The topology of interactions in network dynamical systems fundamentally underlies their function. Accelerating technological progress creates massively available data about collective nonlinear dynamics in physical, biological, and technological systems. Detecting direct interaction patterns from those dynamics still constitutes a major open problem. In particular, current nonlinear dynamics approaches mostly require to know a priori a model of the (often high dimensional) system dynamics. Here we develop a model-independent framework for inferring direct interactions solely from recording the nonlinear collective dynamics generated. Introducing an explicit dependency matrix in combination with a block-orthogonal regression algorithm, the approach works reliably across many dynamical regimes, including transient dynamics toward steady states, periodic and non-periodic dynamics, and chaos. Together with its capabilities to reveal network (two point) as well as hypernetwork (e.g., three point) interactions, this framework may thus open up nonlinear dynamics options of inferring direct interaction patterns across systems where no model is known.

The topology of interactions in network dynamical systems fundamentally underlies their function. Accelerating technological progress creates massively available data about collective nonlinear dynamics in physical, biological, and technological systems. Detecting direct interaction patterns from those dynamics still constitutes a major open problem. In particular, current nonlinear dynamics approaches mostly require to know a priori a model of the (often high dimensional) system dynamics. Here we develop a model-independent framework for inferring direct interactions solely from recording the nonlinear collective dynamics generated. Introducing an explicit dependency matrix in combination with a block-orthogonal regression algorithm, the approach works reliably across many dynamical regimes, including transient dynamics toward steady states, periodic and non-periodic dynamics, and chaos. Together with its capabilities to reveal network (two point) as well as hypernetwork (e.g., three point) interactions, this framework may thus open up nonlinear dynamics options of inferring direct interaction patterns across systems where no model is known.

<p>This work focuses on the collection and analysis of Lithium-ion battery operational and temperature data during nail penetration testing through two different experimental approaches. Raman spectroscopy, machine learning, and internal temperature sensors are used to collect and analyze data to further investigate the effects on cell operation during and after nail penetrations, and the feasibility of using this data to predict future performance.</p> <p><br></p> <p>The first section of this work analyzes the effects on continued operation of a small Lithium-ion prismatic cell after nail penetration. Raman spectroscopy is used to examine the effects on the anode and cathode materials of cells that are cycled for different amounts of time after a nail puncture. Incremental capacity analysis is then used to corroborate the findings from the Raman analysis. The study finds that the operational capacity and lifetime of cells is greatly reduced due to the accelerated degradation caused by loss of material, uneven current distribution, and exposure to atmosphere. This leads into the study of using the magnitude and corresponding voltage of incremental capacity peaks after nail puncture to forecast the operation of damaged cells. A Gaussian process regression is used to predict discharge capacity of different cells that experience the same type of nail puncture. The results from this study show that the method is capable of making accurate predictions of cell discharge capacity even with the higher rate of variance in operation after nail puncture, showing the method of prediction has the potential to be implemented in devices such as battery management systems.</p> <p><br></p> <p>The second section of this work proposes a method of inserting temperature sensors into commercially-available cylindrical cells to directly obtain internal temperature readings. Characterization tests are used to determine the effect on the operability of the modified cells after the sensors are inserted, and lifetime cycle testing is implemented to determine the long-term effects on cell performance. The results show the sensor insertion causes a small reduction in operational performance, and lifetime cycle testing shows the cells can operate near their optimal output for approximately 100-150 cycles. Modified cells are then used to monitor internal temperatures during nail penetration tests and how the amount of aging affects the temperature response. The results show that more aging in a cell causes higher temperatures during nail puncture, as well as a larger difference between internal and external temperatures, due mostly to the larger contribution of Joule heating caused by increased internal resistance.</p>