Dyanmic Energy Partition Calculation And Processes Monitoring Of Robotic Belt Grinding Of Inconel 718 Alloy

With the development and upgrading of manufacturing industris,it has gradually become a trend to use robots to automate complex machining processes,which not only frees workers from harsh and hazardous working environments,but also improve materials processing efficiency.As a flexible and efficient manufacturing process,abrasive belt grinding plays an irreplaceable role in machining difficult-to-cut materials,such as Inconel 718 alloy,and final processing of the complex curved workpieces.However,complicated energy partition,varying belt condition and indeterminate local material removal amount and heat flux distribution caused by implicit local contact status result in difficultis in shape and performance control of robotic belt grinding of Inconel 718 alloy.There is neither systematical research on energy partition calculation nor practicable method to achieve local material removal amount and heat flux distribution online monitoring upon the robotic belt grinding of Inconel 718 alloy.Therefore,aiming at above problem,this thesis systematically and comprehensively analyzed the dynamic robotic belt grinding process from grinding effects evaluation of one grain to thermal energy partitioning with respect to energy partition.On this basis,this thesis acquired multi gridning signals and adopted signal processing technologies,FEM and machining learning technologies to establish grinding process monitoring models which lay the foundation for the subsequent shape and performance control.The main research contents and conclusions of this thesis are as follows:(1)This thesis analyzed the dynamic robotic belt grinding process from the perspective of grinding effects and thermal aspects.Grinding effects are distinguished by combination of single grain scratch tests and force balance of one grain in view of dynamic and elastic contact conditions.Thermal aspects are obtained by a fusion method of FEM and an optimization algorithm.Then,by utilizing the iterative approach,heat accumulating effect and temperature dependent hardness of workpieces are taken into account and the dynamic energy partition is calculated in a continuous grinding process.Validations on three workpieces(respectively made by Inconel718,SUS304 and AA6061-T6)prove that the proposed dynamic energy partition calculation method is effective and accurate.Analysis and calculation results show that heat accumulation is found in grains and reduces the heat dissipation capacity of the belt as the grinding processes continues.Additionally,excellent heatconducting capabilities of the belt grinding system leads to a low grinding temperature.(2)This thesis proposed a novel method based on multi-sensors and machine learning techniques to achieve online dynamic heat input monitoring.Firstly,comparison of the dynamic and static heat input was performed for illustrating the necessity of online heat input monitoring.Secondly,through analyzing correlations bwtween grinding signals(sound and force)and the heat input,a hybrid feature processing method and a BADS-LSSVM(Bayesian adaptive direct research-least squares support vector machine)based model were proposed to achieve heat input monitoring.Test results show that the proposed model has a mean accuracy of no less than 96.7 %,a computed temperature error of ± 6 °C in a complete grinding pass,and takes about 0.6 s for each calculation.With this new method,the subsequent calculation of the heat flux distribution based on contact status can be achieved.(3)By imitating the principle of human eyes to recognize abrasive belt conditions,images of the belt were collected in situ.They images are processed by Bilateral Gabor filter and K-means clustering to extract two stastical features,namely,the number of worn grains and the total area of worn grains.The thesis redefined the belt condition and proposed a belt subregion condition recognization method which is independent of grinding parameters.Verifications based on different types of abrasive belts show that the proposed belt subregion condition recognization method based on vision has high accuracy,generality and stability.(4)After contrastive analysis of performances of FEM,MDR and BEM in obtaining the local contact status in belt grinding,a novel method (7 + BEM,which is the combination of the visual aid to obtain contact width and BEM,was proposed to calculate the contact status quickly.The proposed mothed can obtain the contact status of the point cloud within 18 mm × 18 mm area with the accuracy of 0.2 mm.Compared with the FEM calculation results,the maximum error of the contact status obtained by (7 + BEM is no more than6.0 %.Based on above contact status,considering that material removal reacts on the subsequent contact status,iterative calculation method was proposed to calculate local material removal amount based on material removal rate under a certen abrasive belt condition and update point cloud data.The 5-second grinding experiment shows that the the maximum error of the calculated local material removal amount is within 0.2 mm.Meanwhile,based on above contact status and heat input online monitoring model,this thesis has achieved the calculation of the heat flux distribution.In addition,simulation and experiments found that the heat flux distribution has a linear relationship with the square of the contact pressure,and the average deviation between measured temperature and simulated temperature based on the calculated heat flux distribution is less than 13.4 %.