發(fā)布時(shí)間：2018-11-09 11:28

【摘要】：工藝-組織-性能之間的關(guān)系始終是材料科學(xué)研究的主題。長(zhǎng)期以來(lái),對(duì)材料研究采用的是依賴大量實(shí)驗(yàn)、或傳統(tǒng)的試錯(cuò)法等經(jīng)驗(yàn)或半經(jīng)驗(yàn)的材料研究方法,這需要消耗大量的人力、物質(zhì)資源和時(shí)間。隨著計(jì)算機(jī)技術(shù)的發(fā)展,人們逐漸將人工智能技術(shù)應(yīng)用到材料研究中,試圖通過(guò)較少的實(shí)驗(yàn)獲得較為理想的結(jié)果。人工神經(jīng)網(wǎng)絡(luò)和遺傳算法是人工智能領(lǐng)域技術(shù)純熟,且迄今為止應(yīng)用較多的兩種關(guān)鍵技術(shù)。將這兩種算法融合,既可以發(fā)揮各自的優(yōu)點(diǎn),又可以避免兩者單獨(dú)使用的固有缺點(diǎn),從而使對(duì)工藝的建模更精確有效。本文主要對(duì)人工智能領(lǐng)域的兩個(gè)熱點(diǎn)問(wèn)題——人工神經(jīng)網(wǎng)絡(luò)和遺傳算法進(jìn)行分析和改進(jìn),嘗試將這兩種算法融合,并運(yùn)用到羥基磷灰石-C/C復(fù)合材料的感應(yīng)熱沉積工藝和芳綸纖維化學(xué)鍍鎳工藝優(yōu)化中。論文取得以下主要研究成果:1、設(shè)計(jì)了兩種BP神經(jīng)網(wǎng)絡(luò)與遺傳算法融合的方法:基于遺傳算法的BP神經(jīng)網(wǎng)絡(luò)優(yōu)化方法以及基于神經(jīng)網(wǎng)絡(luò)和遺傳算法的非線性函數(shù)極值尋優(yōu)方法。2、基于神經(jīng)網(wǎng)絡(luò)和遺傳算法的非線性函數(shù)極值尋優(yōu)方法,對(duì)C/C復(fù)合材料表面制備羥基磷灰石工藝進(jìn)行了建模,獲得最佳工藝條件,并對(duì)感應(yīng)熱沉積工藝的熱力學(xué)和動(dòng)力學(xué)的研究具有一定的指導(dǎo)意義。3、基于神經(jīng)網(wǎng)絡(luò)和遺傳算法的兩種融合算法,對(duì)芳綸纖維表面化學(xué)鍍鎳工藝進(jìn)行了建模,獲得最佳工藝條件:鍍液中檸檬酸鈉濃度為12.00 g/L,氯化銨濃度為24.00 g/L,次亞磷酸鈉濃度為28.00 g/L,p H值為9.19,溫度為50℃,化學(xué)鍍鎳的沉積速率為1.91 g/(g?min)。
[Abstract]:The relationship between process-structure and properties has always been the subject of material science research. For a long time, the material research is based on a large number of experiments, or the traditional trial and error methods and other empirical or semi-empirical material research methods, which need to consume a lot of manpower, material resources and time. With the development of computer technology, artificial intelligence technology has been gradually applied to material research, trying to obtain ideal results through fewer experiments. Artificial neural network (Ann) and genetic algorithm (GA) are two key technologies in the field of artificial intelligence. The fusion of the two algorithms can not only give full play to their respective advantages, but also avoid the inherent disadvantages of using the two algorithms separately, thus making the process modeling more accurate and effective. In this paper, two hot issues in artificial intelligence field, artificial neural network (Ann) and genetic algorithm (GA), are analyzed and improved. It was applied to the optimization of induction thermal deposition process of hydroxyapatite / C / C composite and electroless nickel plating of aramid fiber. The main research results are as follows: 1. Two methods of fusion of BP neural network and genetic algorithm are designed: BP neural network optimization method based on genetic algorithm and nonlinear function extremum optimization method based on neural network and genetic algorithm. Based on the nonlinear function extremum optimization method based on neural network and genetic algorithm, the preparation process of hydroxyapatite on the surface of C / C composite is modeled and the optimum process conditions are obtained. It has certain guiding significance for the study of thermodynamics and kinetics of inductive thermal deposition. 3. Based on two fusion algorithms of neural network and genetic algorithm, the electroless nickel plating process on the surface of aramid fiber is modeled. The optimum process conditions were obtained as follows: sodium citrate concentration was 12.00 g / L, ammonium chloride concentration was 24.00 g / L, sodium hypophosphite concentration was 28.00 g / L ~ (-1) H value was 9.19, and temperature was 50 鈩，

本文編號(hào)：2320274