论文部分内容阅读
以Cr、Co、Al、Sr、RE的含量和制备方法为输入层节点参数,以腐蚀电位为输出层节点参数,构建了神经网络分析模型,并对预测能力进行试验验证,同时测试了模型选出的最优耐蚀性镁基储氢合金的性能。结果表明,镁基储氢合金耐蚀性神经网络分析模型的预测精度较高,半连续感应熔炼后再机械球磨的分步法制备出的Mg2NiCr0.3Sr0.1RE0.1具有最优耐蚀性,且吸放氢性能与Mg2Ni相当,循环稳定性明显优于Mg2Ni,循环20次后放电容量衰减率从81.3%下降至39.8%。
Taking the contents of Cr, Co, Al, Sr, RE and the preparation method as the parameters of the input layer node, the neural network analysis model was built based on the corrosion potential as the output layer node parameters. The predictive ability was tested and the model selection Out of the optimal corrosion resistance of magnesium-based hydrogen storage alloy performance. The results show that the predictive accuracy of the model for the corrosion resistance of Mg-based hydrogen storage alloys is high, and the optimal corrosion resistance of Mg2NiCr0.3Sr0.1RE0.1 prepared by the semi-continuous induction melting and ball milling step by step method is obtained. The performance of hydrogen absorption and desorption is comparable to that of Mg2Ni, and the cycle stability is obviously better than that of Mg2Ni. After 20 cycles, the decay rate of discharge capacity decreases from 81.3% to 39.8%.