To access ground truth degradation information, we simulatedcharge and discharge cycles of automotive lithium ion batteriesin their healthy and degrading states and used this informationto determine performance of an autoencoder-basedanomaly detector. The simulated degradation mechanism wasan abrupt increase in the battery’s rate of time-dependent capacityfade. The neural network topology was based on onedimensionalconvolutional layers. The decision-support system,based on the sequential probability ratio test, interpretedthe anomaly generated by the autoencoder. Detection timeand time to failure were the metrics used for performanceevaluation. Anomaly detection was evaluated on five differentsimulated progressions of damage to examine the effectsof driving profile randomness on performance of the anomalydetector.

Lithium ion battery modules have significant capacity left after their useful life in transportation applications. This empirical study successfully tested the used modules in secondary grid applications in laboratory conditions. The selection of the secondary application was based on the construction features of the modules and the growing need for storage in grid operations. Description of the laboratory setup is provided in the context of a critical practical constraint where the battery management system and the usage and health history are not available to the secondary battery integrator. Charge and discharge profiles were developed based upon applications for peak shaving and firming renewables. Techno-economic analysis was focused on peak shaving at the utility level, considering a growing need for an affordable and environmentally friendly replacement to the traditional solutions based on environmentally costly peaker plants. The analysis showed strong evidence that near-term and future storage markets will be characterized by a large mismatch between the demand and supply of reused batteries from automotive primary applications for peak-shaving purposes in the generation side. The paper includes a discussion on successful adoption of cascaded use of batteries and their potential to reduce both economic and environmental cost of peak shaving.

We developed a framework for the risk assessment of delaying the delivery of shipments to customers in the presence of incomplete information pertaining to a significant, e.g., weather-related, event that could cause substantial disruption. The approach was anchored in existing manual practices, but equipped with a mechanism for collecting critical data and incorporating it into decision-making, paving the path to gradual automation. Two key variables that affect the risk were: the likelihood of an event and the importance of the specific shipment. User-specified event likelihood, with elliptical spatial component, allowed the model to attach different probabilistic interpretations; uniform and Gaussian probability distributions were discussed, including possible paths for extensions. The framework development included a practical implementation in the Python scientific ecosystem. Although the framework was demonstrated in a prototype environment, the results clearly showed that the framework was quickly able to show scheduled and in-process shipments that were at risk of delay, while also providing a prioritized ranking of these shipments in order for personnel within the manufacturing organization to quickly implement mitigation actions and proactive communications with customers to ensure critical shipments were delivered when needed. Since the framework pulled in data from various business information systems, the framework proved to assist personnel to quickly identify potentially impacted shipments much faster than existing methods, which resulted in improved efficiency and customer satisfaction.

An NVIDIA Jetson graphical processing unit was evaluated for utilization in a health and usage monitoring system by computing vibration-based condition indicators, evaluating autoencoders for anomaly detectors, and comparing the computational performances to the related performance when the computations were performed on the central processing unit. The comparison included signal preprocessing computations. Two distinct cases of interest were considered with neural network autoencoders: model evaluation and model adaptation with limited training. The experiments found that computations associated with signal preprocessing and computing of condition indicators performed faster on the central processing unit, but neural network model evaluation and adaptation were faster on the graphical processing units. Utilizing the GPU capability of the Jetson Nano, it was determined that 42 accelerometer signals could be evaluated through an autoencoder per second.

No abstract