Real-time feedback control of the impurity emission front in tokamak divertor plasmas.

In magnetic confinement thermonuclear fusion the exhaust of heat and particles from the core remains a major challenge. Heat and particles leaving the core are transported via open magnetic field lines to a region of the reactor wall, called the divertor. Unabated, the heat and particle fluxes may become intolerable and damage the divertor. Controlled 'plasma detachment', a regime characterized by both a large reduction in plasma pressure and temperature at the divertor target, is required to reduce fluxes onto the divertor. Here we report a systematic approach towards achieving this critical need through feedback control of impurity emission front locations and its experimental demonstration. Our approach comprises a combination of real-time plasma diagnostic utilization, dynamic characterization of the plasma in proximity to the divertor, and efficient, reliable offline feedback controller design.

Minimally invasive in vivo human lung tissue bioimpedance measurements during the bronchoscopy procedure.

Respiratory diseases, which include diseases of the lung, pleura, bronchial tree, trachea, upper respiratory tract and of the respiratory muscles and nerves, are a common and important cause of illness and death among the population. Experimental evidences have shown that tissue lesions have different electrical properties compared with normal tissue. Therefore, lung tissues lesions may be differentiated from lung normal tissue by comparing the tissue passive electrical properties. The manuscript reports a feasibility study for minimally invasive in vivo human lung tissue tetrapolar bioimpedance measurements using a catheter during the bronchoscopy procedure based on multisine Electrical Impedance Spectroscopy (EIS) at 10 kHz - 1 MHz.

Learning neural networks with noisy inputs using the errors-in-variables approach.

Currently, most learning algorithms for neural-network modeling are based on the output error approach, using a least squares cost function. This method provides good results when the network is trained with noisy output data and known inputs. Special care must be taken, however, when training the network with noisy input data, or when both inputs and outputs contain noise. This paper proposes a novel cost function for learning NN with noisy inputs, based on the errors-in-variables stochastic framework. A learning scheme is presented and examples are given demonstrating the improved performance in neural-network curve fitting, at the cost of increased computation time.