Automatic monitoring and control of machining systems has been shown to provide substantial economic benefits by improving operation productivity and part quality. However, the challenge of coordinating multiple complex machining modules has received very little attention by the industrial sector and the research community. This challenge will become increasingly important in the future for machining systems incorporating open platforms allowing for cost-effective integration and reconfiguration of modules and for reconfigurable machining systems which will require new module sets each time the system is reconfigured to meet new product demands. In this project, a systematic design approach, known as Supervisory Machining Control, was developed for the intelligent regulation of multiple complex machining modules. The supervisory controller is shown as part of a hierarchical machining control system in Figure 1.

Figure 1: Hierarchical Machining Control System.

The structure of the proposed Supervisory Machining Controller is shown in Figure 2. This controller consists of a state supervisor and an operation supervisor. The state supervisor monitors the machining operation via a set of monitoring modules to determine the instantaneous operation state (e.g., tool-workpiece contact, chatter) and the monitored parameters (e.g., feed estimate, force process model parameters). The instantaneous operation state is utilized by the operation supervisor to regulate the activity of the machining modules, while the monitored parameters are available to all modules which may require them as inputs. Given the operation state, the operation supervisor regulates the activity of the machining modules by turning them on/off, resetting them, etc.

Figure 2: Supervisory Machining Controller Subsystem.

A systematic design approach for constructing supervisory machining controllers was developed. The design approach utilizes the knowledge of the designer combined with graphical techniques (e.g., state transition diagrams, Grafcet) to construct the logic controller which will regulate the machining modules. Given the (a) machining operation and objectives, (b) machine and operation constraints, and (c) available machining modules, the designer constructs the supervisory controller via the following steps:

i) select machining modules to meet the design objectives,

ii) construct fixed machine and operation parameter modules containing the necessary information to support the machining modules, and

iii) construct operation supervisor via a systematic design tool (e.g., Grafcet) and convert the operation supervisor into an implementable module.

A supervisory controller was constructed, via the design approach, for a face milling operation. The operation objectives are to ensure (1) a chatter-free operation and (2) ensure the spindle power constraint is not severely violated (i.e., greater than 125%), which corresponds to a cutting force constraint. There are machine constraints on the maximum feedrate (36 mm/s), maximum spindle power (745.7 W), and servo amplifier inputs (+/- 10 V). There is also an operation constraint on the maximum feed (0.6 mm/tooth). The available machining modules are (A) Servo Controllers, (B) Linear Interpolator, (C) Feedrate Routine, (D) Operator Input Monitor, (E) Feed Hold, (F) Operator Chatter Detector, (G) Automatic Chatter Detector, (H) Chatter Suppresser, (I) Chatter Suppression Routine, (J) Parameter Estimator, (K) Force Signal Processor, (L) Tool-Workpiece Contact Monitor, and (M) Adaptive Force Controller.

The above modules are selected for the face milling operation. The fixed machine parameter module contains servo amplifier saturation values, maximum spindle power, maximum feedrate, and axis model parameters. The fixed operation parameter module contains maximum feed, number of tool teeth, tool radius, module sample periods, reference machining force, and servo and adaptive force controller time constants. The Grafcet representation of the developed supervisory controller si shown in Figure 3. The states are:

1) initialize modules,

2) no tool-workpiece contact and no chatter,

3) operator feedhold active,

4,7,8) reset modules

5) tool-workpiece contact and no chatter, and

6) chatter suppression routine active.

The transition receptivities are:

a) operator feedhold signaled,

b) tool-workpiece contact detected,

c) tool pass begins,

d) chatter manually detected,

e) chatter automatically detected, and

f) chatter suppression routine complete.

Figure 3: Grafcet Representation of Supervisory Machining Controller.

The experimental implementation is shown in Figure 4. When the tool became fully immersed in the workpiece, chatter developed and the chatter suppresser was implemented to halt the operation and rewrite the part program to accommodate a second tool pass. While the chatter suppresser was active, the adaptive force controller was reset, and the cutting force was smoothly tracked during the rest of the tool pass, and again in the second tool pass. The supervisory control approach allowed for a systematic means to create a logic controller which regulated the machining modules to ensure their proper performance.

Figure 4: Experimental Results. The variable F (solid line) is the cutting force normalized by the reference force (190 N). The variable d (dashed line) is the depth-of-cut normalized by the programmed depth-of-cut (2 mm). When d is 0, cutting is not occurring. The operation consists of two tool passes.

The benefits of the developed supervisory machining control approach are:

1) a systematic design tool to reduce the lead and ramp-up time (hence, cost) of high level machine tool logic controllers for conventional and reconfigurable machining systems, and

2) a modular structure which is ideal for implementation in future open architecture machining systems.

This research was conducted in collaboration with Professor A. Galip Ulsoy at The University of Michigan and was financially supported by the National Science Foundation under grant DDM-9313222.

Related Information

"Supervisory Machining Control: Design and Experiments," R.G. Landers and A.G. Ulsoy, Annals of the CIRP, Vol. 47/1, 1998, pp. 301-306.

"Supervisory Machining Control on an Open-Architecture Platform," R.G. Landers and A.G. Ulsoy, CIRP International Conference on Design and Production of Dies and Molds, Istanbul, Turkey, June 19-21, 1997, pp. 97-104.

"Supervisory Machining Control: A Design Approach Plus Force Control and Chatter Analysis Components," R.G. Landers, Ph.D. Dissertation, The University of Michigan, Department of Mechanical Engineering and Applied Mechanics, Ann Arbor, Michigan, April 1997.

"A Supervisory Machining Control Example," R.G. Landers and A.G. Ulsoy, International Conference on Recent Advances in Mechatronics, Istanbul, Turkey, August 14-16, 1995, pp. 990-997.