Non-Invasive Temperature Estimation and Control in Titanium Casting

本项目旨在开发一种精确的、非侵入性的方法，用于真空电弧重熔（VAR）铸造过程中熔融钛温度的估计;以及通过调整电弧参数（电压和间隙）和浇注起始时刻来确保目标浇注温度的控制策略。估计和控制算法将采用VAR过程中热能守恒的组成经验模型。该模型将通过分析过程中测量的信号（例如电弧功率或冷却坩埚和电极夹持器的水的温度和流量）来不断更新。特别地，在每次浇注期间，将通过红外高温计测量流过坩埚唇缘的熔体的实际温度，以关闭估计过程中的反馈回路，并确保下一次执行过程的准确温度预测。 在线识别关键的物理特性，如熔体和冷却水之间的热阻，将有助于预测控制电弧功率，以实现目标浇注温度，尽管在该过程中存在强烈的干扰。 一个原型系统将在一家制造用于航空、国防、太空探索和医疗应用的钛铸件的工厂开发和实施。本计画将产生一个非侵入式温度估测与控制系统的原型，对钛铸件的品质控制有价值。 所开发的算法和系统架构也可适用于其他高性能铸造工艺。

This project is to develop an accurate, non-invasive method for the estimation of molten titanium temperature in the Vacuum Arc Remelted (VAR) casting process; and, a control strategy to assure target pour temperatures by adjusting arc parameters (voltage and gap) and the pour initiation instance. The estimation and control algorithms will employ a constitutive-empirical model of thermal energy conservation in the VAR process. This model will be continuously updated by analyzing in-process measured signals, such as the arc power, or temperature and flow of water that cools the crucible and electrode gripper. In particular, the actual temperature of the melt flowing over the crucible lip will be measured during each pour by infrared pyrometers to close the feedback loop in the estimation procedure and assure accurate temperature prediction for the next execution of the process. On-line identification of crucial physical properties, such as the thermal resistance between the melt and cooling water, will facilitate predictive control of the arc power to accomplish target pour temperatures despite strong disturbances present in the process. A prototype system will be developed and implemented in a plant that manufactures titanium castings for aviation, defense, space exploration and medical applications. This project should produce the prototype of a non-invasive temperature estimation and control system valuable to quality control in titanium casting. The developed algorithms and system architecture may be applicable to other high performance casting processes as well.