Một phương pháp cải tiến nhận dạng đặc tính gia công cơ khí tự động từ mô hình cứng 3d

Journal ofScience & Technology 100 (2014) 001-005 An Improved Method of Automatic Machining Feature Recognition from 3D Solid Model Phung Xuan Lan*, Hoang Vinh Sinh Hanoi Umversity ofScience and Technology, No I Dai Co Viel Str., Hai Ba Trung, Ha Noi, Vie! Nam Received: February 18, 2014; accepted: Aprd 22, 2014 Abstract Automatic machining feature recognition from 3D solid model is seen as one of the most complicated tasks in creating communication bridge between design and manufacturing This paper presents a rule-based recognition method for extracting machining features from 3D solid model Overcoming the limitations of eartier studies, this proposed system is capable of not only recognizing many kinds of design features such as cut, boss, sweep and revolution but also separating into unit-machining features from a multiple-contour sketch based design feature It is also effective to recognize both isolated and complex interacting feature in a uniform way The capability of proposed method was demonstrated by two specific case studies The extracted features are equipped with the essential manufacturing information that can be directly applicable for various applications such as computer aided process planning (CAPP) Keywords: Feature recognition Rule-based method Machining feature I Introduction Integrated design and manufacturing systems are considered as an effective approach to increase productivity The initial step in designing such integration system is machining feature recognition It is process of converting from design featiue mto machining featiire Design feature is a form feature or geometric feature of interest to the designer such as boss, sweep, and hole Machining feature is a volume or material to be moved to produce a form feature [I] The design feature contains detailed geometric infonnation of part; however they are not manufactiumg information directly used in the manufacturing activities such as process planning In the last two decades, many feature recognition methods have been proposed for exttacting machining information from geometric information There are two main approaches: extemal approach and intemal approach The first approach uses neutral files in formats such as DXF, IGES or STEP to exttact the desired features One of limitations of this approach is that significant amount information associated with the model is lost in the ttanslation process [2] In the second approach, both the creation of the CAD part model and the feature recognition are done m the same feature based design CAD system such as Solidworks, Pro-Engineer, Autodesk and Inventor Although this approach is restricted to one CAD system, it has some advantages such as the accessibihty to the object library of design features • Corresponding author: Tel: (+84) 935.^ 8.435 Email: lan.phungxuan@hust.edu.vn easily, the reduction of time for recognition process [2] Because of these benefits, it has been focused in this research One of the most difficult tasks in feature recognition is the interaction feature which can distorts the relationship between faces, edges and characteristics of vertexes in the part and thus it results in the significant errors in feature recognition Among the intemal approaches, M.T.Hayasi etal recognized machining feature through the number of faces for each design feature and the edge number of its base face [2] The method was able to effectively recognize non-interacting features but couldn't applied for recognizing features whose boundaries at the 3D bounding box interacting each other such as P M F a n d P M F I ( F i g 2) X Zhou et.al [3] used similar method to M.T.Hayasi et.al Although it could recognize more kinds of design features, it still had same drawback None of these algorithms were able to handle the multiple-contour sketch based features This problem can be explained as follow The designer can fabricate a design feature from multiple-contour sketch For instance, there are three specific cases including (PMF2, PMF3), (PMF5, PMF6, PMF7) and (PMFI5, PMFI6) (Fig 4) which are constmcted from only one design feature named boss-extmde2, cutextmde2 and cut-extrude7 respectively Both mentioned methods could only recognize them as one machining feature although they should be different machining featiues because of the differences of machining operations or cutting tools It is surely an undesirable result in feature recogmtion Journal ofScience & Technology 100 (2014) 001-005 I Open Solidworks applicabon with active part | Algonlhm 2: Oeteimining ihe sketch face and base face AlgortJiin 3' Recognizing feature interaction through finding interaciion algorithm Fig The flow chart of recognition Other method from V.S, Muniappan was based on checking inner loop and outer loop of faces, determining transition characteristic of each vertex and the concavity of all edges on base face [4] It also failed to recognize the interacting features such as PMFI, PMF6 and PMF13 (Fig.2) The main reason is that the number of vertex transitions and edges were changed due to interaction while the feature definition was still not changed Hence, the recognition result could be distorted One more drawback of this method is that it limited to the simple features such as rectangular pocket, slot or step but couldn't recognize the complex features including revolve pocket, loft pocket and sweep slot such as PMF4 (Fig 4) This paper seeks to overcome the drawbacks of the existing methods through the development of a proposed method as presented in next section Algorithms of separating into unitmachining features and determining the feature interaction type according to each edge type are the key to success of the proposed method Proposed Method The mle-based method is used to identify both interacting and isolated featture in uniform way The procedure of proposed method for feature recognition is shown in Fig Through this procedure, most of major machining features according to ISO 10303224 [5] such as pocket, slot, step, planar, boss, fillet and chamfer with many kinds of shapes are recognized Fig Feature interaction example Stepl Determine the design feature composing of multiple imit-features Step2: Get all faces of the design feature Step3: Determine the similarity coefficient of faces This similarity coefficient is based on the shortest distance between each pair of faces Two faces sharing the same edge give the highest similarity coefficient Slep4: Use the linear cell clustering algorithm [6] to group together these faces that have the highest value of the similarity coefficient to form face group Step5 Determine the real type of unit features (cut, boss, sweep, e t c ) Algorithm2: Determining the sketch face and the base face of all features Stepl: Determine the sketch face The sketch face is the face on which the sketch was drawn Step2: Determine the base face The base face is the planar face having maximum number of concave or convex edges and its norma! is compared with too! axis, [4] Step3 Determine the bottom condition of feature The bottom condition code is or I in case of through, blind feature respectively Algorithm3: Recognizing feature interaction This process is only earned out in case of blind cut feature Algorithml: Separating into unit-design features if exist Stepl Determine the concavity of all edges of outer loop on base face [4], If the design feature composes of many sketch contours and it is not boss-extmde-thin type, algorithm I is applied to re-group faces of design feature into multiple unit-groups Stepl: Determine the interaction characteristic of each edge of outer loop on base face Based on the basic types of interaction such as volume interaction, adjacent interaction and ttansition interaction [7], this paper presents a different method to determine al! Journal ofScience & Technology 100 (2014) 001-005 interaction types as shown in Table It only considers the interaction characteristic of each edge type The example of all interactions for each edge type is shown in Fig Table I Name and code of interaction TvDe of edge Concave edge (inner edge) Convex edge (outer edge) Tvoe of interaction No interaction Volume interaction Boss interaction Side interaction Transition interaction No interaction Nested interaction Side interaction Volume interaction Transition interaction Code 2" 2' 2= 2= 2" 2^ 2^ 2^ 1^ 2" Step3: Calculate the number of edge interaction for each type defined by No-Edge(i) where i respects to the superscript of interaction code If many edges are on a same line or circle, the counting result is not increased This point of algorithm is very important to prevent error due to volume interaction which happens between PMFI and PMF8 as shown in Fig.2 Algorithm4: Recognizing machining feature Rulel Cut-blmd recognition For each cut-blmd feature, Stepl: Machining type feature definition If the feature is constmcted from a slot sketch or it has only two planar side faces and they are parallel to each other, it is a slot If it is not a slot and No-Edge(5) is a)