Available online at ScienceDirect. Procedia Manufacturing 11 (2017 )

Transcription

1 Available online at ScienceDirect Procedia Manuacturing 11 (2017 ) th International Conerence on Flexible Automation and Intelligent Manuacturing, FAIM2017, June 2017, Modena, Italy The architecture o intelligent system or CNC machine tool rogramming Jan Duda*, Janusz Pobozniak Institute O Production Engineering, Cracow Unviersity o Technology, Al. Jana Pawla II 37, Cracow 31-86, Poland Abstract The goal o the aer is to resent the architecture o CNC machine tool rogramming system based on the meta-knowledge and recognition o workiece intermediate states. Meta-knowledge reresented as hierarchical decision nets has the control role over manuacturing knowledge stored in the orm o roduction rules. This is a Comuter Aided Process Planning System (CAPP) with the unctionality limited to the CNC machining. Additionally, both the manuacturing knowledge as well as eature recognition and rocessing algorithms were develoed secially or the CNC machining The The Authors. Authors. Published Published by by Elsevier Elsevier B.V. B.V. This is an oen access article under the CC BY-NC-ND license (htt://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under resonsibility othe scientiic committee o the 27th International Conerence on Flexible Automation and Peer-review under resonsibility o the scientiic committee o the 27th International Conerence on Flexible Automation and Intelligent Manuacturing Manuacturing. Keywords: CAPP; Comuter Aided Process Planning; CNC; machining eature recognition; machining eature interaction; manuacturing knowledge reresentation 1. Introduction CAM (Comuter Aided Manuacturing) systems are widely used in nowadays industry. These systems are used or CNC (Comuter Numerical Control) machine tool rogramming. The workiece model is irst created in CAD system. Next, the manuacturing rocess lan describing how to machine the art and listing the tools and machine * Corresonding author. Tel.: ax: address: duda@mech.k.edu.l The Authors. Published by Elsevier B.V. This is an oen access article under the CC BY-NC-ND license (htt://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under resonsibility o the scientiic committee o the 27th International Conerence on Flexible Automation and Intelligent Manuacturing doi: /j.romg

2 502 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) tools is reared. Based on the rocess lan, the manuacturing engineer creates the rogram. The CNC rogram is created in CAD system environment or in a standalone CAM sotware. Taking into account the oularity o CNC machine tools and the growing geometrical comlexity o arts, it is obvious that CAM system are widely used. It is relected, or examle, in the number o localised CAM systems, i.e. oered with Polish language interace available on the market. For examle, the Polish language version o AlhaCAM, ESPRIT, HyerMill, MasterCAM, and Siemens NX are available. It is diicult to list all available CAM systems. The trends to automate CNC machine tool rogramming can be noticed. There are two directions in the automation trends: manuacturing eature recognition and use o manuacturing knowledge. These trends were ormulated on the base o author ersonal exerience resulting rom long-term oeration o various CAM systems. Therese exeriences come rom the articiation in the trainings authorized by CAM system roviders, teaching activities, articiation in many industrial trade airs, conerences and research rojects using CAM systems. There is no research works estimating unctionality o commercial CAM systems. Because o this, the literature review was ocused on CAPP (Comuter Aided Process Planning) systems, as many solutions develoed or CAPP systems is later used in CAM sotware. This is esecially true in case o solutions increasing the intelligence and automation o these systems, including eature recognition and the tools using manuacturing knowledge. As the unctionality o leading CAM systems is similar, no trade names o articular sotware ackages were given. The goal o this work is to describe the general unctionality o CAM systems and to roose the ways to increase it, and not to criticize the seciic CAM systems having strong ositions among their industry users due to their eatures. There are several literature ositions on the manuacturing eatures recognition like ockets, holes, slots, stes, etc. The recognition methods are using CGM or B-Re reresentation. CSG reresentation loses its oularity and because o it, the recognition algorithms based on this reresentation will be not discussed. Three commonly used recognition methods or B-Re ormat are grah-based algorithms, volumetric decomosition techniques and hint based geometric reasoning [1]. The CAM system roviders doesn t seciy which recognition aroach is used in their sotware, but all these three methods can be used. The literature review [1] clearly shows that there is no roblem to recognize the isolated manuacturing eatures, i.e. the manuacturing eatures without the common elements with other eatures. One o the main roblems blocking the eature technology is the recognition o interacting manuacturing eatures i. e. manuacturing eatures sharing the elements with other eatures [1, 2]. The interacting manuacturing eatures generate two roblems: Due to the interaction, some eature aces are removed, which makes diicult to recognizes the eatures. The eature attern contains usually the eature descrition with all aces. I some aces are missing, the eature doesn t it to the attern. There could be more than one interretation o interacting manuacturing eatures. For examle, two interacting slots with the same deth can be interreted as one ocket or as two or more slots. The manuacturing eature recognition algorithms in commercial CAM system don t deal with roblems related to interacting manuacturing eatures [1, 2], but recognize only simle, non-interacting eatures. This work outlines the algorithm or dealing with interacting eatures. The general idea o this algorithm seems to be more simle in comarison to other works, as or examle [2]. Nevertheless, it shall be thoroughly checked by ractical imlementation. CAM systems have also limitations in the area o manuacturing knowledge use. Tyically they allow to assign machining cycles to recognized manuacturing eatures. For examle, i there is a hole in the art, taking into account its arameters like diameter, length, tye o bottom, the thread, etc., the sequence o machining cycles can be automatically determined, or examle, re-drilling, drilling, rough reaming and inal reaming. Usually, each cycle can be assigned the value determining its riority. When assigning the machining oeration to manuacturing eatures, CAM systems uses rules (or similar logic constructions). This aroach is also used in knowledge-based CAPP systems [4]. The examle o such rule based system is given in [5]. Our exerience shows that the main actor deciding about the ractical value o knowledge-based CAPP system is the use o meta-knowledge having the suervisory role over rule-based knowledge database. This work resents the idea o meta-knowledge relecting the

