Based On Improved Bp Neural Network Grinding Quenched Numerical Simulation And Experimental Study

In manufacture process of many steel workpieces, surface quench is usually used to improve the surface capability and fatigue strength of the workpieces. Grinding is widely used in machining, but the grinding heat is regarded as a negative factor. Grinding quench is a technology that the surface of workpiceces is quenched using the grinding heat in grinding. Author summarizes the research of grinding quench and grinding simulation currently, and research works are given in the following.BP arithmetic is improved, and a software is developed using improved BP arithmetic, which can simulate the condition of workpieces in grinding. The software can forecast the grinding results and select the grinding parameters that are trained. The grinding parameters include grinding wheel material, grinding wheel granularity, workpiece feed speed, grinding wheel speed, grinding fluid dosage, grinding depth and workpiece material. The grinding results include quench depth, grinding top temperature, workpiece surface residual stress, workpiece surface degree of roughness, workpiece case hardness, grinding force of perpendicular direction and long direction. The software includes data management function, BP arithmetic function, quench results forecast function and parameters auto-select function.The train data of BP networks are obtained by experiment, and several grinding parameters are selected to process 40Cr and AISI SAE1045 steel. The grinding force is obtained online using piezoelectricity milling and grinding ergometer, electric charge amplifier, A/D card and data collection software. The grinding temperature is obtained online using thermoelectric couple, voltage amplifier, A/D card and data collection software. The case hardness, surface quench depth, surface degree of roughness are gathered after grinding. All the data are inputed to the software.To contrast the simulation data with the experimental data, the results show that the simulation data getting from improved BP arithmetic have a high precision. It shows that it is feasible to research grinding quench using BP arithmetic, and it will have a good foreground.