TW201223690A - Method and apparatus of machine tools for intelligently compensating thermal error - Google Patents

Abstract

A method and apparatus of machine tools for intelligently compensating thermal error is disclosed. It comprises capturing heat behavior data, selecting warming-up stable situation by experts, constructing warming-up characteristic description with method of support vector data description with stable temperature information, screening by warming-up characteristic description to conform to the heat behavior in stable situation, constructing thermal error model with method of support vector regression, and detecting warming-up on line and computing thermal error.

Description

201223690 VI. Description of the invention: [Technical field to which the invention belongs] = The invention is related to the thermal error adjustment of the machine tool, especially regarding the method of warm-up judgment and adaptive learning. u (4) tool machine thermal error intelligent adjustment device and [prior art] mechanical advancement 卫 wei process 'regardless of motor, liquid I system and Ding energy conversion' regardless of the conversion path 'most of the i caused: ' ΐ some heat 1 caused inside the body and The temperature changes around it, the most accurate, _ error (the problem of the hot problem of the figure is that the precision machine development process must always face _ is contributed by the thermal error: see the difference has an impact, play Absolutely critical role ~ machining accuracy can be 'in order to reduce thermal error interference, the machine 11 temperature rise must be reached, (or called the state), then start to go to the second eve of the reliance / extensive judgment to warm up Whether or not, as shown in Fig. 3 and Fig. 4. The response strategy of the method is mainly divided into two kinds of prediction models, which are soft and squatting, and the thermal error of the construction tool is based on the active error quantity. In the design stage, I will try my best to e-Ting Li Shan

Reduced' is intended to control or avoid thermal errors. °, to produce I and passive compensation research and pseudo-, and. The research and technical overview of the active anti-Bei-Jie and the technical overview are summarized in Table- and Table 2 201223690. Table Workers Take the means to represent the literature __ Heat flow control 1 · External heat source: environmental thermostatic control and structural thermal insulation [2], structural active cooling [3, 4]. ------------ A cold part [2, 5, 6] and heat source temperature control [7]. 1 Other material selection [8]. Structural optimization design 2. Error analysis and symmetrical configuration design [5,9,1〇]. --------- Soil rabbit ^ configuration and hardware compensation design [11 j 21. Table 2, tool machine thermal error passive compensation | taking means connotation and representative literature _5 return analysis compensation model mathematical statistical model, suitable for steady-state thermal error compensation "13141. Earth lexical network model mathematical statistical model" suitable for steady-state heat Error compensation C13 151. Prime analysis model _^ type seam contains the physical caution of the machine tool structure, the actual application is still difficult 丨13]. __^ state model can adapt to different operating conditions or environmental variation error model [丨6 _丨9]. Industry technology '----- OKUMA thermal affinity concept [5], MAZAK smart thermal protection [1〇], MIKRON system [20], FANUC artificial intelligence thermal compensation [21]. Lu Xiang is more active than The design of thermal suppression is more convenient and economical by adopting the thermal error software compensation. It is not directly removing or less thermal error generated by the machine tool, but using the experimental measurement results for the analysis of the difference. By means of software to compensate for the impact of errors, this method is also widely used in foreign tool machine factories, such as 曰本Mazak and 〇kUma, Switzerland Mikrcm, etc. Therefore, how to improve the existing compensation technology, research and development more accurate 4 201223690 The more reliable thermal error compensation method is the long-term goal of the tool machine industry. However, from the past research results, it is found that the static compensation model using mathematical statistics can be used for the thermal error of the steady state of the machine tool. Obtaining good results, but the thermal error of the temporary (moving) state is quite tricky and difficult to handle. So far, the domestic industry is still unable to solve such problems. Based on the above problems, the inventor proposed a tool machine heat. Error intelligence adapting device and method thereof to overcome the defects of the prior art. Φ References: [1] Bryan, J. B,, 1990, "International status of thermal error research, 55 Annals of the CIRP 39/2, pp. 645-656.

[2] MAKIN0 website www.makino.co.ip [3] Muto, A·, 200.5, “Machine tool with a feature for preventing a thermal deformation,” U.S. Patent, No. 6,923,603.