3 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) basic manuacturing rinciles. The use o meta-knowledge increases the usability o CAM systems. The simliied solutions occurring in nowadays commercial CAM systems solution doesn t allow to determine the roer order o machining cycles, which shall result rom the basic manuacturing rinciles. The goal o the aer is to resent the architecture o CNC machine tool rogramming system based on the metaknowledge and recognition o intermediate workiece states. This is a Comuter Aided Process Planning System (CAPP) with the unctionality limited to the machining. The roosed solution will allow to increase the degree o automation o CNC machine tool rogramming with CAM system. 2. Recognition o intermediate states The task o manuacturing eature recognition is the grouing o low-level entities like aces, loos, edges. etc. into the manuacturing eatures. Low-level entities are usually read rom the drawing database o commercial CAD systems based on B-Re. Three commonly used recognition methods or B-Re ormat are grah-based algorithms, volumetric decomosition techniques and hint based geometric reasoning [1]. Indeendently rom the selected recognition method, the common oinion is that interacting manuacturing eatures revent the usage o eature technology [2]. Due to the interactions, the eatures become distorted, or examle some aces are removed. This signiicantly imedes the eature recognition. The other roblem accomanying the recognition o interacting eatures, not aroriately considered in the research works, is the decomosition o the interacting eatures. Figure 1a resents the examle. The art has two interacting through slots. They can be interreted as various simle eatures. For examle, these slot can be treated as the oen ocket (Figure 1b), machined with the one machining cycle (Figure 1e). In the other decomosition, there could be one long, through slot (Figure 1c) and two shorter through slots. The longer slot is machined irst, and then the two shorter slots. The decision, which slot should be selected as the longer one must be taken (Figure 1c or Figure 1d). Sometimes, it is very imortant rom the manuacturing oint o view. I the interacting slots have dierent widths, the slot with larger width should be machined irst (Figure 1). This is because o the basic manuacturing rule recommending the removal o the larger volume, when the cutting orces are higher, on the beginning o the manuacturing rocess. So, the slot with the larger width, as having the greater volume, should be machined irst. The other reason is that the machining with the tool having greater diameter is more eicient end economically justiied. Because o this, the common art o these two slots should be machined with the tool having the greater diameter. The ersonal exerience o the author suggests, that manuacturing eatures cannot be correctly recognised using only the geometrical reresentation, or examle in B-Re (Boundary Reresentation) ormat. Unortunately, nearly all research works on eature recognition ignore the act that manuacturing knowledge lays imortant role in eature recognition. This also limits the usability o the eature technology or ractical imlementations. The other roblem, which does not have yet the solution, is the recognition o the eatures occurring in the intermediate states o the art during the machining. Such eatures very oten are not resent in the machined art in its inal state. For examle, during the machining o the art shown on Figure 2 there is the acing cycle. The cycle uses the lane manuacturing eature. The standard eature recognition systems do not ind such lane eature during the recognition rocess o this art. Similar situation occurs in the case o the roile contouring cycle, which uses the boss eature. To solve these roblems, the aroach with the ollowing stes is used: machining eature recognition, transormation o machining eature selected on the base o decision nets described below and the next eature recognition. The eatures are transormed by using the reverse method, rom the workiece in its inal state to the stock by adding material. The eature transormation can result in the aearance o new eatures, which are then recognized. The order o manuacturing eature transormation is given by manuacturing knowledge in the orm o hierarchical decision nets. Such aroach allows or the recognition o interacting eatures and eatures not visible in the workiece inished state. The detailed descrition o the whole rocedure is given in [3]. 3. Manuacturing knowledge reresentation in the orm o hierarchical decision nets CAPP systems are evolving toward knowledge based and exert systems [4]. Manuacturing rocess lanning requires the aroriate manuacturing knowledge database. Taking into account the usage o the knowledge, the ollowing knowledge subsets were distinguished [6, 7]:

