The computer-aided design method of cabinet based on style imagery

（School of Mechanical Engineering，Southeast University，Nanjing 211189，China）

The computer-aided design method of cabinet based on style imagery

Shen Zhangfan Xue Chengqi Wang Haiyan Niu Yafeng Shao Jiang Zhang Jing

（School of Mechanical Engineering，Southeast University，Nanjing 211189，China）

Due to the practical problems of the high costs and the long development cycle of China’s cabinet production，a computer-aided design method of the cabinet based on style imagery is proposed．According to the principle of the conjoint analysis method，the rough set theory and the weight coefficient of different components of the cabinet，a multidimensionalmodel of style imagery to evaluate the cabinet is built．Then the related constants of style imagery are calculated and the cabinet components library is also built by the three-dimensional modeling．Finally，with recombinant technology and the mapping model between cabinet style and external characteristics，the prototype system based on Visual Studio is proposed．This system actualizes the bidirectional reasoning between product style imagery and the shape features，which can assist designers to producemore creative designs，greatly improve the efficiency of cabinet development and increase the profits of companies．

cabinet；computer-aided design；style imagery；component recombinant；shape features

In recent years，w ith the development of industry technology，the demand of industry cabinets is boom ing．The cabinetmarket in China has reached an unprecedented stage of development．However，the domestic potential market is dom inated by many famous international cabinet brands，such as Rittal，APC，HP and IBM．The main reason is that domestic cabinet manufacturers lack advanced design concepts，design styles and deep understanding of the design process．Therefore，how to effectively change the status quo，strengthen the overall design ability and improve the efficiency of cabinet production have become urgent issues for these domestic cabinet manufacturers at present．

The cabinetmodeling design uses principles and rules of the space design to process all kinds of product structures，functions，materials and the relationships between humans and the environment．However，most cabinet designers design products only according to their experience because they lack theoretical support．Besides，there are few studies on the product features of cabinets．Cheng［1］analyzed the framework of cabinet case library by his indepth know ledge of cabinet products and tried to create an aided design system to facilitate the improvement of designing industrial cabinets．Han［2］summarized themodeling characteristics of cabinets in three different stages and proposed a theory model for humanization design of cabinet by combining humanized product design concepts．However，these studies only focused on designers’concepts and they were not comprehensive and systematic，which caused a failure to establish image mapping between cabinet characteristics and styles．

As an importantaspect of emotional cognition，product style imagery becomes a key factor for distinguishing the consumermarkets，and therefore it is of great significance to the design of industrial products．The perceptual technology based on the semantic differencemethod has been w idely used in recent years．M cCormack etal．［36］carried out the detailed analysis of the style of different kinds of industrial products．Dem irbilek et al．［7］discussed the key role of product characteristics and users’emotions during the process of design．Chen et al．［8］suggested that the product style should bemade up of a series ofmodel elements through differentmethods of special forms which was themain carrier of the mental function of the product．Huang et al．［910］studied the application of perceptual technology in a computer-aided product design system．Zhu et al．［11］proposed a new method which overcame problems of the old interior design method and improved the validity of the interior design．Pan et al．［12］developed a prototype system about self-adaptive product design on modularity．Potocnik et al．［13］presented a know ledgebased system which is capable of giving the designer high-quality supportwhen making decisions from the aspect ofmodeling the reinforcement of a plate-pressw ithin a position of themaximum compressive load．

1 Related W ork

In this paper，the characteristics-style mapping mechanism and the reasoningmethod based on style imagery of cabinet are studied．Then，the reference basis for cabinet design and technical support are proposed．Finally，thecabinet CAD prototype system based on the image style is created．This system can enhance the dynam ic and continuous innovation of the cabinets，which has practical guiding significance and application value to the product design research and development of the cabinet enterprise．Based on people’s cognitive principles，themethod of cabinet style imagery modeling design takes user perceptual imagery semantic information into the quantization scheme to develop conceptual design，considering exteriormodeling as themain object．Fig．1 illustrates the research framework．

