Technology


Research works on efficient recycling of EV battery systems


Te BattBox research project headed by Austrian technology firm, Fill, aims to develop cost-efficient recycling of EV battery systems. A consortium consisting of Fill, Graz


University of Technology, AVL and the Automotive Cluster is developing a multi- stage recycling concept in the BattBox project, since currently there’s lack of standardisation in the chemistry, design and dismantling of batteries. At each process stage, the exposed components will be dismantled and assessed for economic and safety-critical aspects. Tis structured approach is intended to develop methods for the extraction of high-quality and pure- grade raw materials with the best possible recyclability. “No e-vehicle battery is too difficult for us. Te more complex the batteries under


investigation, the more exciting and far-reaching the results of the project will be,” said Josef Ecker, Fill’s Project Manager. Te initiative is expected to


yield extensive insights and expertise into viably, economically and sustainably managing and handling battery systems aſter their end of life. Until now the activities in the area of batteries for e-mobility were mainly focused on the energy storage sector, but with the BattBox project, expertise is being extended to the entire battery life cycle, making an important contribution to their sustainable reuse and recycling. Te BattBox project has been recognised by the Austrian Research Promotion Agency


Fill team working on battery dismantling


(FFG) as a two-million-euro pioneering project. Te BattBox project is also a strategic


milestone for Fill, signifying the company’s foray into electric mobility.


Memristor-based Bayesian neural network enables better handling of real-world tasks


A team comprising CEA-Leti, CEA-List and two CNRS laboratories have developed a memristor-based Bayesian neural network for real-world tasks. Considering medical-diagnosis and other


safety-critical sensory-processing applications that require accurate decisions based on a small amount of noisy input data, the team’s study found that Bayesian neural networks excel at these tasks but require more energy


CEA-Leti wafer-level characterisation of memristor crossbar arrays using a prober


for their increased computations. Te increase is caused by the networks’ nature, which to be implemented in hardware needs a random number generator to store the synaptic weights – or, probability distributions. So instead of using random number


generators, the team used memristors because of their intrinsic variability, to store the probability distributions. A second major challenge was the network’s


inferencing, which require massive parallel multiply-and-accumulate (MAC) operations, which also requires a lot of power. “Tese operations are power-hungry when


carried out on CMOS-based ASICs and FPGAs, due to the shuttling of data between processor and memory,” said Elisa Vianello, CEA-Leti chief scientist of the study. “In our solution, we use crossbars of memristors that naturally implement the multiplication between the input voltage and the probabilistic synaptic weight through Ohm’s law and the accumulation through Kirchhoff’s current law, to significantly lower power consumption.”


04 February 2024 www.electronicsworld.co.uk Te team also had to reconcile the


imperfect nature of memristors with the Bayesian neural networks, which led to a new training algorithm. “Our work overcomes the challenge


of blending the two with a new training algorithm – variational inference augmented by a ‘technological loss’ – that accommodates device non-idealities during the learning phase,” said Damien Querlioz, a CNSR scientist, who is also part of the team. “Our approach enables the Bayesian neural network to be compatible with the imperfections of our memristors.” For example, if a traditional neural


network trained to recognise cats and dogs is presented with an image of a giraffe, it misclassifies it as a cat or a dog, said Vianello. “In contrast, a Bayesian neural network would respond: ‘I am not entirely sure what this is because I have never seen it.’ In critical environments like medical diagnosis, where incorrect predictions can have severe consequences, this uncertainty-capturing ability becomes crucial.”


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44