Wind energy is one of the main drivers of the renewable energy market. The wind turbine gearbox, used as a multiplier - is one of the main components directly related to wind turbine efficiency and lifespan. Therefore, the strict control of the gearbox project and its manufacturing processes or even minor improvements in this component heavily and positively impact energy production/generation over time. This work diagnoses a 1.6 MW gearbox fracture mechanics to identify its main reason and how the chain of events took place, leading to the catastrophic failure. The diagnoses underwent numerical simulation (Finite Element Analysis - FEA) and further analysis of the lubrication system, bearings, planetary stage gears, helical stage gears, and high-speed shaft. In conclusion, although the numerical simulation showed high contact stresses between the teeth, the region with the crack nucleated was the top land (tip of the tooth). The most likely factors that led to the premature failure were the missed lubrication for the planetary bearings, lack of gear raw material cleanliness, and excessive case depth at the tip of the sun shaft gear teeth. With the sun gear, planet bearings, and planet gears broken in pieces, those small and large pieces dropped into the oil, between the gears, and into the tooth ring, ending up as a premature gearbox failure.

The ubiquitous deployment of Li-ion batteries (LIBs) in more demanding applications has reinforced the need to understand root-causes of thermal runaway. Herein we perform a forensic simulation of a real-case failure scenario, using localised heating of Li(Ni0.5Mn0.3Co0.2)O2 versus graphite 18650 cylindrical cells. This study determined the localised temperatures that would lead to venting and thermal runaway of these cells, as well as correlating the gases produced as a function of degradation pathway. Catastrophic failure, involving melting (with internal cell temperatures exceeding 1085°C), deformation and ejection of cell componentry, was induced by locally applying 200 °C and 250 °C to a fully charged cell. Conversely, catastrophic failure was not observed when the same temperatures were applied to the cells at lower state of charge (SOC). This work highlights the importance of SOC, chemistry and heat in driving the thermal failure mode of Ni-rich LIB cells, allowing for a better understanding of battery safety and associated design improvements.

The use of mindfulness techniques as a cutting-edge tactic to reduce catastrophic risk in the chemical sector is examined in this study. Constructed from fundamental design principles and conventional safety measures, mindfulness provides an additional method for addressing the behavioural and cognitive aspects of risk management. The study looks at the safety paradigms' development, the historical background of significant industrial accidents, and the theoretical foundations of mindfulness in high-reliability businesses. Through the promotion of increased consciousness, better judgement, and better communication, mindfulness has the potential to build a safer culture that is more resilient. The study investigates practical implementation challenges and the measurable outcomes of mindfulness-based interventions in high-risk industrial settings. This multidisciplinary study adds to the continuing conversation about industrial safety by offering insightful information to safety experts, business executives, and legislators. The necessity for additional empirical research to measure the effectiveness of mindfulness therapies in various chemical industry situations is emphasised in the paper's conclusion.

Deep learning models are part of the family of artificial neural networks and, as such, it suffers of catastrophic interference when they learn sequentially. In addition, most of these models have a rigid architecture which prevents the incremental learning of new classes. To overcome these drawbacks, in this article we propose the Self-Improving Generative Artificial Neural Network (SIGANN), a type of end-to-end Deep Neural Network system which is able to ease the catastrophic forgetting problem when leaning new classes. In this method, we introduce a novelty detection model to automatically detect samples of new classes, moreover an adversarial auto-encoder is used to produce samples of previous classes. This system consists of three main modules: a classifier module implemented using a Deep Convolutional Neural Network, a generator module based on an adversarial autoencoder; and a novelty detection module, implemented using an OpenMax activation function. Using the EMNIST data set, the model was trained incrementally, starting with a small set of classes. The results of the simulation show that SIGANN is able to retain previous knowledge with a gradual forgetfulness for each learning sequence. Moreover, SIGANN can detect new classes that are hidden in the data and, therefore, proceed with incremental class learning.