4 504 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) a) Features to be recognised consisting o two interesecting through slots b) Oen ocket c) d) Slot 2 Slot 1 Slot 2 Slot 1 2) Facing(eature: lane) 3) Proile contouring (eature: boss) Slot 3 Slot 3 e) ) 1) Stock 4) Slot milling (eatures: slots) 5) Final art Figure 1: Dierent views on intersecting eatures Figure 2: Features in the intermediate states o the art knowledge covering the rules or the classiication o machine workiece to select the aroriate generalized structure o manuacturing rocess SK, knowledge covering the rules or the selection o the tye o stock or already determined tye o machined workiece SF, knowledge covering the rules and rinciles o rocess lanning or stock SW knowledge covering the rules and rinciles or machining rocess lanning SP. Manuacturing knowledge covering the rules and rinciles o machining rocess lanning SK, SF, SW, SP is reresented as hierarchical decision nets, describing the deendencies between extracted knowledge areas. This is the knowledge o I tye. The structure o machining rocess on =1 P levels o exansion shows Figure 2. There is a set o decisions dp DP describing the rinciles or the selection o activities tp TP occurring in the machining system. The relations RP is described on the countable sets DP i TP : RP DP TP DP, =1..P The general structure o machining rocess lan results rom the roduction rocedures and know-how, and can be resented as multi-level model o the rocess structure: GS = DP,TP,RP where: DP - set o decision nodes GS reresenting the decision sets on level or the selection o roduction activities. TP - set o edges GS reresenting the activities on level o exansion o the machining rocess structure RP - relations on the sets o nodes and edges =1..P level o exansion o rocess lan structure The set o rules or machining activities selection deending on the machining system status deined on the decision sets deines the manuacturing knowledge o tye I. The net can be deined on the rocess structure GS P : D T SP = GS,, { } where: GS - grah o the general rocess structure on -level o exansion, D T -set o unctions described on the set o decision nodes on -level o exansion o grah GS, - set o unctions described on the set o edges o grah GS P. Function D describes the relations between the decision and the machining activities having dierent level o

5 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) details and it is deined on the set o decisions dp DP D : D GD It assignes the grah tree describing the conditions o the decisions determining the next machining activity to each decision dp DP. The knowledge SP determines the set o rules and rinciles o machining rocess execution. The machining activities determined with knowledge SP can be described in more details using the net SP +1 o level +1. So, on the set o activates tp TP or =1,... P-1, the ollowing unction can be deined: T : TP SP +1 The machining rocess lan is generated using the multi-level net model reresented as the rames and decisions rules in exert system. S11 S12 S13 S14 S15 I tp12 C B tp15 A tp11 tp14 tp16 dp11 tp13 dp12 dp13 dp14 dp S21 S22 S23 H tp21 tp22 D tp22 tp23 dp21 dp22 dp A tp16 S14 B tp14 S13 C tp13 S23= S13 D tp23 S33= S23 E tp32 S32... F tp S= S-1... G tp31 S31= S22 H tp21 S21= S12 I tp11 S11 G E tp31 tp32 dp31 dp32 dp tp33 S31 S32 S33 Si i-ty stan ośredni -tego oziomu rozwinięcia struktury uogólnionej SP1 SP2 F tpp1 dpp1 dpp2 P1 Figure 3: Model o manuacturing knowledge in the orm o hierarchical decision nets, 4. Procedure or machining rocess lanning Machining rocess lanning is done in stages resented on Figure 4. In irst stage, the generalized structure o the manuacturing rocess is selected based on the manuacturing and geometrical characteristics o the roduct. The roduct is evaluated taking into consideration the values o its attributes, according to the sequence resulting rom the knowledge SK describing the rules or art classiication in the view o manuacturing. The classiication into manuacturing class determines the manuacturing knowledge or the stock tye selection SF and manuacturing knowledge or structuring the rocess lan SP. Second stage is realized in the ollowing stes: lanning o the intermediate shaes o the roduct,

