Multi-objective Process Parameters Optimization Of Granite Robot Machining Based On Neural Network

Store carving,as an art with great artistic value and economic value,is the inheritance of historical culture.As living standards improve,stone carving has become ubiquitous in modern life,and the traditional way of stone carving is being replaced by automatic machining.Compared with other automatic equipment,because of higher degrees of freedom and flexibility,robots are more suitable for stone carving,and have become the future development trend of stone carving equipment.Processing parameters have a very high impact on machining efficiency and quality,which are still determined based on technician's experience and some factors can't be taken into account,such as efficiency and quality etc.Shanxi Black,one kind of high hardness granite,which is more difficult to carve than marble,is increasingly being used in stone carving.Therefore,Shanxi Black was selected as the object of this paper to search the best processing parameters by studying the process of robot machining special-shaped granite.In this subject,the process of special-shaped granite robot machining was divided into roughing machining and finish machining.The influence of processing parameters,including spindle speed,depth of cut,width of cut and feed rate on force and size error was studied by orthogonal experiment of rough machining.After the experiment,the actual material removal rate was analyzed,and the debris were observed to study the manner of material removal of different processing parameters.Then the factor experimental design was carried out based on orthogonal experiment.Factor experimental design was also carried out to investigate the impact of processing parameters on contour error and surface roughness of finish machining.After that,the results of the experiments were applied to artificial-neural-network training and verification.Based on full factor experiments,the matlab toolbox function was used to construct predictive models for processing parameters and output results of rough machining and finish machining based on BP neutral network.The validation test indicated that the predictive models can meet machining requirements.The parameters optimization models were proposed and solved by genetic algorithm based on BP neural predicting network.The validation test showed that the efficiency of rough machining has increased 38.3%,the surface roughness and contour error have decreased 15.2% and 50.1% respectively.In this study,the predictive models and optimization models were built to optimize processing parameters of robot carving granite,and the validation test indicated that the efficiency and quality were both improved.Which can change the situation that processing parameters are determined artificially,and is conductive to beef up the spread and application of robot in stone carving.