data group.
[0006] 1 is a block diagram illustrating a configuration of a thermal displacement data processing device according to an embodiment of the present disclosure. 2 is a table illustrating an example of derivation of recommended conditions by a condition derivation unit in FIG.
[0007] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present disclosure will now be described with reference to the accompanying drawings. Fig. 1 is a block diagram showing the configuration of a thermal displacement data processing device 1 according to an embodiment of the present disclosure.
[0008] The thermal displacement data processing device 1 is a device that processes a group of thermal displacement data in a machine tool (drive mechanism) 2. The thermal displacement data processing device 1 can be realized by causing one or more computers having a memory, a processor, an input / output interface, etc. to execute an appropriate data processing program. The thermal displacement data processing device 1 may be configured integrally with a control device 22 that controls the machine tool 2 or a management server that manages the machine tool 2.
[0009] The thermal displacement data group processed by the thermal displacement data processing device 1 is a data file or data set containing multiple multidimensional (multi-series) thermal displacement data tables (data sheets) with multiple data series (data items). Each thermal displacement data table contains one or more data series of thermal displacement amounts in the machine tool 2 and data series of measurement values from multiple temperature sensors 21. A data series refers to a collection of repeatedly acquired data (individual acquired values) acquired from the same data source (e.g., a sensor) under the same collection conditions, and generally includes multiple data acquired at regular time intervals or whenever a specific event occurs. The thermal displacement data table includes a data series recording thermal displacement values, multiple data series recording temperature sensor measurement values corresponding to each thermal displacement data, and a data series consisting of data identifying the acquisition timing of the data in the other data series (e.g., time, serial number, etc.). A thermal displacement data group may include multiple data tables acquired under different collection conditions for the same multiple data items. Collection conditions may include, for example, spindle rotation speed, heat generation pattern, ambient temperature, etc. Furthermore, the collection conditions in each data table do not need to be strictly constant, but may be within the range of a roughly set division.
[0010] The thermal displacement data processing device 1 includes a data acquisition unit 11 that acquires a thermal displacement data group, a dimension reduction unit 12 that reduces the dimension of the thermal displacement data group, and a presentation information creation unit 13 that creates presentation information to be presented to a user based on the dimension-reduced thermal displacement data group. Note that the components of the thermal displacement data processing device 1 are classifications of the functions of the thermal displacement data processing device 1, and do not necessarily have to be clearly distinguishable in terms of physical configuration and program configuration.
[0011] The data acquisition unit 11 acquires a group of thermal displacement data (data files or data sets) that may include multiple multidimensional (multi-series) thermal displacement data tables (data sheets) having multiple data series (data items), for example, from the control device 22 of the machine tool 2, a management server (not shown) that manages the machine tool 2, etc.
[0012] The dimension reduction unit 12 generates a dimension-reduced data group (a dimension-reduced thermal displacement data group) by replacing the features of each data table in the thermal displacement data group with a data table that represents a smaller number of data series. In other words, "dimension reduction" means reducing the number of data series. The dimension reduction unit 12 may be configured to reduce the dimension of the thermal displacement data group by well-known "feature selection," or may be configured to reduce the dimension of the thermal displacement data group by well-known "feature extraction," or may be configured to switch between "feature selection" and "feature extraction" as needed. The dimension reduction unit 12 of this embodiment includes a feature selection unit 121 that performs dimension reduction by feature selection and a feature extraction unit 122 that performs dimension reduction by feature extraction.
[0013] The feature selection unit 121 selects data series highly correlated with thermal displacement and generates a dimension-reduced data set (dimensionally reduced thermal displacement data set) consisting of the thermal displacement data series and the selected data series. In other words, the feature selection unit 121 generates a dimension-reduced data set with a reduced number of data series by deleting data series with low correlation with thermal displacement. A known method for selecting data series involves calculating an evaluation value for each data series and selecting data series in descending order of evaluation value. Specific methods for calculating the evaluation value include MDA (Mean Decrease Accuracy), mRMR (Minimum Redundancy Maximum Relevance), and LASSO (Least Absolute Shrinkage and Selection Operator). The thermal displacement data set may include data series useful for analysis regardless of their correlation with thermal displacement, such as measurement time. However, such data series are preferably excluded from the dimension reduction process and retained in the dimension-reduced data set. By performing dimension reduction using the feature selection unit 121, a dimension-reduced data set can be obtained relatively easily.
[0014] The feature selection unit 121 preferably classifies multiple temperature sensors into multiple groups and suppresses duplicate selection from the same group. As a specific example, the feature selection unit 121 may be configured to calculate an evaluation value for each data series, and then multiply the evaluation value of each data series by a larger correction coefficient the higher the ranking of the data series within the group, thereby suppressing duplicate selection from the same group. Alternatively, the feature selection unit 121 may select data series with the highest evaluation values in order from among the data series with the highest evaluation values within a temperature sensor group. A temperature sensor group is composed of temperature sensors that are considered to have a similar relationship with thermal displacement. As a specific example, classifying temperature sensors arranged in the same component of a drive mechanism into the same group can prevent the temperature of a specific component from being evaluated as being excessively high.