Fig.1 Research framework

1）Cabinet samples and related imageries are collected and organized．Cabinet samples are collected from the Internet and figure publications．Corresponding imagery linguistic variables are obtained through questionnaires conducted by cabinet designers and users．The typical cabinet samples are achieved by cluster analysis，and the related imagery linguistic variables are further screened in order to obtain the adjectives，which can correctly reflect cabinet perceptual cognition as far as possible．

2）The chosen imagery phrases and representative cabinet samples are performed in the imagery evaluation experiment by the semantic differencemethod．Factor analysis is implemented to the evaluated results in order to select several representative imagery semantics，which w ill make up the cabinet imagery semantic set A．In addition，cabinet designers and experts are invited to analyze the appearance of typical cabinet samples and decompose the cabinetmodeling characteristics in order to establish the modeling feature set B．

3）Image semantic set A and themodeling feature set B obtained from the second stage are performed in the semantic difference evaluation experiment，in which the data is carried in the quantification analysis．Finally，the linear relationship between semantic and cabinetmodeling characteristics is obtained．

2 M ethod

2.1 The description of the design style

Generally speaking，users are unable to make a direct and specific description of cabinet products by imagery perception．This study picks outmore than 100 linguistic variables，which can be used to describe the cabinet products．Then，30 subjects（including 20 people w ith solid design background and 10 people w ithout design background）are invited to take part in the questionnaire experiment．They are required to choose the most suitable linguistic variables to describe the perceptual imagery semantics of cabinet products．The selected 15 pairs of linguistic variables are as follows：modern-traditional；simple-complex；individual-public；light-heavy；technological-fundamental；expensive-cheap；elegant-rugged；static-dynam ic；ordinary-luxury；plump-lean；rational-emotional；rigid-pliable；graceful-tedious；deluxe-elementary and m inimalist-mutative．

More than 120 pictures of cabinet samples produced by domestic or foreign companies are collected．A fter the discussion among the designers w ith solid cabinet design background，the sim ilar samples are removed．As a result，39 representative samples are prelim inarily screened out．

2.2 Rep resentative sem antics selection

30 subjectsmeasure the 39 cabinet samples w ith the 15 linguistic variables by the subjective evaluation method（The value is from－2 to 2）．Correlation analysis is carried outon the15 linguistic variables by SPSS．The result shows that the semantic correlation coefficients of simplecomplex，ordinary-luxury and m inimalist-mutative are extremely high，and the coefficients of other linguistic variables are very low．Therefore，these three linguistic variables are separated and the remaining 12 pairs of linguistic variables are analyzed by the factor analysismethod．The value of KMO（Kaiser-Meyer-Olkin）test is 0．664，which means that the results conform to the requirement of the factor analysis．The principal componentanalysis is used to analyze the average value of the evaluated 12 linguistic variables．Result shows that factor 1，2 and 3 can explain 97．602%of the total variance of the original variables，so it is rational to use the factor analysismethod to analyze the data．

The data is analyzed by themethod of orthogonal rotation．Tab．1 shows that the five linguistic variables（light，rational，lean，rigid，static）have a high load onthe first factor，which can be defined as the feeling of being light．Another four linguistic variables（smooth，expensive，luxury，graceful）can be defined as the feeling of elegance and the three linguistic variables（modern，individual，technological）can be defined as the feeling of modernity．The last three linguistic variables（simple，ordinary，m inimalist）can be classified into one category，defined as the feeling of simpleness．

Tab.1 Componentmatrix of the 12 linguistic variables

2.3 Shape features reduction

Fig.2 Cabinet experimental samples

Fig．2 shows the 12 typical cabinets selected as experimental samples．In order to investigate the most important shape features of cabinet，10 senior cabinet designers conductmorphological analysis of these sam ples by the Delphimethod．They extract and organize as many as possible of the morphological features of cabinet．In the first round，designers listmore than one hundred morphological features to describe cabinets，such as square frame，round handle，thick base and so on，then they screen them into 20 typical features．Tab．2 shows the final 8 typical features extracted by the second round of morphological analysis．Tab．3 also shows the feature-style decision table by using the data discretization method．

Tab.2 Feature-attribute table of cabinet

Tab.3 Features-style decision table of the 12 cabinet samples shown in Fig．2_______________________________________

Finally，we determ ine some key features of the cabinet appearance by the rough set theory．The definitions of redundancy，attribute reduction and attribute core in the rough set theory are as follows：

