Tungsten-based (W-based) nanoparticles are produced through electrochemical spark erosion process. In this investigation, the parametric effects of voltage, tool rotation and pulse on time on production rate of W-based nanoparticles are analyzed. The shape and size of the produced nanoparticles are controlled through proper controlling of the referred parameters. Small size particles are obtained with low voltage and pulse on time, but with high tool rotation speed. The ANN-predicted values of this study are in close agreement with the observed experimental values for all the test formulations. It can be concluded that the process optimization via ANN modeling has been found to be very efficient for determining the performance linked with the electrochemical spark erosion process. The devised neural network provided an average prediction error of 1.52% for training and 3.78% in case of testing. The formulated models can predict results which are in close agreement with the test results. The produced W-based nanoparticles are used for sensing the NO 2 and CO 2 gases.

中文翻译：

---

通过电火花腐蚀工艺生产 W 基纳米粒子及其在气体传感器中的实际应用的表征和优化

钨基（W 基）纳米颗粒是通过电化学火花腐蚀工艺生产的。在这项研究中，分析了电压、工具旋转和脉冲对时间的参数影响对 W 基纳米粒子的生产率的影响。通过适当控制所涉及的参数来控制所产生的纳米颗粒的形状和尺寸。使用低电压和准时脉冲获得小尺寸颗粒，但工具旋转速度高。本研究的人工神经网络预测值与所有测试配方的观察实验值非常一致。可以得出结论，已经发现通过 ANN 建模的过程优化对于确定与电化学火花腐蚀过程相关的性能非常有效。设计的神经网络提供的平均预测误差为 1.52% 的训练和 3.78% 的测试。公式化的模型可以预测与测试结果非常一致的结果。产生的基于 W 的纳米粒子用于感测 NO 2 和 CO 2 气体。