[0015] The feature extraction unit 122 may be configured to treat the thermal displacement data group as a matrix and generate a dimension-reduced data group by performing a matrix operation on the thermal displacement data group. Examples of matrix operation methods for generating the dimension-reduced data group include PCA (principal component analysis), LSA (latent semantic analysis), LDA (linear discriminant analysis), and ICA (independent component analysis). By performing dimension reduction using such a feature extraction unit 122, it is possible to generate a dimension-reduced data group that more accurately reflects the overall characteristics of the thermal displacement data group.
[0016] The presentation information creation unit 13 preferably creates, as presentation information to be presented to the user, at least one of statistics calculated from the dimension-reduced data group, a graph of the dimension-reduced data group, and recommended conditions recommended for collecting a new thermal displacement data table. In this embodiment, the presentation information creation unit 13 includes a statistics calculation unit 131 that calculates statistics of the dimension-reduced data group, a graph creation unit 132 that creates a graph of the dimension-reduced data group, and a condition derivation unit 133 that derives recommended conditions for collecting a new thermal displacement data table based on the dimension-reduced data group. The presentation information creation unit 13 is preferably configured to display the created presentation information on the screen of the display device 3 to present it to the user. The display device 3 may be connected to the thermal displacement data processing device 1 or may be integrated with the thermal displacement data processing device 1.
[0017] The statistical quantity calculation unit 131 preferably includes, as the presented information, at least one of standard deviation, variance, kurtosis, skewness, and density, or at least two of maximum, minimum, local maximum, and local minimum, as the presented information. By presenting such statistical quantities, the user can relatively easily grasp the trend of thermal displacement.
[0018] The graph creation unit 132 graphs the data of the dimension-reduced data group and presents it to the user. The graph created by the graph creation unit 132 is preferably in a form that allows the user to determine the range and bias of the data. As a specific example, by plotting the data of the dimension-reduced data group generated by feature selection on a three-dimensional graph with thermal displacement, the temperature detected by the sensor, and the distance from the sensor reference point as three axes and allowing the user to change the viewpoint, the user can relatively easily identify areas with low data density, particularly areas where it is preferable to supplement the data in order to understand the trend of thermal displacement from the trend of other data.
[0019] The condition derivation unit 133 derives recommended conditions, which are collection conditions that are considered to be capable of acquiring a thermal displacement data table that allows new data to be added to areas with low data density in the dimension-reduced data group. By acquiring a new thermal displacement data table in accordance with the recommended conditions, it becomes possible to more accurately estimate thermal displacement, thereby optimizing the settings of the machine tool 2 and enabling more accurate operation.
[0020] A specific example of the derivation of recommended conditions by the condition derivation unit 133 will be described with reference to the exemplary data shown in FIG. 2 . First, each item (environmental temperature change pattern, lower limit of environmental temperature, upper limit of environmental temperature, machine tool heat generation operation pattern, and spindle rotation speed) of the collection conditions (collection conditions 1 to 6) in each thermal displacement data table of the dimension-reduced data group is classified into categorical variables and continuous variables. Note that a "categorical variable" is a variable with discrete values, such as the environmental temperature change pattern and the machine tool heat generation operation pattern in this example. A "continuous variable" is a variable with continuous values, such as the lower limit of environmental temperature, the upper limit of environmental temperature, and the spindle rotation speed in this example. Next, acquisition conditions for which all categorical variable values match are classified into the same group. In Table 1, there are Group 1 (collection conditions 1 and 2) and Group 2 (collection conditions 5 and 6). Collection conditions for which all categorical variable values do not match other collection conditions (collection conditions 3 and 4) are not considered to belong to any group. Furthermore, the distance between data items in the same group is calculated. The distance between data items can be calculated using the sum or average of the Euclidean distances for all sampling points in the data at the same time. An intermediate condition between the collection conditions (collection conditions 1 and 2) with the largest distance between the data, that is, a collection condition (collection condition 7) in which the categorical variables are equal and the continuous variables are at their intermediate values, is calculated and designated as the recommended condition.
[0021] As described above, the thermal displacement data processing device 1 of this embodiment reduces the dimension of the thermal displacement data group by reducing the dimension, making it easier for the user to grasp the overall picture, making it easier to collect the data necessary to more accurately grasp the thermal displacement trends, and enabling the machine tool 2 to be configured so that the thermal displacement can be accurately compensated for.
