TAE Technologies Inc., PO Box 7010, Rancho Santa Margarita, CA, 92688, USA.
Nat Commun. 2018 Feb 15;9(1):691. doi: 10.1038/s41467-018-03110-5.
Active control of field reversed configuration (FRC) devices requires a method to determine the flux surface geometry and dynamic properties of the plasma during both transient and steady-state conditions. The current tomography (CT) method uses Bayesian inference to determine the plasma current density distribution using both the information from magnetic measurements and a physics model in the prior. Here we show that, from the inferred current sources, the FRC topology and its axial stability properties are readily obtained. When Gaussian process priors are used and the forward model is linear, the CT solution involves non-iterative matrix operations and is then ideally suited for deterministic real-time applications. Because no equilibrium assumptions are used in this case, inference of plasma topology and dynamics up to Alfvenic frequencies then becomes possible. Inference results for the C-2U device exhibit self-consistency of motions and forces during Alfvenic transients, as well as good agreement with plasma imaging diagnostics.
主动控制场反转配置（FRC）设备需要一种方法来确定在瞬态和稳态条件下的等离子体的磁面几何形状和动态特性。电流层析成像（CT）方法使用贝叶斯推理，通过来自磁测量的信息和先验中的物理模型来确定等离子体电流密度分布。这里我们表明，从推断出的电流源中，可以很容易地得到 FRC 的拓扑结构及其轴向稳定性特性。当使用高斯过程先验并且正向模型是线性时，CT 解涉及非迭代矩阵运算，因此非常适合确定性实时应用。由于在这种情况下不使用平衡假设，因此可以推断出直到阿尔芬频率的等离子体拓扑结构和动力学。C-2U 装置的推断结果显示了阿尔芬瞬变期间运动和力的自洽性，并且与等离子体成像诊断很好地吻合。
