The purpose of this project was to develop an Adaptive Control Algorithm for a CNC milling machine. The milling machine is controlled by a 2100A Hewlett Packard mini-computer. The Adaptive Control Software has to operate in unison with an already available Numerical Control Software. Both these programmes are stored in the computer and the computer operates on them with the aid of the interrupt pulses received from the Time Base Generator located in the Controller. The Adaptive Control Software should be capable of optimising the milling process, that is enabling the milling machine to operate at the highest feed-rate without violating or overriding the maximum permissible values of the horizontal force and torque acting on the cutter. These maximum values of the force and torque are determined from the tool strength and capacities of the servo drives and spindle motor. Further, the machine should be able to arrive at the above feed-rate in the shortest possible time interval without causing cyclic variations in the feed-rate which could lead to an unstable system. The programme should be able to obtain ten samples of the parameters per revolution of the spindle. The feed-rate thus obtained (after comparing with the maximum and minimum feed-rates of the machine and making any corrections, if necessary) should be stored in a memory location accessible to the Numerical Control Programme. The instantaneous values of the force and torque are transmitted to the computer via the transducers attached on the spindle of the machine and the Analog-to-Digital Processor, therefore, the Adaptive Control Software will have to communicate with the Analog-to-Digital Processor in order to receive the values of the forces and torque. Thus the above mentioned requirements will have to be met by this piece of software. With this end in view, the following algorithm was developed. The algorithm consists of two portions, namely, the Data Reading Routine and the Policy Routine. The former accepts the two horizontal forces (which are phase shifted by 90°) and the torque acting on the cutter by communicating with the Analog-to-Digital Processor. However, all these three parameters are received through the same channel from the Analog-to-Digital Processor as such a method of identifying the variables was necessary. For this purpose, the Data Reading Routine consists of software capable of communicating with the Analog-to-Digital Processor at time intervals of 10 m.sec. and receiving the data in a digital form, decoding the input and ascertaining which input parameter was received. The Policy Routine has two modes of operation viz., the constraint and optimizing modes. This routine ascertains the critical error and arrives at the new feed-rate depending on the Policy used. After checking the value of this feed-rate with the maximum and minimum feed-rates available on the machine (and corrections made if necessary), the suitable value of this feed-rate is stored in a memory location accessible to the Numerical Control programme. This gives the general structure of the Adaptive Control Algorithm developed in this project.

Master of Engineering (MEngr)

Thesis