[4] YASDA website www.vasda.co.ip 丨· [5] OKUMA website www.okuma.co.ip [6] Yintai Technology Co., Ltd. www.pmi-amt.com [7] M0RISEIKI website www.moriseiki .com/dixi/en2lish/products/control.html [8] Rahman, M., Mansur, MA, and Karim, MB, 2001, “Non-conventional materials for machine tool structures,” JSME Int. J., Series C , Vol. 44, No. 1, pp.1-11 · [9] Slocum, AH, 1992, Precision Machine Design, Society 201223690 of Manufacturing Engineers, 1st Edition.

[10] MAZAK website www.mazak.com [11] Kobari, T. and Takada, R., 1999, "Shuttle table device,, 5 Japan patent, No. 11-267938.

[12] Kato, K. and ITO, T., 2006, "Machine tool and posture maintenance device,5, Japan patent, No. 2006-341328.

[13] Ramesh, R., Mannan, MA and Poo AN, 2000, “Error Compensation in Machine Tools - A Review Part II: Thermal Errors, 55 Int. J. Mach. Tools Manufact., Vol. 40, pp.1257 -1284.

[14] Yang, JG, Ren, YQ, Liu, GL, Zhao, HT, Dou, XL, Chen, WZ, and He, SW, 2005, "Testing, variable selection and modeling of thermal errors on an INDEX-G200 Turning center,,5 Int. J. Adv. Manuf. Technol., Vol. 26, pp.814-818.

[15] Kang, Y" Chang, CW, Huang, Y" Hsu, C丄., Nieh, IF, 2007, "Modification of a neural network utilizing hybrid filters for the compensation of thermal deformation in machine tools," Int. J Mach. Tools Manufact., Vol. 47, pp.376-387.

[16] Ramesh, R., Mannan, MA Poo, AN, and Keerthi, SS, 2003, “Thermal error measurement and modelling in machine tools. Part I. Influence of varying operating conditions,” Int. J. Mach. Tools Manufact ., Vol. 43, 201223690 pp.391-404.

[17] Ramesh, R., Mannan, MA Poo, AN, and Keerthi, SS, 2003, “Thermal error measurement and modelling in machine tools. Part II. Hybrid Bayesian Network—support vector machine model, 5, Int. J. Mach. Tools Manufact., Vol. 43, pp.405-419.

[18] Yang, H. and Ni, J., 2005, “Adaptive model estimation of machine-tool thermal errors based on recursive dynamic modeling strategy,'' Int. J. Mach. Tools Manufact., Vol. 45, pp. 1-11.

[19] Yang, H. and Ni, J., 2005, "Dynamic neural network modeling for nonlinear, nonstationary machine tool, thermally induced error/5 Int. J. Mach. Tools Manufact., Vol. 45, pp.455- 465.

