In the machining process, the machining performance which mainly refers to machined surface quality and cutting forces is hard to predict under different tool wear status. In this work, an improved case-based reasoning (ICBR) method is proposed to predict both the cutting force and machined surface roughness. With the emergence of new problem, ICBR method obtains solutions to new problem through case retrieval and reuse. In case retrieval stage, K similar cases to the new problem were retrieved using K nearest neighbor method. By means of the K similar cases, support vector regression machine (SVR) model was established to give the solution of the new problem in case of the reuse stage of ICBR method. Artificial neural network (ANN) was introduced to estimate the influence of machining parameters and tool wear on machining performance. As the ANN and SVR models contain unknown parameters, the novel particle swarm optimization algorithm was proposed to train these models for its capability of fast convergence and global optimum. The proposed ICBR method was used to predict the surface roughness and cutting force. The results showed the proposed ICBR method can give superior prediction accuracy and lower Mean square error than other popular intelligent models. Meantime, the ICBR method possesses good robustness and can be used for the actual machining process.

在加工过程中，很难预测在不同的刀具磨损状态下的加工性能，而加工性能主要是指加工的表面质量和切削力。在这项工作中，提出了一种改进的基于案例的推理（ICBR）方法来预测切削力和机加工表面粗糙度。随着新问题的出现，ICBR方法通过案例检索和重用获得了新问题的解决方案。在案例检索阶段，使用K最近邻方法检索了K个与新问题相似的案例。通过K在类似的情况下，建立支持向量回归机（SVR）模型以解决在ICBR方法重用阶段出现的新问题。引入了人工神经网络（ANN）来估计加工参数和刀具磨损对加工性能的影响。由于ANN和SVR模型包含未知参数，因此提出了一种新颖的粒子群优化算法来训练这些模型，使其具有快速收敛和全局最优的能力。提出的ICBR方法用于预测表面粗糙度和切削力。结果表明，与其他流行的智能模型相比，所提出的ICBR方法具有更高的预测精度和更低的均方误差。同时，ICBR方法具有良好的鲁棒性，可用于实际的加工过程。