Definition 1 For the decision system S＝（U，CD，V，f），set c0∈C，if PC（D）＝P（c＼｛c0｝）（D），c0is the necessary attribute of D．If each c0in C is the necessary attribute of D，then C is the irreducible set of D．The set of irreducible attributes of D in C is called as the D core of C，marked as c′D（C）．Set C′?C，C′≠?，if①PC＝PC′（D）；②C′is the irreducible set of D，and it is called as the reduced set of C，marked as rD（C）．

Definition 2 The significance of C′for D is defined asαCD（C′），αCD＝［c（PC（D））－c（P（C－C′））（D）］／c（U）．Based on Definition 1，Ohrn［14］proposed a genetic algorithm for attributes reduction，f（B）＝（1－β）·

is the unrecognizable set in the decision table；βis the weight ratio between the loss reduction and them inimum classification rate；B is the subset of A，obtained by the evolution search algorithm；function cos t is the data actualizing correct classification；εis the threshold value of the classification rate．

Using the method of the genetic algorithm，we delete feature attributes one by one to achieve the goal of subtracting the feature attribute，and obtain the set of the most significant（irreducible）features．Setα＝0．4，ε＝1．0，and the set A＝｛framework，panel，handle and base｝is obtained；therefore，the cabinets can be decomposed into four different components as framework，insert box panel，handle and base．

2.4 Cabinet appearance and style imagery mapping

W ith the method of conjoint analysis，the impact of style imagery of various components on the whole cabinet can be quantized as the weight coefficient and the mapping model between modeling features，and style imagery can be established．30 subjects w ith solid design background are invited to conduct SD evaluation experiment w ith a 5-point scale method．Tab．4 shows the values of different features on different style semantics．

Tab.4 Values of four different features of 12 typical samples on four style semantics

3 Results

According to the principle of the conjoint analysis method，the score of the whole cabinet Y can be expressed as the multi-dimensional linear model composed of independent variables of components scores：

where A is the value of frame；B is the value of panel；C is the value of handle；D is the value of base．Tab．5 shows the results of mutiple linear regression analysis．W ith the weight coefficient of different parts of the cabinet in different style imageries and the constants shown in Tab．5，the calculation formulae of style imagery values can be obtained（see Tab．6）．

Tab.5 Results ofmultiple linear regression analysis

Tab.6 Formulae of different style imagery values

Fig．3 shows the cabinet computer-aided design system established by using Visual Studio．It consists of the calculation and recognition of cabinet style imagery，alternative selection and edition of the cabinet samples，presentation of 3Dmodels and representative values．The libraries of style imagery and product characteristics play a key role in the whole system．The specific steps are as follows：

1）Select different imagery variables and give each variable corresponding value，and the system w ill present a series of eligible alternatives．

2）Select an alternative，and the 3D cabinetmodelw ill be presented．

3）Click the“Edit”button，the four components can be replaced by different options，while the four imagery sores w ill change accordingly．

4）Click the“Picture”and“Model”buttons to export the corresponding pictures and models to aid the designer to design cabinets．

Fig.3 Cabinet computer-aided design system

4 Conclusion

In this paper，the style imagery model of the cabinet is built and the aided innovation design system of cabinet is proposed．According to the conjoint analysis method，a series of weight coefficients of various components and related constants is obtained．Then，the mapping model between the style imagery and external characteristics is established．Finally，w ith the cabinet product style imagery evaluation model and 3D recombinant technology，the cabinet aided design system is developed．The system is verified by practical application and feasibility，in addition．At present，the optimized version of this system is adopted by some cabinet manufacturers，and thus the companies have obtained certain econom ic profits．

［1］Cheng S F．Study on the design system of industrial cabinet［D］．Nanjing：School of Mechanical Engineering，Southeast University，2009．（in Chinese）

［2］Han F．Cabinetmodeling design and humanized design research［D］．Nanjing：School of Mechanical Engineering，Southeast University，2010．（in Chinese）

［3］McCormack J P，Cagan J，Vogel C M．Speaking the Buick language：capturing，understanding，and exploring brand identity w ith shape grammars［J］．Design Studies，2004，25（1）：1- 29．

