Electrochemical Data Analysis and Simulation via Artificial Neural Intelligence for Pyroprocessing Safeguards Application

Pyroprocessing technology is a high-temperature method utilizing molten salts for used nuclear fuel treatment. The heart of this process is an electrorefiner (ER) which contains different fission, rare-earth, and transuranic chloride compositions during the operation. As a result, materials detection and accountability towards safeguards within the ER are extremely important in order to advance this technology. Development of a smart signal detection program toward pyroprocessing safeguards will require full understanding of massive ER systemic parameters. To obtain this desired goal, a novel electrochemical data analysis and simulation using an artificial neural intelligence (ANI) method has been developed and explored. One of the common electrochemical methods, cyclic voltammetry (CV), has been chosen to successfully test this approach as the first stepping stone. Massive collected data sets by Hoover (2014), over 77,000 data values, for 0.5 to 5 wt% of zirconium (Zr) in LiCl-KCl molten salt at 773 K with different scan rates has been considered to provide multi-variables (e.g. scan rate, voltage, current, time, and concentration) for ANI technique [1]. The computational code which has been performed using the commercial software MATLAB, weighted each inputs and contrasted with the sum of inputs to the threshold value to produce the outputs. That is, ANI can be used to mimic the system by driving the data sets of currents, potentials, concentrations, scan rates and process time through interrelation between variables to provide a current and potential simulated data set. Optimizing the ANI process is related to choice a proper number of second layer which is called intermediate hidden layer. Preliminary results demonstrate that the model can predict the current versus potential diagram with an error around 16%.