[20] MIKRON website www.mikron.com [21] FANUC website www.fanuc.co.ip Lu [22] Vapnik, V., Golowich, S. and Smola, AJ, 1997, “Support vector method for function approximation, Regression estimation, and signal processing, ''in Advances in Neural Information Processing Systems, Vol. 9, pp. 281-287 [23] D. Tax and R. Duin, “Support vector data description,” Machine Learning, vol. 54 , pp. 45-66, 2004 [24] Yi-Hung Liu, Yu-Kai Huang, and Ming-Jui Lee, 201223690 "Automatic inline-defect detection for TFT-LCD array process using locally linear embedding and support vector data description, Measurement Science and Technology, vol. 19, August 2008 [25] Yi-Hung Liu, Szu-Hsein Lin, Yi-Ling Hsueh, and Ming-Jiu Lee, Automatic target defect identification for TFT-LCD array process inspection using kernel fuzzy c -means based fuzzy SVDD ensemble", Expert Systems with Applications, vol. 40, 2008 [Invention] The present invention aims to provide a support vector regression (Support Vector R) Egression, SVR) [22] as a thermal error model, using Support Vector Data Description (SVDD) [23] to construct a steady-state mode range, to instantly adjust the thermal error of the machine tool thermal error smart adaptation device and Its method. Another object of the present invention is to provide a machine tool thermal error smart adaptation apparatus and method thereof that can perform incremental learning for different external temperature changes. In order to achieve the above objectives, this issue (4) provides a tool-intelligence error adjustment device, including: - warm-up feature description to construct a single-hearted machine tool body temperature distribution vector - steady-state mode range; - thermal error model construction Early 兀 'training out - nonlinear thermal error model; and a warm-up judgment and thermal error calculation unit, by Green South # 5 from the machine tool on the temperature sensing signal k and the red machine - operating condition information Determine the state of the warm-up 201223690 of the machine tool and calculate the thermal error compensation amount of at least one node. „„ Ί. Xuan"1* has a machine heat § Wu difference wisdom 5 week suitable device also includes - incremental learning early 兀 'adaptable parameter adjustment to ensure the accuracy of the thermal error model. For the purpose of the appropriate party, this issue (4) provides a tool frequency error wisdom adjustment f IP, sudden inclusion. Extraction hot riding; selected by the expert warm-up sorrow, to stabilize the state temperature information:: coffee screening The heat setting in accordance with the steady state:: error, transport = signal construction thermal error model; online warm-up judgment and advance (4) - Λ中 'f machine tool thermal error wisdom adjustment method includes incremental learning, hunting to adapt to environmental changes and Machine history literature [24] [25]. Number drifting 'similar to green reference' [Embodiment] Although the present invention uses several preferred embodiments, only the preferred embodiment of the present invention is described. (4) The specific embodiment disclosed below is only a sub- The following figures and specific implementation block diagrams. The machine tool thermal error intelligent adjustment device of the present embodiment with the smart sweater smart adjustment skirt is mainly included - the warm-up feature description construction unit 2, the -1 type construction unit 3, the hot county 2, = quantity Learning unit 5. For the 4th and 1st increase, please refer to Figure 6. The warm-up feature γ + — 射田述建早兀2 to support vector data 201223690 Description (SVDD) establishes a steady-state mode range of the body temperature distribution vector of the machine tool, ie The support vector data description (SVDD) confirms that the temperature state of each node a~d in the machine tool reaches a steady state until the support vector data description (SVDD) has the same dimensional feature vector distribution and the preset steady state mode range. . ^ The support vector data description (SVDD) is described in detail below. Purpose of SVDD: The judgment function D(x) is estimated from the training data, wherein

D(x) = 1 - and 2 + Ζ α where exp, ί ιί ——. |xf-xf 丫, “μ 2 s V ά J -2£a,exp

The SVE>D model uses: for the round-in variable x, to determine whether x falls within the range that can be explained by the past data; Z)(x)S〇 means applicable, _D(x)>〇No. Modeling process: 1) . Set parameters σ, C to verify all combinations in the range, select one of the groups &, C); 2) · Perform k-fold, randomly divide the steady-state experimental data into I Group, take 々1 group as model training data (marked as

ixU 'the other group plus transient data for testing (denoted as {X; H 3). For the training and verification combination of group a, solve the quadratic programming problem L, subject t0 ~ x') ,|2^2 Η, ^, >0, V/= is equivalent to solving its dual problem, 201223690

Max m 1 ~ Σ a>ai εχρ* f if I train v/rom| ' \x' ~xj 1 2 σ L , / 7) . Repeat steps 丨~6 for all parameter combinations to select the best performing parameters. Combine ((? e); 8) . Complete SVDD modeling. Referring to FIG. 7, after the machine reaches a steady state, prediction is performed, and the thermal error model construction unit 3 trains out with a support vector regression (SVR)-nonlinearity.

Σ«,=1> /=1 〇<α,. <C,V,= 1,...,W determines {xf} (the support vector 'SVs' is defined as 'the corresponding 0'), 7VSK (support vector number) and % 4). Select any support vector xfK of $^0, calculate subject to

R 1-2^]^. exp 7=1 σ ^ 1 ^2lajak exPi'- »1 5夂 "The test of jj double-into the model test, get a single 立 stand: a few points of the steady state mode D (x < 〇tp (true positive) type knot D(x) > 〇fn (false negative) fruit transient fp (false positive) R sy ν5Κ||λ2 X — σ material tn (true negative)

Fn + 2-tp + fp ). Repeat steps 3 to 5 for the different training and verification combinations of the * group, averaging all the errors, and defining the classification performance of F((T, c)q &amp; =l 201223690 Thermal error model, support vector regression (SVR) has fewer operational parameters, and in terms of prediction accuracy, a lower MAPE (mean absolute percentage error) can be obtained, and its core spirit is the kernel method. Support Vector Regression (SVR) is described in detail. SVR Objective: Estimate the function from the training data: μ = /(χ), where: + b /(x)=Z(a&lt;*_a&lt;)'exp&lt; In the formula, each training instance has its corresponding 6 and <; to determine whether the training instance can fall outside the scope of ε. The role of C is to adjust whether the training model has passed. Insufficient or insufficient information (overfitting or underfitting). When the core definition is clear, the following three parameters can be adjusted:

Gamma: Adjust the std of the Gaussian kernel function, which is σ in the above equation. C: The weighting measure used to adjust the error term during training can determine overfitting or underfitting.