6 506 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) design o stock. The base or the determination o the workiece intermediate states a generalized structure o manuacturing rocess selected during the irst stage. On the basis o geometrical and manuacturing characteristics o the workiece and the manuacturing knowledge, the intermediate states and the stock tye may be determined by using the reverse (rom the inished art to the stock) method. This method includes the identiication o workiece suraces with characteristics matching the lanning activities selected according to generalized structure, rom end to start nodes on each level o hierarchical net. The tye o stock is selected on the basis o the set o rules or classiication o stocks SF. The necessary arameters are collected rom three sources: art tye; it is the outut o irst stage, Selection o the general rocess structure and tye o stock Reverse lanning o stock and intermediate shae generation Generation o rocess documentation Figure 4: The stages o rocessing geometrical and manuacturing characteristic (material, mass, accuracy requirements, etc.) rom eature/object roduct database, dimensions o standard stocks rom catalogue. As the result o the reasoning, the stock tye is roosed (rolled, orged, etc.). The selected tye o stock and intermediate states determined or the start node orm the basis or the lanning o stock and its rocess lan with knowledge RW. The determination o the intermediate shaes allows to decomose the overall lanning goal, the develoment o rocess lan, into sub-goals. The lanning goal can be deined as ollows: Select the means (machine tools ixtures, tools) to realize the sub-goals reresented by the intermediate shaes while creating the hierarchical, object oriented structure o rocess lan or the given workiece. The actions are carried out based on the ollowing data: characteristics o the stock material and intermediate states o the workiece being the outcome o the stage II, characteristics o the manuacturing system caabilities selected or the rocess lan imlementation. The algorithm or rocess lan generation covers the generation o oerations, set-us, ositions and cycles.

7 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) The structure o comuter system The comuter system was imlemented in Windows environment and its structure is resented on Figure 5. Shell exert system Exsys Proessional is the main element o this system. The knowledge is reresented in this system as the logical rules IF THEN ELSE. Very imortant eature o this tool is the ossibility to modiy the reasoning rocess using the scrit written in the secial Command Language (CL). This scrit language allows to select the grous o rules or reasoning in the determined order or run the rograms develoed using the rocedural languages. Command Language was used to imlement the meta-knowledge. The meta-knowledge in the orm o multi-level decision nets determines the grous o rules used in the given context o machining rocess lanning. The hierarchical decision nets are reresented in the orm o aroriately ormatted text iles. These iles are read and interreter by Command Language. The second task o CL scrits is the running o rocedures written in rocedural languages or machining eature recognition. These rocedures are called both beore as well as during rocess lanning. Figure 5: System architecture The workiece model is created in 3DXExerience PLM system. This system oers FT&A (Functional Tolerancing & Annotations) module, in which 3D model is directly used to add non geometrical inormation instated o rearing the traditional 2D worksho drawing. This inormation can be directly rocessed by comuter systems, which is not ossible in the case o traditional documentation with 2D art views. The non-geometrical data modeled with FT&A module include the art dimensions, dimension tolerances, surace roughness, location and shae tolerances, etc. The data rom 3DXExerience system are imorted to CGM grahic kernel by Satial Cororation. CGM grahic kernel imorts 3DEXerience iles, including non-geometrical data (CGM grahic kernel calls them as PMI (Product Manuacturing Inormation) data, gives access to all geometry and toological data, allows to execute tyical geometrical oerations like Boolean subtraction and oers also the tools or visualization (through Hoos Visualize sotware). The manuacturing eature oriented workiece database is created using the outcome o eature recognition sotware [3]. During the machining rocess lanning, the workiece is transormed rom its inal state to the stock state by the series o manuacturing eature transormation. The manuacturing eature transormation removes existing manuacturing eatures (by adding material) and cause the occurrence o new manuacturing eatures. The manuacturing eature transormation by adding material is imlemented using the Boolean add oeration, surace removal or surace oset oeration. The CGM grahic kernel has the tools or all such oerations. Generated manuacturing rocess is stored in extended STEP-NC ormat. STEP (Standard or the Exchange o Product Model Data) (ISO 10303) is a comlete and the unambiguous roduct data descrition throughout its lie

