Search

In the cylindrical grinding process, the residual grinding allowance is left owing to the rubbing of the abrasive grains and the elastic deformation of the grinding system caused by the large normal grinding force. Therefore, a sizing gauge is generally used to monitor the workpiece diameter during the process. However, the sizing gauge disturbs the grinding operations, such as loading and unloading the workpiece. In this study, a new method for estimating the grinding stock of a ground workpiece is proposed. In the proposed method, the grinding stock can be estimated without a sizing gauge. The removed volume of the workpiece can be calculated from the power consumption of the wheel motor, which is easily measured. Before the grinding experiments, the specific grinding energy is determined from several grinding results. During the grinding process, the grinding stock of the ground workpiece was estimated using the determined specific grinding energy. It was confirmed that the grinding stock could be accurately estimated under several grinding conditions.

The form-shaping process is the main functional process performed by a machine tool. A theory of the form-shaping system (FSS) investigates interrelationships between, on the one hand, geometry and kinematics of the machine tool and, on the other, geometry and accuracy of the machined part. In this paper, the FSS theory is developed in the following ways: (1) the general approach spreads also to the FSS with parallel-type links (previously, only chain-type systems were discussed); (2) direct and inverse form-shaping problems are formulated.

Summary In this paper, a decision making process model for grinding operations is proposed. It has a multistage structure and consists of different two types of neural network : the Feed-Forward network and the Brain-State-in-a-Box network. As an application example of the model. the relationship between the dressing conditions and the surface roughness of ground components is simulated. The results indicate that the proposed model is capable of learning the stochastic data of surface roughness and recalling the dressing conditions which attains the required surface roughness. The simulated process can be considered to be quite similar to the decision making process of experienced operators.