| Precision grinding is widely used to attain good surface quality and high accuracy with relatively high efficiency and low cost.Applications of grinding technology can be found in most industrial sectors such as aerospace,automotive,shipbuilding,precision instruments,nuclear energy,optical electronics and semiconductor engineering.However,the grinding process is characterized by a huge number of irregular abrasive grains undergoing nonuniform wear.It is recognized as a complex and highly non-stationary process.Moreover,abrasive grains in grinding tools are high negative rake angle geometry.Hence,the grinding process requires large energy per unit volume of material removed as compared with other mechanical machining processes.High grinding energy consumption indicates high machining force and large friction thermal,which leads to fast wear of the grinding wheel/abrasive tool as well as potential surface and subsurface damage of the workpiece,e.g.,grinding burn.This work aimed at common difficulties in current grinding practices by the processing experience of operators,e.g.,high energy consumption,frequent burn,poor surface integrity,unstable grinding performance and difficult prediction of wheel tool wear,grinding overheat.A portable power monitoring and intelligent decision-making system with specially designed grinding analytical software was developed.Grinding experiments of two different material,No.45 steel and quartz ceramics,were carried on,and power signal in machining process was monitored.BP neural network and particle swarm optimization algorithms were employed and improved to optimize grinding energy consumption and surface roughness.Grinding process strategies in high efficiency and low consumption was proposed.The main research contents are as follows:(1)The intelligent monitoring hardware platform acquiring real-time grinding power signal was setup based on SMART-B818 Ⅲ grinder.Real-time sampling method of spindle power signal was designed by use of Portable Power sensor(Portable Power Cell Model-3,PPC-3).A kind of data acquisition system,NI c DAQ 9174 and NI 9203,was selected to convert analog signal to digital format and demodulate to voltage or current signal.Monitoring data was transmitted by USB.Then intelligent monitoring of grinding power and energy consumption was completed.(2)Full factorial and orthogonal grinding experiments were arranged for the two different materials,No.45 steel and quartz ceramics.The grinding characteristic signal under different processing conditions was collected to calculate active power consumption in grinding,and the surface roughness of the grinding workpieces were measured.The relationship between the characteristic signal and different processing condition was explored,and the influence rules of different grinding parameters on the surface roughness and grinding energy consumption of the workpieces were analyzed.(3)A prediction model of grinding energy consumption was established by using three-layer error back propagation neural network,in which wheel velocity,feeding velocity and grinding depth were inputs.The optimal process parameters were iterative optimized to get minimum energy consumption,taking the improved the particle swarm by introducing the dynamic inertia weight as optimization algorithm and the prediction model of BP neural network as the fitness function.(4)The prediction method of grinding wheel wear and grinding burn was studied based on the eigenvalue of power signal varying with grinding.An analysis and decisionmaking system was developed in Lab VIEW software,with function of data acquisition,filtering procession,calculation of energy consumption,condition monitor and comparison of grinding wheel,prediction of grinding burn,optimization of grinding parameters,et.al.Intelligent online warning and steerable controlling in grinding process with high efficiency and low energy consumption was validated through further tests. |
PDF Full Text Request