基于自适应支持向量机的磨粒识别技术研究

Research on Wear Particle Recognition Based on Self-Adapting Support Vector Machine

通过对铁谱磨粒类型进行识别,可有效监测机械装备磨损状态,有利于尽早发现和消除故障隐患。对粒子群优化算法进行改进,采用改进的粒子群算法同时优化支持向量机的惩罚参数和核函数参数,建立了自适应磨粒识别模型。通过对磨粒样本进行仿真实验,识别正确率达到98%,并与BP神经网络方法进行对比,结果表明了该方法的有效性及优越性。

Through the pattern recognition of ferrograghy wear particle, the wear condition of mechanical equipment can be monitored effectively to prevent the malfunction problems. The raditional particle swarm optimization algorithm is improved, in addition, the improved PSO is used to optimize the error punish modulus and kernel function parameter, and the slf-adapting recogniyion model for wear particle was established. The method is applied in simulation, and the classification accuracy rate is 98% . The method is compared with the BP neural network, the results show superiority and effectivity of the new method.