Epsilon: The size of the error tolerance band, which is e in the above equation. The SVR model uses: For the input variable X, the value of the corresponding output variable /(4) is predicted. Modeling process: Ϊ)·Determining all combinations of parameters σ, c, and f in the range to be verified, select one of them (σ, 〇, £·); 2)·K-fold, randomize experimental data Divided into the sputum group, taking the moxibustion-1 group as the model training (recorded as 12 201223690 {(〇厂)丨) 3; one group for testing (denoted as min, the first 'and practiced , the certificate combination 'solution quadratic planning problem such as I Kw;

Jfain subject to ,

Max a;, % r, etc. "Price is solved by its dual question («;-«,) («*-α) εχρ (<w,&lt;,屮乂~Η·,ν/=1 一ζχΛΐ ^ 〇5 V/ = Title: Σ subject to m Σ(α; '〇:,) = 0 /=1 7 , mt, TOin — x her 丨丨, 2 σ £Σ(α&lt; + α&lt;)+ Σ )&gt; ί (a. + a ) /:1 1 , 疋{^{:} (the support vector is defined as "the corresponding a, 〆."), SK (the number of support vectors), "; and % ; ) Any ^ support vector X factory with 6=0 or 〇0, calculate έ, Vsv SV l!\2' &gt;*ι a;)*exp &lt;1 σ , sv iVsv Σ (pan / -%). Exp&lt; f\\xsv -.sy\\\2 -ni σ ε if 0&lt;〇:.&lt; C/ k-^s if 0&lt;a* &lt;C/ 5) , calculation error e_·, j| ;|/(xm|; 6) · Repeat steps 3 to 5 for a different set of training and verification combinations.

All errors 'define (4) (2) = |i>% is the prediction error of this group of parameter combinations; A = I). Repeat steps 1 to 6 for all parameter combinations, and select the parameter combination when the prediction error is the smallest (6) illusion; 13 201223690 8)·Complete SVR modeling. Referring to FIG. 8, the warm-up determination and thermal error calculation unit 4 breaks the warm-up state of the machine tool by reading a plurality of temperature sensing signals provided at each node and the information of the machine tool. Calculate at least one of the ''', and the error compensation amount before processing. He constructs a single picture 9, in the external environment temperature change, such as the machine tool moves to the basin-unit ^ operation, the incremental learning unit 5 can describe the thermal error support vector data description (SVDD) of the original data for the warm-up feature, and Support Vector Regression (SVR) receiving environment of the constitutive unit 3 (7) Knowing the pre-material, the warm-up judgment and the thermal error computing unit 4 and 2 are used as reference data to confirm whether to perform incremental learning to adapt to the government. π, to ensure the accuracy of the thermal error model. *, the second detailed operation process will be detailed later. Clear reference to &quot;Jg| 1. The step includes: the tool error thermal error adjustment method of the present invention, step S1: step S2: step S3: step S4: extracting thermal behavior data; selecting the warm state of the warm-up state by the expert; constructing the warm-up feature with the steady state temperature information description;

