【摘要】：建立精確的數(shù)控機(jī)床誤差模型是對(duì)機(jī)床誤差修正的前提和基礎(chǔ)。數(shù)控機(jī)床在加工過(guò)程中產(chǎn)生的幾何誤差與所承受的切削力有關(guān),而大部分已有的數(shù)控機(jī)床幾何誤差建模方法沒(méi)有考慮切削力的影響。數(shù)控機(jī)床光柵測(cè)量系統(tǒng)在傳統(tǒng)的固定方式下,機(jī)床床身的熱變形會(huì)通過(guò)光柵固定螺釘傳遞到光柵測(cè)量系統(tǒng)上,對(duì)光柵測(cè)量系統(tǒng)零位誤差和示值誤差產(chǎn)生附加影響,很難通過(guò)誤差建模補(bǔ)償理論進(jìn)行精確補(bǔ)償。 為了提高數(shù)控機(jī)床幾何誤差建模精度和補(bǔ)償效果,本文通過(guò)測(cè)力環(huán)等儀器模擬施加和測(cè)量機(jī)床主切削力,采用激光干涉儀同步測(cè)量機(jī)床俯仰角和偏擺角誤差,應(yīng)用粒子群優(yōu)化算法(PSO)優(yōu)化支持向量機(jī)(SVM)的相應(yīng)參數(shù),利用實(shí)際測(cè)量數(shù)據(jù)進(jìn)行訓(xùn)練,建立了基于PSO-SVM的力-幾何誤差預(yù)測(cè)模型。 本文根據(jù)形體熱變形理論,提出了一種基于熱變形臨界點(diǎn)的數(shù)控機(jī)床光柵測(cè)量系統(tǒng)新的固定方式。定義了數(shù)控機(jī)床三維熱變形臨界點(diǎn),利用有限元仿真分析對(duì)三軸數(shù)控機(jī)床仿真模型進(jìn)行分析,確定了機(jī)床不同方向床身的熱變形臨界點(diǎn),作為該方向上的光柵測(cè)量系統(tǒng)的固定點(diǎn),使床身熱變形無(wú)法傳遞至光柵上。光柵兩端采用浮動(dòng)支撐的方式使其只能在測(cè)量方向自由變形,這樣就減小甚至消除數(shù)控機(jī)床床身的熱變形對(duì)光柵測(cè)量系統(tǒng)產(chǎn)生的附加影響。最后根據(jù)光柵測(cè)量系統(tǒng)本身熱變形理論和上述光柵固定方法,建立了光柵測(cè)量系統(tǒng)零位誤差和示值誤差精確預(yù)測(cè)模型,用于后續(xù)的機(jī)床熱誤差綜合補(bǔ)償。為了提高數(shù)控機(jī)床各類誤差的綜合補(bǔ)償效果,本文建立了數(shù)控機(jī)床綜合誤差數(shù)學(xué)模型,為提高數(shù)控機(jī)床的加工精度提供了很好的理論基礎(chǔ)。 通過(guò)實(shí)際試驗(yàn)驗(yàn)證,所建立的PSO-SVM誤差預(yù)測(cè)模型輸出的偏擺角誤差預(yù)測(cè)值與實(shí)際測(cè)量數(shù)據(jù)的最大差值僅為0.6μrad,俯仰角誤差預(yù)測(cè)值與實(shí)際測(cè)量數(shù)據(jù)的最大差值僅為0.21grad,遠(yuǎn)遠(yuǎn)小于利用BP神經(jīng)網(wǎng)絡(luò)以及常規(guī)方法優(yōu)化的SVM所建立的力-幾何誤差預(yù)測(cè)模型的建模誤差,因此所建立的誤差預(yù)測(cè)模型可以用于后續(xù)數(shù)控機(jī)床幾何誤差的高精度實(shí)時(shí)補(bǔ)償。
[Abstract]:Accurate error model of NC machine tool is the premise and foundation of error correction. The geometric error produced by NC machine tool in machining process is related to the cutting force, but most of the existing modeling methods of NC machine tool geometric error do not consider the influence of cutting force. In the traditional fixed mode, the hot deformation of the machine bed will be transferred to the grating measuring system through the grating fixed screw, which has additional influence on the zero position error and the indication error of the grating measuring system. It is difficult to accurately compensate by error modeling compensation theory. In order to improve the modeling accuracy and compensation effect of NC machine tool geometric error, this paper simulates and measures the main cutting force of the machine tool by means of force measuring loop, and uses laser interferometer to measure the error of pitch angle and deflection angle of machine tool synchronously. The particle swarm optimization algorithm (PSO) is applied to optimize the parameters of support vector machine (SVM). The force-geometry error prediction model based on PSO-SVM is established by training the actual measurement data. Based on the theory of body thermal deformation, a new fixed method of grating measuring system for CNC machine tool based on critical point of thermal deformation is proposed. The critical point of 3D thermal deformation of NC machine tool is defined. The simulation model of 3-axis NC machine tool is analyzed by finite element simulation analysis, and the critical point of hot deformation of machine tool in different directions is determined. As the fixed point of the grating measuring system in this direction, the hot deformation of the bed can not be transferred to the grating. The two ends of the grating can only deform freely in the measuring direction by means of floating support, which reduces or even eliminates the additional influence of the hot deformation of the bed of the NC machine tool on the grating measurement system. Finally, according to the thermal deformation theory of the grating measurement system and the grating fixing method mentioned above, an accurate prediction model of the zero error and the indication error of the grating measurement system is established, which can be used to compensate the thermal error of the subsequent machine tool. In order to improve the comprehensive compensation effect of various kinds of errors of NC machine tools, the mathematical model of synthetic errors of NC machine tools is established in this paper, which provides a good theoretical basis for improving the machining accuracy of NC machine tools. Experimental results show that the maximum difference between the predicted value of deflection angle error of the established PSO-SVM error prediction model and the actual measurement data is only 0.6 渭 rad, pitch angle error prediction value and the maximum difference value of the actual measurement data is only 0.21grad. It is much smaller than the modeling error of the force-geometry error prediction model established by the BP neural network and the SVM optimized by the conventional method, so the established error prediction model can be used to compensate the geometric error of the NC machine tool with high precision and real time.
【學(xué)位授予單位】：安徽理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：TG659

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 傅建中;姚鑫驊;賀永;沈洪W，

本文編號(hào)：2367622