In this paper, anew intelligent adaptive methodology was proposed for the optimization ofcutting parameters in finish hard turning of AISI D2 using PSO as anoptimization technique and ANN and GP as modeling techniques.
Since itconsiderers the variations in machining process, which in this paper isprogressive tool wear, the methodology can be defined as an adaptiveoptimization system. The core of the system is composed of three process partitions:a genetic model for predicting the values of tool flank wear, a neural networkmodel for predicting the surface roughness and a PSO optimization unit to findoptimum cutting parameters in each step of cutting. The results canbe summarized as follows:1.
Surface roughness reduces withthe increase of cutting speed and decrease of feed. Tool flank wear alsoadversely affect the surface quality. The feed has greater influence on surfaceroughness followed by tool flank wear. The tool wear is influenced more by thecutting speed as compared to feed rate. Tool wear rate is intensive in earlymoments.
By passing the time, this rate decreases. By considering both the toollife and the surface roughness, a combination of optimum cutting parameters accordingto defined performance index was obtained as v=67.5 m/min and f=0.
425mm/rev.2. Comparing the results ofexperiments and results obtained by intelligent methods shows that geneticprogramming and artificial neural networks have unique ability and precision inmodeling of machining characteristics. Models developed for surface roughnessand tool wear using intelligent technique are very useful for predicting newexperiments and can be used reliably in adaptive optimization system. Closecorrelation between predicted and measured values was established previously. 3.
The adaptive optimizationtechnique presented in this paper can vary cutting conditions to operate atmaximum efficiency based on defined performance index that results in highermaterial removal rate accompanying with lower cutting costs. The results of comparisonsmade between proposed methodology and optimization with constant parameters showsthat the proposed method improved the sum of performance indexes by 40%compared to machining with constant optimum cutting parameters.Future workcould be directed to application of various intelligent models and optimizationtechniques to machining process optimization, investigation of adaptive controlwith optimization by implementation of various sensor systems to assess thevariation in cutting process, such as tool wear, and apply appropriate cuttingparameters accordingly, and performing adaptive optimization experiments withdifferent performance indexes and different constraints such as power,temperature, vibration and other machining characteristics.