When attempting to quantify future harms caused by carbon emissions and to set appropriate energy policies, it has been argued that the most important metric is the number of human deaths caused by climate change. Several studies have attempted to overcome the uncertainties associated with such forecasting. In this article, approaches to estimating future human deaths tolls from climate change are compared and synthesized, and implications for energy policy are considered. Several studies are consistent with the “1000-ton rule,” according to which a future person is killed every time 1000 tons of fossil carbon are burned (order-of-magnitude estimate). If warming reaches or exceeds 2°C this century, mainly richer humans will be responsible for killing roughly 1 billion mainly poorer humans through anthropogenic global warming. Such mass manslaughter is clearly unacceptable. On this basis, relatively aggressive energy policies are summarized that would enable immediate and substantive decreases to carbon emissions. The limitations to such calculations are outlined and future work is recommended to accelerate the decarbonization of the global economy while minimizing the number of sacrificed human lives.

This paper examines cognitive biases which affect the ability of decision makers to make rational decisions in an organizational context. The motivation for this analysis begins with the observation of catastrophic accidents caused by human error but in an organizational context. This paper expands on the concept of cognitive bias to define organizational biases which are the factors that affect decisions in an organizational context. The paper distinguishes between organizational biases, which are the focus of this paper, and individual biases, which are biases experienced by individuals but may have organizational consequences. The purpose of this paper is to identify methods to mitigate the risks of organizational accidents, accidents which involve many people operating at different levels of an organization. The methodology is to identify those decisions that would address the specific organizational biases. The focus of this paper is the decisions for mitigating the risks associated with decisions in an organizational context. Results are shown for seven organizational biases, six specific case studies, and four decision options. This paper concludes that organizational biases are intrinsically different from individual biases and that these differences lead to different decision options from those that mitigate individual biases; however, they may exist concurrently.

To safeguard against meat supply shortages during pandemics or other catastrophes, this study analyzed the potential to provide the average household’s entire protein consumption using either soybean production or distributed meat production at the household level in the U.S. with: 1) pasture-fed rabbits, 2) pellet and hay-fed rabbits, or 3) pellet-fed chickens. Only using the average backyard resources, soybean cultivation can provide 80%-160% of household protein and 0%-50% of a household’s protein needs can be provided by pasture-fed rabbits using only the yard grass as feed. If external supplementation of feed is available, raising 52 chickens while also harvesting the concomitant eggs or alternately 107 grain-fed rabbits can meet 100% of an average household’s protein requirements. These results show that resilience to future pandemics and challenges associated with growing meat demands can be incrementally addressed through backyard distributed protein production. Backyard production of chicken meat, eggs, and rabbit meat reduces environmental costs of protein due to savings in production, transportation, and refrigeration of meat products and even more so with soybeans. Generally, distributed production of protein was found to be economically competitive with centralized production of meat if distributed labor costs were ignored.

Continual learning (CL), a.k.a lifelong learning, is an emerging research topic that has been attracting increasing interest in the field of machine learning. With human activity recognition (HAR) playing a key role in enabling numerous real-world applications, an essential step towards the long-term deployment of such systems is to extend the activity model to dynamically adapt to changes in people’s everyday behavior. Current research in CL applied to HAR domain is still under-explored with researchers exploring existing methods developed for computer vision in HAR. Moreover, analysis has so far focused on task-incremental or class-incremental learning paradigms where task boundaries are known. This impedes the applicability of such methods for real-world systems. To push this field forward, we build on recent advances in the area of continual learning and design a lifelong adaptive learning framework using Prototypical Networks, LAPNet-HAR, that processes sensor-based data streams in a task-free data-incremental fashion and mitigates catastrophic forgetting using experience replay and continual prototype adaptation. Online learning is further facilitated using contrastive loss to enforce inter-class separation. LAPNet-HAR is evaluated on 5 publicly available activity datasets in terms of its ability to acquire new information while preserving previous knowledge. Our extensive empirical results demonstrate the effectiveness of LAPNet-HAR in task-free CL and uncover useful insights for future challenges.

