案例检索算法冲压模具英文文献和中文翻译
1. Introduction Design method based on Case-Based Reasoning (CBR) is in virtue of the preterit cases and experiences which regard design reference to solve new design problems, overcome the bottleneck of knowledge acquisition and solve complex problem difficult to control of the reasoning system. Case retrieval is the key technology of CBR. It has biggish effect on the efficiency of obtaining the answer and reasoning veracity. Because design cases have many characteristic attributes whose importance degree is different and weighting coefficient is difficult to calculate, we usually obtain the results by the supposition or average weighting coefficient method based on experiences and knowledge. So this result can not prove its correctness and guarantee the validity of case retrieval. Rough Set Theory (RST) is used to process weighting coefficient problem of the case attributes. It makes use of the existing information to calculate the importance degree of all attributes and provides impersonal and reasonable evidence for the similar matching [1]. So the paper brings forward the case retrieval technology based on rough set theory which realizes the similar matching between the object case and the original case.
2. Rough set theory In the rough set theory, the knowledge is represented by the form of information system. An information system is a data set which expresses with the two-dimensional form. Row expresses the attribute and column expresses the object. In order to realize the intelligentization of the processing data, there need some signs which to express knowledge [2]. The knowledge representation is to research the set of objects. So the foundation of RST is its basic concept, and then a few main concepts are introduced as follow [3]: Definition 1 (Decision Table) S = {U, A, V, F}, U is an objects collection (Domain), C∪D=A is an attribute set. Subset C is called condition attribute and D is called decision attribute. V=Ur∈RVr is a set of attributes. Vr is the attribute values range of r ∈ A. F: U×A→V is an information function, which designates each object x’s attribute value in U. Definition 2 (Indiscermibility Relation) S = {U, A, V, F} for every attribute subset B ⊆ A to define an indiscernibility relation Ind(B): Ind(B)={(x, y)∈U2:fa(x)= fa(y),∀a∈B}. Ordered pair BS = ( U, IND( B) ) is called approximation space. According to the offered information of attribute set B, the included entity equivalent [X]B is defined to be [X]B={y∈U|xIND( B) y}. Definition 3 (Object Set) S = {U, A, V, F}, A=CUD and B ⊂ C. Defined D’s positive region of B: PosB(D)=∪{B(X):X∈IND(D)}. Definition 4 (Attribute Set) S = {U, A, V, F}, A=CUD and the dependability between D and C is defined: K =γc(D) =|POSC (D)| / |U|（0≤K≤1）. For the attribute of c∈C, if γC (D) =γC−c (D), then attribute c is redundant attribute which is relative to decision-making attribute D. Definition 5 (Attribute Reduction) S = {U, A, V, F}, A=CUD and B⊆C. If B and D are independent, and γC (D) =γB (D), then B is called C’s relative reduction of D, and marks as Rred