Study On Electrochemical Properties Of Hydrogen Storage Alloys Influenced By Their Compositions Via SVR

As the environmental problem became increasingly serious, the hydrogen as aclean energy has attracted widespread concerns. But how to enhance the hydrogenstorage capacity is still the most important issue in the use of hydrogen energy. Amongvarious hydrogen storage materials, hydrogen storage alloys have been extensivelyinvestigated because of its steady electrochemical property, and have been utilized asthe electrode materials of rechargeable batteries （Ni-MH batteries） in the industry.However, some of the shortcomings and deficiencies of the hydrogen storage alloy arestill need to be researched and improved. There are two common approaches that can beapplied to improve the performance of hydrogen storage alloy. One is the elementsubstitution, i.e., using different chemical elements to replace the original elements inthe A-side or B-side of the hydrogen storage alloy; the other is the surface modificationmethod. Both of above strategies require a large number of experiments to find thesuitable chemical substitution elements, or to optimize the experimental parameters inthe surface modification process. Therefore, a reliable modeling approach that can beemployed to guide the experimental design and analyze the experimental data hasbecome an indispensable part of the research.In this thesis, the influences of the contents of alternative elements in the hydrogenstorage alloy, and the experimental parameters of copper plating process on thehydrogen storage capacity and electrochemical property were investigated by usingsupport vector regression （SVR） modeling approach combined with particle swarmoptimization （PSO）. The modeling results were validated by leave one out crossvalidation and the verifications reveal that SVR approach possesses excellent predictionaccuracy and reliability. And the comparative analysis between the calculation results ofSVR model and those of the conventional quadratic polynomial model demonstrate thatthe accuracy of SVR model is superior to that of the traditional polynomial model whendealing with relatively small size of experimental data. According to the establishedSVR models, we investigated the influence of the contents of substitution elements （Ce,Pr, Nd） on the maximum discharge capacity and high rate dischargeability, also studiedthe effect of substitution of Cr with Fe and Mn on the maximum and effective hydrogenstorage capacity of Ti-Cr-V hydrogen storage alloys, and optimize the experimentalparameters （temperature, pH value, Ni2+concentration） of copper plating on the AB₅ alloy surface. In addition, the effect of experimental conditions on the purity andaverage particle size of the hydroxyapatite powder in the sol-gel experimental processwas investigated as well. The results show that, in the experimental researches of thehydrogen storage alloy, support vector regression is an effective modeling technique toanalyze the experimental data, and play a significant role in the optimization of theexperimental design.