| 扬州漆木柜典型度预测方法 |
| |
| 作者姓名： | 方方 关惠元 卢章平 李明珠 |
| |
| 作者单位： | 南京林业大学家居与工业设计学院;扬州工业职业技术学院艺术设计学院;江苏大学艺术学院 |
| |
| 基金项目： | 江苏省高校哲学社会科学研究项目(2018SJA1081);江苏高校优势学科建设工程资助项目(PAPD)。 |
| |
| 摘 要： | 为设计出具有传承性的新中式漆木家具,以扬州漆木柜为例开展典型度预测研究,提出基于支持向量分类机的典型度预测方法。首先,综合应用扎根理论、德尔菲法和设计形态分析法提取出扬州漆木柜的9个典型特征作为典型度评价指标;其次基于语义差异法进行典型度调查,获取典型度分级值;最后分别采用BP神经网络和支持向量分类机作为典型度预测方法,建立"典型特征-典型度"之间的映射关系,构建基于典型特征的扬州漆木柜典型度分级预测模型。在此基础上,利用混淆矩阵比较2种模型的预测精度,结果表明支持向量分类机构建模型的拟合精度和泛化精度分别为95.5%和94.0%,高于BP神经网络构建模型的对应精度值77.3%和90.0%,且支持向量分类机构建模型的F1值无论在训练分析中还是在预测分析中均大于或等于BP神经网络构建模型,预测效果显著占优。在实际应用层面,典型度是个隐性的模糊感觉量,建立扬州漆木柜典型度预测模型,不仅可以为设计师设计具有传统特征基因的扬州漆木家具提供科学指导,也能协助企业降低人工评估误差和创新风险,有效缩短设计周期。在理论层面,承继对象的新选择和细化以及定量研究方法的引入,为新中式家具研究提供了一个全新的研究思路和研究方法。
|
| 关 键 词： | 扬州漆木柜 新中式 典型度预测 支持向量分类机 BP神经网络 |
Typicality prediction method study of Yangzhou lacquered cabinets |
| |
| Authors: | FANG Fang GUAN Huiyuan LU Zhangping LI Mingzhu |
| |
| Institution: | (College of Furnishings and Industrial Design,Nanjing Forestry University,Nanjing 210037,China;Art Design College,Yangzhou Polytechnic Institute,Yangzhou 225127,Jiangsu,China;Art School of Jiangsu University,Zhenjiang 212013,Jiangsu,China) |
| |
| Abstract: | Aimed at playing a helpful role for the traditional design of new Chinese style lacquered furniture,this essay proposed typicality prediction methods of support vector classification(SVC) by taking cabinets of this regard as an example.Firstly,nine typical characteristics of Yangzhou lacquered cabinets were extracted through the integrated application of grounded theory(GT),Delphi method as well as design format analysis(DFA) before these nine characteristics were utilized as the evaluation criteria of typicality;secondly,grading values of typicality were obtained through the typicality investigation based on semantic differential(SD) method;lastly,by respectively adopting back propagation neural network(BPNN) and SVC as typicality prediction methods,the mapping relation between "typical characteristics-typicality" was constructed to build the typicality grading prediction model for Yangzhou lacquered cabinets based on typical characteristics.Based on the above,the prediction accuracy of these two models was compared by using confusion matrix.Results had shown that the fitting precision and generalization precision of models constructed by SVC were 95.5% and 94.0% while those built by BPNN were 77.3% and 90.0%.It had also been discovered that the F1-Score of models constructed by SVC were no lower than that of BPNN no matter in training analysis or in prediction analysis,showing apparently better prediction power.In the practical application,typicality is an implicit factor that is usually perceived in a vague manner.The typicality prediction model for Yangzhou lacquered cabinets constructed in this study can not only provide scientific guidance for designers in making other Yangzhou lacquered furniture with conventional genes,but also be of help with enterprises in reducing human error in evaluation and risks in innovation,and thus effectively shortening design cycle.In the theoretical research,through the re-selection of inheriting objects and the introduction of quantitative research methods,this essay provided a new research idea and method for the study of new Chinese style furniture. |
| |
| Keywords: | Yangzhou lacquered cabinet new Chinese style typicality prediction support vector classification back propagation neural network |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