［4］Giannini F，Monti M，Podehl G．Styling properties and features in computer aided industrial design［J］．Computer-Aided Design and Applications，2004，1（4）：321-330．

［5］Giannini F，Montani E，Monti M，et al．Semantic evaluation and deformation of curves based on aesthetic criteria［J］．Computer-Aided Design and Applications，2011，8（3）：449- 464．

［6］Wu H H P，Lee M T，Yu Y H．A novel hybrid method in new style product design patent retrieval［J］．Transactions on Information Science and Applications，2007，4（6）：1257- 1262．

［7］Dem irbilek O，Sener B．Product design，semantics and emotional response［J］．Ergonomics，2003，46（13）：1346- 1360．

［8］Chen K，Owen C L．Form language and style description［J］．Design Studies，1997，18（3）：249- 274．

［9］Huang Q，Sun S Q，Pan Y H．Product style recognition based on feature matching［J］．China Mechanical Engineering，2003，14（21）：1836- 1838．（in Chinese）

［10］Luo S J，Zhu S S．Users’and designers’product form perceptual image［J］．Chinese Journal of Mechanical Engineering，2005，41（10）：28- 34．（in Chinese）

［11］Zhu H J，Wu H R．Aided design method on geophysical prospecting for petroleum based on 3D visualization［J］．Journal of Information＆Computational Science，2014，11（4）：1303- 1311．

［12］Pan B S，Jiang L Z．Study on self-adaptive product design system based on modularization［J］．Journal of Information＆Computational Science，2004，1（3）：227-232．

［13］Potocnik D，Ulbin M，Dolsak B．Know ledge-based system for supporting the design of a plate-press［J］．Journal of Information＆Computational Science，2012，12（2）：024502-1- 024502-5．

［14］Ohrn A．Discernibility and rough sets in medicine：tools and applications［D］．Trondheim，Norwegian：Department of Computer and Information Science，Norwegian University of Science and Technology，2004．

基于風(fēng)格意象的機(jī)柜計(jì)算機(jī)輔助設(shè)計(jì)方法

沈張帆 薛澄岐 王海燕 牛亞峰 邵 將 張 晶

（東南大學(xué)機(jī)械工程學(xué)院，南京211189）

針對(duì)國內(nèi)機(jī)柜產(chǎn)品成本高和開發(fā)周期長的實(shí)際問題，提出了一種基于風(fēng)格意象的機(jī)柜計(jì)算機(jī)輔助設(shè)計(jì)方法．根據(jù)聯(lián)合分析法原理、粗糙集理論以及機(jī)柜各組成部件的權(quán)重系數(shù)，建立了機(jī)柜風(fēng)格意象評(píng)價(jià)的多維模型．計(jì)算得到了各風(fēng)格意象的相關(guān)常量及系數(shù)，并通過3D模型建立了機(jī)柜組件庫．最終，結(jié)合組件重組技術(shù)以及機(jī)柜產(chǎn)品風(fēng)格與特征之間的映射模型，基于Visual Studio軟件開發(fā)了原型系統(tǒng)．該系統(tǒng)實(shí)現(xiàn)了產(chǎn)品風(fēng)格意象與造型特征的雙向推理，輔助設(shè)計(jì)師進(jìn)行創(chuàng)新設(shè)計(jì)，大大提高了機(jī)柜研發(fā)的效率，提升了企業(yè)的經(jīng)濟(jì)效益．

機(jī)柜；計(jì)算機(jī)輔助設(shè)計(jì)；風(fēng)格意象；部件重組；造型特征

TP391

10．3969／j．issn．1003－7985．2015．03．012

2015-02-27．

Biographies：Shen Zhangfan（1988—），male，graduate；Xue Chengqi（corresponding author），male，doctor，professor，ipd_xcq＠seu．edu．cn．

s：The National Natural Science Foundation of China（No．71271053），the Scientific Innovation Research of College Graduates in Jiangsu Province（No．CXLX13_082）．

：Shen Zhangfan，Xue Chengqi，Wang Haiyan，etal．The computer-aided designmethod of cabinet based on style imagery［J］．Journal of Southeast University（English Edition），2015，31（3）：369- 374．

10．3969／j．issn．1003－7985．2015．03．012