[0022] The following supplementary notes are further disclosed regarding the above-described embodiment and modified examples: (Supplementary Note 1) A thermal displacement data processing device (1) processes a thermal displacement data group including a plurality of thermal displacement data tables each having a data series of one or more amounts of thermal displacement in a drive mechanism (2) and a data series of measurement values from a plurality of temperature sensors (21), and includes a data acquisition unit (11) that acquires the thermal displacement data group, a dimension reduction unit (12) that reduces the dimension of the thermal displacement data group, and a presentation information creation unit (13) that creates presentation information to be presented to a user based on the dimension-reduced thermal displacement data group.
[0023] (Supplementary Note 2) In the thermal displacement data processing device (1) of Supplementary Note 1, the dimension reduction unit (12) may reduce the dimension of the thermal displacement data group by feature selection.
[0024] (Supplementary Note 3) In the thermal displacement data processing device (1) of Supplementary Note 2, the dimension reduction unit (12) may classify the plurality of temperature sensors into a plurality of groups and suppress duplicate selection from the same group.
[0025] (Supplementary Note 4) In the thermal displacement data processing device (1) of Supplementary Note 1, the dimension reduction unit (12) reduces the dimension of the thermal displacement data group by feature extraction. The thermal displacement data processing device according to claim 1.
[0026] (Supplementary Note 5) In the thermal displacement data processing device (1) of Supplementary Notes 1 to 4, the presented information may include one or more statistics.
[0027] (Supplementary Note 6) In the thermal displacement data processing device (1) of Supplementary Note 5, the one or more statistical quantities may include at least one of standard deviation, variance, kurtosis, skewness, and density, or at least two or more of maximum value, minimum value, local maximum value, and local minimum value.
[0028] (Appendix 7) In the thermal displacement data processing device (1) of Appendices 1 to 6, the data acquisition unit (11) further acquires collection conditions for the thermal displacement data table, and the presented information may include recommended conditions derived from the collection conditions for the thermal displacement data table and the dimension-reduced thermal displacement data group as new collection conditions recommended when a new thermal displacement data table is collected.
[0029] (Supplementary Note 8) In the thermal displacement data processing device (1) of Supplementary Notes 1 to 7, the presented information may include a graph of a group of dimension-reduced thermal displacement data.
[0030] Although the present disclosure has been described in detail above, the present disclosure is not limited to the individual embodiments described above. Various additions, substitutions, modifications, partial deletions, etc. are possible to these embodiments without departing from the gist of the present disclosure or the spirit of the present disclosure derived from the content of the claims and their equivalents. For example, the thermal displacement data processing device according to the present disclosure may be used to process thermal displacement data groups in drive mechanisms other than machine tools, such as robots.
[0031] REFERENCE SIGNS LIST 1 Thermal displacement data processing device 11 Data acquisition unit 12 Dimension reduction unit 121 Feature selection unit 122 Feature extraction unit 13 Presentation information creation unit 131 Statistical quantity calculation unit 132 Graph creation unit 133 Condition derivation unit 2 Machine tool (drive mechanism) 21 Temperature sensor 22 Control device 3 Display device
Claims
1. A thermal displacement data processing device that processes a thermal displacement data group including multiple thermal displacement data tables each having a data series of one or more thermal displacement amounts in a driving mechanism and a data series of measurement values of multiple temperature sensors, comprising: a data acquisition unit that acquires the thermal displacement data group; a dimension reduction unit that reduces the dimension of the thermal displacement data group; and a presentation information creation unit that creates presentation information to be presented to a user based on the dimension-reduced thermal displacement data group.
2. The thermal displacement data processing device according to claim 1, wherein the dimensionality reduction unit reduces the dimensionality of the thermal displacement data group by feature selection.
3. The thermal displacement data processing device according to claim 2, wherein the dimensionality reduction unit classifies the plurality of temperature sensors into a plurality of groups and suppresses duplicate selection from the same group.
4. The thermal displacement data processing device according to claim 1, wherein the dimensionality reduction unit reduces the dimensionality of the thermal displacement data group by feature extraction.
5. A thermal displacement data processing device according to any one of claims 1 to 4, wherein the presented information includes one or more statistics.
6. A thermal displacement data processing device according to claim 5, wherein the one or more statistics include at least one of standard deviation, variance, kurtosis, skewness and density, or at least two or more of maximum value, minimum value, local maximum value and local minimum value.
7. A thermal displacement data processing device as described in any one of claims 1 to 6, wherein the data acquisition unit further acquires collection conditions for the thermal displacement data table, and the presented information includes recommended conditions derived from the collection conditions for the thermal displacement data table and the dimensionally reduced group of thermal displacement data as new collection conditions recommended when newly collecting the thermal displacement data table.
8. A thermal displacement data processing device according to any one of claims 1 to 7, wherein the presented information includes a graph of the group of thermal displacement data whose dimensions have been reduced.
Citation Information
Patent Citations
Real-time error compensation for machine tools using full differential wet modeling
JP2002515995A
Machine learning device and thermal displacement correction device
JP2018153901A
Numerical control device
JP2019025552A
Fluctuation amount estimation device in machine tool, and correction amount calculation device
JP2020179429A