The warm-up feature description is used to filter the information that meets the steady state; Step S 5 · Construct the thermal error wedge by _ a with the thermal behavior information of the cattle. The second step S6 • The online warming machine # breaks and performs thermal error calculation v Township S7: Zeng Jinjue 曰 曰 予 ' ' 借 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应For the classification in the historical data 14 201223690, the warm-up feature description in step S3 is constructed by the support vector data description (SVDD), and the thermal error model of step S5 is constructed by support vector regression (SVR). The S6's online warm-up judgment is performed by reading information such as temperature sensing signals on the power tool and operating conditions of the machine tool. Please refer to FIG. 11 for description of the single-axis feed tool machine 11; temperature sensors T1 to D24 are disposed in each part of the machine tool, and the position of the branch is as shown in FIG. 12 to FIG. 18; the thermal behavior information extraction plan is as follows: Itinerary planning: full stroke ll〇〇mm operation. Feed speed gauge: Four feed speeds of 9m/min, 18m/min, 27m/min and 36m/min. Measuring time gauge: In order to facilitate analysis of the system's transient and steady-state thermal error behavior mode, it is measured every 15 minutes after power-on, and the transient information is retrieved. After a long time (in 150 minutes as an example), Change it to measure every 30 minutes and retrieve the steady state information of the system. Please refer to FIG. 19, which is a flow chart showing a method for extracting thermal behavior data. The step of the thermal behavior data extraction method of the present invention comprises: Step SA1: selecting the feed speed; Step SA2: setting the machine tool parameters and setting the temperature sensor at several nodes of the machine tool; Step SA3: the machine tool is turned on and enters the operation mode Step SA4: Whether the operation mode reaches 150 minutes (predetermined operation time), if not, enter every 15 minutes (first time interval) 15 201223690 If yes, then enter each SA42); - time interval) measurement mode ( Step SA5: Perform temperature measurement on each node of the work and her AA step SA6: store ^; and program 2 = SA6 can manually fill in the form or use the temperature to please refer to the figure U~18, the tool track 13, the servo The motor 14 includes a saddle 11, a screw 12, a slippery thumb, a seat 15, a foot screw 16, and the temperature variable includes an external heat source (a record of the temperature variable of the number of turns, a ugly degree) and the inside. The heat source, so for the twenty-four structures, the external environment temperature, the degree includes: five _ seat u ^ 4 ^ - fourteen structure body temperature π ghost know, the degree, a nut seat (screw 12 at W Degree, eleven The base 15 has two sides ((4) 13) temperature, four bottoms f 15": foot end structure (foot screw 16) temperature, one motor seat bearing end temperature, - tailstock bearing end temperature, and a motor interface seat temperature , that is, at each node of the fourteen temperature sensors T1 to T24; in addition, according to the thermodynamics, the temperature change rate (derivative) should be included in the reference. The thermal clamp correlation variable is based on the vehicle. The stroke is planned. In order to match the position of the nodes in the segmentation model constructed by the virtual sensing, the experiment sets the measurement of the positioning accuracy at intervals of 125 mm. The measurement positions (mm) are: 131, 256, 381, 506, 631, 756, 881, and 1006, as shown in Fig. u. The present invention uses support vector regression (SVR) as a thermal error prediction model to construct a machine body temperature distribution vector with support vector data description (SVDD). 16 201223690 The range of evil modes is used to define the applicability of the thermal error prediction model. SVDD is a novel machine learning algorithm whose purpose is to construct a minimum hypersphere (minimum-volum) e hypersphere) to surround the training set. Since the hypersphere is constructed in the feature space, the surface of the sphere becomes very elastic in the input space. For a new machine tool body temperature distribution vector variable, we only need to calculate it. The distance to the center of the sphere can be used to calculate the confidence index of the warm-up state corresponding to the input: the greater the distance, the greater the difference between the input and the training set, the smaller the output φ confidence index . There are two advantages to using SVDD: 1) its solution does not have a local minimum problem, because its dual problem is also a QP problem, so there is no need to manually set the threshold, and 2) regardless of the distribution of the input data training set, With SVDD you can find a boundary that can tightly surround it. Its fine technical details have been revealed as before. Through the above structure and method 'acquiring the thermal displacement behavior data through the temperature sensors set in each part of the machine tool, the steady state mode support vector regression (SVR) thermal error model and the support vector data are described to describe the hypersphere, the energy φ The warm-up state is judged under the variation of the ambient temperature, and the compensation information is directly predicted on the line to compensate the thermal error. Although the present invention has been explained in connection with the preferred embodiments, it is not intended to limit the invention. It should be noted that many other similar embodiments can be constructed in accordance with the teachings of the present invention, and are within the scope of the present invention. 17 201223690 Figure l Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 1 Figure 11 [Simple diagram of the diagram] A schematic diagram of the thermal error of a conventional tool machine. Another schematic diagram of the thermal error of the conventional machine tool is shown. The curve circle of the thermal error of the tool machine is not known. It shows a graph of the conventional tool machine to judge the thermal error by the operator. The figure showing the thermal error intelligent adjustment device of the machine tool of the present invention represents the last name of the thermal error intelligent adjustment device of the machine tool of the present invention. The data describes a schematic representation of the steady state description. The vector represents the graph of the steady-state prediction error amount of the tooling machine thermal error smart adaptation device of the present invention. ...set = this. The Japanese version of the machine is equipped with a thermal error wise 慧 慧 (4), the judgment of the operation of the warm error calculation unit. table. Show this day red machine thermal error wisdom adjustment device t incremental learning early yuan judgment diagram. A flow chart of the method for adjusting the thermal error of the Japanese red machine. Figure no. ^The stroke planning machine shown as an example of the stroke feeding machine is shown in the figure; I show the structure diagram of the uniaxial feeding machine as an example. In the early axis feed guard machine, the temperature sensor is arranged. The temperature sensor is arranged in the single-axis feed tool machine. In the single-axis feed guard machine, the temperature sensor is distributed. &quot; Ding Ben Mouyue distributes the temperature sensor in the single-axis feed tool machine 18 201223690 Layout 4. Fig. 17 is a view showing the fifth layout of the present invention for arranging a temperature sensor in a single-axis feed tool. Fig. 18 is a view showing the distribution of the temperature sensor of the present invention in a single-axis feed tool machine. Fig. 19 is a flow chart showing the steps of extracting heat behavior in the thermal error intelligent adjustment method of the machine tool of the present invention. [Main component symbol description] 1 Tool machine thermal error intelligent adjustment device 10 Machine tool 11 Saddle 12 Screw 13 Slipper 14 Servo motor 15 Base 16 Foot screw 2 Warm-up feature description Construction unit 3 Thermal error model construction unit 4 Warm-up Judgment and thermal error computing unit 5 Incremental learning unit a~d Nodes T1 to T24 Temperature sensor steps S1 to S7 The machine tool thermal error smart adaptation method according to the present invention 19 201223690 Steps SA1 to SA6 According to the extraction thermal behavior of the present invention Step 20

