Flow stress during hot deformation is essentially controlled by the chemistry of material, initial microstructure/texture, strain, strain rate, strain path, stress triaxility and the temperature of deformation. A comprehensive literature survey has been performed to realize this fact completely. In the present research, a neural network model under Bayesian framework has been created to correlate the complex relationship between flow stress with its influencing parameters in various grades of zirconium alloys at different deformation conditions. The network has been trained with published experimental database obtained from the different hot deformation experiments of zirconium alloys. Performance of the model has been evaluated; and excellent agreements between experimentally measured and model calculated data are obtained. The analysis permits the estimation of error bars whose magnitude strongly depends on their position in the input space. The model has been employed to different grades of zirconium alloys to confirm that the predictions are reasonably accurate in the context of basic metallurgical/solid mechanics theories and principles. The work has clearly identified the regions of the input space where further experiments should be encouraged and necessary. This model will be useful to design and manufacture the new generation zirconium alloys in future for the nuclear power plant components according to the needs of nuclear engineers/scientists by controlling the alloying elements and other possible conditions. The result shows that neural computation is a very effective tool to model the complex \(\textit{non-linear}\) behaviour of flow stress of different zirconium alloys under any deformation conditions.

热变形过程中的流动应力基本上由材料的化学性质，初始微观结构/组织，应变，应变速率，应变路径，应力三轴性和变形温度控制。已经进行了全面的文献调查，以完全认识到这一事实。在本研究中，建立了贝叶斯框架下的神经网络模型，以将不同应力条件下不同等级的锆合金中流动应力与其影响参数之间的复杂关系关联起来。该网络已使用从锆合金的不同热变形实验获得的公开实验数据库进行了训练。该模型的性能已得到评估；并获得了实验测量和模型计算数据之间的极佳一致性。该分析允许估计误差棒，误差棒的大小很大程度上取决于其在输入空间中的位置。该模型已用于不同等级的锆合金，以确认在基本冶金/固体力学理论和原理的基础上预测是合理准确的。这项工作清楚地确定了输入空间的区域，应该鼓励和必要进行进一步的实验。通过控制合金元素和其他可能的条件，该模型对于将来根据核工程师/科学家的需求设计和制造用于核电站部件的新一代锆合金很有用。结果表明，神经计算是建模复杂模型的有效工具\（\ textit {非线性} \）不同锆合金的流变应力的任何变形的条件下的行为。