摘要
为了实现智能制造,在运行机器学习算法的过程中,需要对运算环境、运算能力和程序算法做出优化,以面对日渐增长的数据量和处理压力。将具备自学习能力的算法从传统的.Net平台迁移至针对机器学习优化设计的Tensor Flow平台,证实该平台有能力承担故障诊断的计算,并验证了该平台在面对大量数据时可以提供更可靠的处理能力和更优秀的弹性计算能力,证明了以Tensor Flow为基础开发数控机床故障诊断系统的可行性。
To achieve sustainability and the vision of smart manufacturing, algorithms should be constantly optimized along with the computing environment and the computing units. In this analysis,the traditional computing structure. Net was replaced by machine learning oriented computing platform Tensor Flow,which shows the capability of building fault diagnosis system on Tensor Flow. This new platform was also proved to be capable of serving better computing performance and better scalability. It is feasible to rewrite current algorithms on Tensor Flow and design a new diagnosis system for CNC machines on basis of Tensor Flow.
出处
《机电一体化》
2018年第2期36-41,共6页
Mechatronics