There are potentially promising mitigation activities for epidemic and pandemic scenarios that are not currently the subject of significant research effort. Large epidemics and pandemics pose risks that are important to mitigate, even if the likelihood of the events is low and uncertain. While some efforts are the subject of extensive funding and consideration, other approaches are neglected. Here, we consider such neglected interventions which could significantly reduce the impact of such an epidemic or large-scale pandemic. These are identified via a narrative literature review of extant literature reviews and overviews of mitigations in epidemic and pandemic situations, followed by consideration of the economic value of information of further study of heretofore neglected interventions and approaches.Based on that analysis, we considered several classes of mitigations, and conducted more exploratory reviews of each. Those discussed include mitigations for (1) reducing transmission, such as personal protective equipment and encouraging improved hygiene, (2) reducing exposure by changing norms and targeted changes for high-risk or critical professions and activities, (3) reducing impact for those infected, and (4) increasing large scale resilience using disaster and infrastructure continuity planning.Some proposed mitigations are found to be of low marginal value. Other mitigations are likely to be valuable, but the concepts or applications are underdeveloped. In those cases, further research, resources, or preparation are valuable for mitigating both routine and extreme disease outbreak events. Still more areas of research are identified as having uncertain value based on specific but resolvable uncertainties. In both of the latter cases, there is no guarantee that mitigations identified as worthy of further consideration will be valuable, but the argument for further research is clear.

As an extension of the (univariate) Birnbaum-Saunders distribution, the Type II generalized crack (GCR2) distribution, built on an appropriate base density, provides a sufficient level of flexibility to fit various distributional shapes including heavy-tailed ones. In this paper, we develop a bivariate extension of the Type-II generalized crack distribution and study its dependency structure. For practical applications, several specific distributions, GCR2-Generalized Gaussian, GCR2-Student’s t and GCR2-Logistic, are constructed. The expectation-maximization algorithm is implemented to estimate the parameters in the bivariate GCR2 models. The model fitting results on the catastrophic loss dataset show that the bivariate GCR2 distribution base on the generalized Gaussian density fits the data significantly better than other alternative models.

In the event of an abrupt sunlight reduction scenario there is a time window that occurs between when food stores would likely run out for many countries (~6 months or less) and ~1 year when resilient foods are scaled up. A promising temporary resilient food is leaf protein concentrate (LPC). Although it is possible to extract LPC from tree biomass (e.g. leaves and needles), neither the yields nor the toxicity of the protein concentrates for humans from the most common tree species has been widely investigated. To help fill this knowledge gap, this study uses high-resolution mass spectrometry and an open source toolchain for non-targeted screening of toxins on five common North American coniferous species: Western Cedar, Douglas Fir, Ponderosa Pine, Western Hemlock, and Lodgepole Pine. The yields for LPC extraction from the conifers ranged from 1% to 7.5%. The toxicity screenings confirm that these trees may contain toxins that can be consumed in small amounts and additional studies including measuring the quantity of each toxin are needed. The results indicate that LPC is a promising candidate to be used as resilient food, but future work is needed before LPCs from conifers can be used as a wide-scale human food.

Continual learning, a crucial facet of machine learning, involves the perpetual acquisition of valuable insights from incoming data, sans the necessity for full dataset access. Esteemed as a fundamental goal in artificial intelligence, continual learning grapples with an enduring challenge—catastrophic forgetting. A proficient model must exhibit adaptability for new data assimilation and robustness to retain existing knowledge. Class-incremental learning (CIL) aids the gradual integration of knowledge from newly introduced classes, forming a universal classifier. However, directly training the model with fresh class instances triggers a problem—forgetting distinguishing features of prior classes, causing a performance decline. Addressing such issues in machine learning, this survey aims to delineate significant challenges, outcomes, and recent advancements, including our contributions to CIL techniques, especially in image classification.