Claims (1)

201223690 VII. Scope of application for patents: In the machine tool package: the tooling machine thermal error wisdom adjustment skirt, set in the machine to describe the construction unit, establish the work, the degree of steady-state mode of the knife cloth vector; type; T-differential model Kawasaki, T-Korean thermal error mode, a warm-up judgment and thermal error calculation unit, a number of temperature sensing signals and the machine tool ";: cookware: with machine warm-up state, at least the difference: " The tool machine thermal error wisdom adapts to ensure that the thermal two can be adapted to the parameter adjustment, and the maintenance of the warm-up feature is described in (4). The steady-state mode range. Line Bai Libei Ke (4) 1 item of the machine tool thermal error wisdom adjustment = the thermal error model construction unit to support vector regression (SVR) training, east out of the nonlinear thermal error model. The utility model relates to the machine tool thermal error intelligent adjustment device 6 of the patent scope. [The various components on the machine tool are provided with a plurality of temperature sensors. The tooling machine thermal error wisdom adapts its method, and the steps thereof include Extracting thermal behavior data; Experts select warm-up state of the warm-up; 21 201223690 Constructing a warm-up feature description with steady-state temperature information; Filtering the hot-line with stable state by a warm-up feature description to construct a thermal error model γ ° On-line warm-up judgment and thermal error operation. The work described in item 7 of the “thermal error wisdom adjustment method μ _^= learning” is adapted to the environmental change by parameter drift. The tooling machine thermal error intelligent adjustment square m, the thermal behavior includes at least the temperature and positioning error. The profit range: the tooling machine thermal error intelligent adjustment method described in item 7 / Ί '$ extraction hot behavior data step further includes: The feed rate is selected; the tester sets the machine tool parameters and sets the temperature sense tool on several points of the machine tool to start up and enter the operation mode; - the first transport reaches a predetermined running time, if not, the first time interval is entered. Measurement mode; if yes, enter one;: separate measurement mode; wherein 'the first time interval is less than the second time to warm the nodes of the machine tool Measurement; and storage of data. Fang, go to Shen ° 1 1 special tool for the machine tool thermal error as described in the first paragraph of the Liganwei wisdom adjustment is 4 == save the data step including filling out the form manual record or X / 撷 撷The program automatically records. 22