8 508 Jan Duda and Janusz Pobozniak / Procedia Manuacturing 11 ( 2017 ) cycle. STEP-NC is the alication o STEP into the CNC machine tool rogramming [7]. Unlike the ISO 6983 (G codes) standard describing the tool motions relative to the art and inorming "how to make" that art, STEP-NC standard inorms "what to do", describes the sequence o machining cycles, tool ath style, the required accuracy and the tools. The STEP-NC concet uses the manuacturing eatures. This is the only available standard allowing to save all data rom CAPP system, including the ull descrition o workiece geometry, recognized machining eatures, machining cycles done on these eature, arameter o machining cycles, machine tool unctions (coolant, etc.). These data are urther used to generate G code. Such solution is orced by the situation on commercial CNC control systems oer. Nowadays there is no commercial CNC systems suorting STEP-NC standard. It should be noted, that the use o STEP-NC standard simliies the develoment o CAPP system, as this standard oers a lot o data structures. Esecially useul are the data structures or storing eature geometric data and the geometries o workiece and stock. This CAPP system structure allows to veriy the roosed aroach or CNC machine tool rograming using the meta-knowledge in the orm o hierarchical decision nets and the recognition o workiece intermediate states. 6. Conclusions The algorithm was artially veriied by the develoment o sotware. Base on the results o the work, the ollowing conclusions can be ormulated: the roosed algorithm seems to be relatively simle in comarison with other algorithms or interacting eature recognition, while it has a good eiciency, the current state o the art is always subjected to recognition, without redicting the otential shae o the art ater the machining o the subsequent manuacturing eatures, the use o manuacturing knowledge signiicantly increases the ractical value o this algorithm. It is not ossible to make the correct recognition o manuacturing eatures without using the manuacturing knowledge. The develoed algorithm can also store the intermediate states, the use o hierarchical decision nets simliies the decision rocess during machining rocess lanning and decomoses the comlex lanning task into more simle artial tasks. manuacturing knowledge model used as meta-knowledge reduces the comlexity o roduction rules, hels to kee their consistency and oers the better results during the reasoning rocess. The goal o the urther works will be to continue the imlementation o CAPP system or the ull veriication o roosed aroach. This will include the urther works on machining eature recognition and transormation, manuacturing knowledge acquisition through the analysis o already existing machining rocesses o arts manuactured in aerosace industry, works and imrovement o algorithm or set-u lanning and urther develoment o sotware. Reerences [1] Babic B., Nesic N., Miljkovic Z., A review o automated eature recognition with rule-based attern recognition, Comuters in Industry, Volume 59, Issue 4, 2008 [2] Mesbahi A. E., Jaider O., Rechia A., Automatic Recognition o Isolated And Interacting Manuacturing Features In Milling Process, Int. Journal o Engineering Research and Alications, 2014 [3] Pobozniak J., An aroach or the recognition o interacting manuacturing eatures, Proceedings o the Ninth International Conerence on Engineering Comutational Technology, Nales, Italy, 2014 [4] Xu, X; Wang, L; Newman, St. Comuter-aided rocess lanning - A critical review o recent develoments and uture trends. International Journal o Comuter Integrated Manuacturing. 24, 1, 1-31, Jan ISSN: X. [5] Kumar S.P. L., Jerald J., Kumanan S., Feature-based modelling and rocess arameters selection in a CAPP system or rismatic micro arts, International Journal o Comuter Integrated Manuacturing, 2015 [5] Duda, J.; Habel, J.; Pobożniak, J. Semi-generative CAPP system CA systems and technologies : III Interntional Workshos CA System and Technologies, Cracow, [6] Duda, J., Comuter aided rocess lanning in the concurrent roduct develoment, Second International CAMT Conerence Modern trends in manuacturing, Wroclaw, 2003 [7] ISO :2004 Industrial automation systems and integration -- Physical device control -- Data model or comuterized numerical controllers -- Part 10: General rocess data