classification for F2 (feature 2 in Fig.11). F2 has nine inspection single features; Side Plane 2(4 EA), Side Quarter Cylinder 2(4 EA) and Bottom Plane 2(1 EA). The number of measuring points and their location for each inspection feature are determined by the OMM algorithm, so the decision of measuring points for one characteristic feature is completed. Fig.13 shows the determination of measuring points which is applied to Fuzzy algorithm and Hammersley’s method for the representative inspection features of F2. Fig.14 shows the number of measuring points, their location and probe path. SP2a, SP2b, SC2a, BP2a are composed with 7, 7, 5, 7 measuring points, so the total number of measuring points is 26 in F2.
Fig.13 Distribution of inspection points for Typical Feature of F2
Inspection planning
STEP 1
STEP 2
Fig.11 Feature classification for part
Fig.12 Inspection feature classification for F2
Fig.14 Distribution of inspection points for F2
4.3 Determination of Non-contact point
The measuring points of F1, F3 which are machining features are determined by the same method to be applied to F2. If S1~S5 are applied to Hammersley’s algorithm, however, non-contact point could be generated as in Fig. 15(a). Most of S-group has non- contact points. In this case, the positions of non-contact points are determined again. In this study, for the inspection feature which has non-contact points, the locations of measuring points are determined again by the rectangle mesh. Fig.16 shows the partition of S1. The method of location decision, at first, detects whether the measuring points exist or not in each rectangle mesh and locates the measuring points inpidually on mesh-center again as many as the number of non-contact points from the largest mesh to the smallest mesh which doesn’t have the measuring point. If the number of non-contact points is more than the number of rectangle mesh, the pided surfaces are regenerated using the measuring points of
each rectangle mesh and the measuring points are located again in order of large area. Besides, if the range between the surface and the rectangle mesh is larger than two- third of the minimum range, there is possible to be distributed the measuring points, otherwise, it is impossible. The real S1 has three non-contact points and the position of measuring point is decided again in large area order; M1a1, M1b2, and M1a3. The result is showed in
Fig. 15(b). [13][14]
(a) Before (b) After
Fig.15 Relocation of non-contacting measuring points for S1
Fig.16 Rectangular mesh generation
5. Simulation & Experiment
The OMM program was made of Visual C++ ver.
6.0 based on the proposed inspection method. This program was loaded into DPROM which is the dialog based program of PC-NC, and the machining simulation and inspection were executed. The workpiece was designed and the machining simulation was executed for testing the developed OMM-Module.(Fig. 17) The workpiece material is an MC(Methylene Chloride). The real machining and the OMM were performed in a CNC M/C(V100, HYUNDAI).(Fig. 18) The results were obtained by performing the above inspection method. The number of measuring points is 60 and the time which is required for inspection is 10 min. 50 sec. Fig. 19 shows the moving path of the probe and the location of
measuring points based on the inspection plan. Table.2 and Table.3 show the analytic result of geometric tolerance after inspecting the machined object. Table.2 shows the position tolerance of inspection surface. SC1a, the side surface of F2, was estimated the worst machining state and TC1a, the upper surface of Island (cylinder), was estimated the best machining state. Table.3 shows the estimated results of geometric tolerances except for that. In case of performing the inspection mode, the precision of machine tool is an input variable. The purpose is to improve the precision of a machine tool, so the geometric tolerance of a machine tool was measured by laser interferometer and applied to this system. The pre-traveling error which is generated by contacts of probe was gained by the experiment and the result of inspection was calibrated by this error. [15]~[19]