为了提高电火花加工放电间隙的实时识别准确率,文中提出了一种基于 FPGA 高速数据采集的神经网络的放电间隙在线识别方案.利用 FPGA 和 AD 模块对加工间隙电压数据进行高速采集,通过FPGA内部的FIFO和串口协同实现周期性数据传输.在Python的Pytorch库构建了CNN-GRU 融合神经网络模型,对所采集的数据进行在线识别,验证了该采集系统的可行性.实验结果表明:文中方案对加工电压脉冲的离线识别准确率达到了 97.35%,优于CNN和GRU方法,满足了电火花加工对间隙电压识别的高精度要求,实现了神经网络的实时在线识别放电间隙.
WEDM Online Pulse Recognition Based on FPGA High-Speed Data Acquisition
In order to improve the accuracy of real-time recognition of EDM discharge gap,the paper presents a neural network based on FPGA high-speed data acquisition as an online recognition scheme for discharge gap.High-speed acquisition of machining gap voltage data is carried out with FPGA and AD modules,and periodic data transmission is realized through the internal FIFO of FPGA and serial port cooperatively.A CNN-GRU fusion neural network model was constructed in Python's Pytorch library to recognize the collected data online,which verified the feasibility of the acquisition system.The experimental results show that the offline recognition accuracy of the machining voltage pulse by the scheme proposed in the paper reaches 97.35%,higher than that of the CNN and GRU methods.It is concluded that this scheme meets the high-precision requirements of gap voltage recognition in EDM machining and realizes the real-time on-line recognition of the discharge gap by the neural network.
electrical discharge machiningfield programmable gate arraydischarge gap identificationhigh-speed data acquisition
NETL
NSTL
万方数据
