功能材料
功能材料
공능재료
JOURNAL OF FUNCTIONAL MATERIALS
2012年
8期
1079-1083
,共5页
周春华%王帮峰%刘曌%陈珩%牟常伟
週春華%王幫峰%劉曌%陳珩%牟常偉
주춘화%왕방봉%류조%진형%모상위
主动变形%波纹型蒙皮结构%形状记忆合金增强复合材料%人工神经网络
主動變形%波紋型矇皮結構%形狀記憶閤金增彊複閤材料%人工神經網絡
주동변형%파문형몽피결구%형상기억합금증강복합재료%인공신경망락
active deformation%rippled skin structure%shape memory alloy reinforced composites%artificial neural network
为满足变体机翼蒙皮连续光滑以及大变形要求,提出了一种可主动变形的SMARC波纹型蒙皮结构,并制备了两种蒙皮实验样件,在不同环境温度下,进行了拉伸与驱动实验。拉伸实验结果表明SMARC样件拉伸变形性能优于不含SMA丝样件,但SMARC材料受温度影响较大,温度升高,结构刚度减小;驱动实验中,初始位移线性增加,当激励电流超过3.4A时,复合材料基体受热软化,导致位移呈非线性变化趋势。最后在实验基础上建立了一个以电流强度为输入参量,驱动位移为输出参量的RBF神经网络模型,其逼近曲线较为准确和贴近实际驱动特性,最大预测相对误差〈6%,为智能蒙皮的主动控制研究提供了一定的理论及实验依据。
為滿足變體機翼矇皮連續光滑以及大變形要求,提齣瞭一種可主動變形的SMARC波紋型矇皮結構,併製備瞭兩種矇皮實驗樣件,在不同環境溫度下,進行瞭拉伸與驅動實驗。拉伸實驗結果錶明SMARC樣件拉伸變形性能優于不含SMA絲樣件,但SMARC材料受溫度影響較大,溫度升高,結構剛度減小;驅動實驗中,初始位移線性增加,噹激勵電流超過3.4A時,複閤材料基體受熱軟化,導緻位移呈非線性變化趨勢。最後在實驗基礎上建立瞭一箇以電流彊度為輸入參量,驅動位移為輸齣參量的RBF神經網絡模型,其逼近麯線較為準確和貼近實際驅動特性,最大預測相對誤差〈6%,為智能矇皮的主動控製研究提供瞭一定的理論及實驗依據。
위만족변체궤익몽피련속광활이급대변형요구,제출료일충가주동변형적SMARC파문형몽피결구,병제비료량충몽피실험양건,재불동배경온도하,진행료랍신여구동실험。랍신실험결과표명SMARC양건랍신변형성능우우불함SMA사양건,단SMARC재료수온도영향교대,온도승고,결구강도감소;구동실험중,초시위이선성증가,당격려전류초과3.4A시,복합재료기체수열연화,도치위이정비선성변화추세。최후재실험기출상건립료일개이전류강도위수입삼량,구동위이위수출삼량적RBF신경망락모형,기핍근곡선교위준학화첩근실제구동특성,최대예측상대오차〈6%,위지능몽피적주동공제연구제공료일정적이론급실험의거。
In the research of variant skinned wing, a type of SMARC rippled skin structure with active deforma- tion was proposed at first for meeting requirements of continuous smooth and large deformation. And then, two kinds of skin testing samples were made and tested for stretch and driving ability in different environment tem- perature. Results of tensile test show that SMARC rippled skin test sample has better stretching deformation performance than the ordinary rippled skin type. However, the temperature has more influence on SMARC rip- pled skin test samples. When the temperature is higher, structure stiffness of SMARC rippled skin test sample is smaller. In the driving test, the initial displacement of SMARC rippled skin test sample has linear increase. When the drive electric current is more than 3.4A, matrices of composite material are heated to soften. As the result of that, the displacement has a nonlinear variation trend. At last, on the basis of driving test data, a type of RBF neural network model was established where current strength was input parameter and the driver dis- placement was output parameter. The approximation curve of this neural network model is more accurate and close to the practical characteristics, and the biggest driver relative prediction error is less than 6 %, which provides for active control of intelligent skin with a certain theoretical and experimental basis.