Catastrophic Loss Research Articles

How will better data (and better use of data) enable us to save the planet?

The potentially catastrophic loss of biodiversity happening around the globe is largely caused by economic activity that is not connected to its environmental impacts. To improve positive and reduce negative outcomes for nature, investment decision-makers in companies, governments, and the finance sector need data on the impacts of economic activity, especially production of food and other commodities, on biodiversity, at fine geographical scales. This Essay argues that the data allowing us to identify the most important factors causing biodiversity loss are already available. However, we need more data to track impacts on biodiversity up value chains into pathways, toolkits, and approaches that will facilitate verified, concrete actions by companies and consumers to reduce threats to biodiversity in particular places. Our current knowledge is insufficient to deliver complete responses to the biodiversity crisis, but this is no excuse for delaying action.

Wood Chipper Design for Biofuel Production in a Global Catastrophic Loss of Infrastructure Scenario

A variety of events such as high-altitude electromagnetic pulses, extreme solar storms, and coordinated cyber attacks could result in a catastrophic loss of infrastructure on a continental or global scale. The lengthy repair of critical infrastructure creates a need for alternative fuels such as wood gas. Wood gas is produced by heating wood in a low-oxygen environment and can be used to power combustion engines. This work investigates a novel wood chipper, designed as an energy-efficient tool for producing wood gas stock, wood chips, aiming to speed up the transition to alternative fuel. A prototype is built and tested to determine the energy efficiency and production rate of the device. The results suggest that the wood chipper could produce one cord of wood chips, 3.6 m3, in less than a day and is a viable alternative to other manual wood-processing methods. In addition, the global scaling up of production of the wood chipper is considered, indicating that the mass production of the wood chipper could accelerate the transition of wood gas production methods from manual to machine-driven immediately after a catastrophic event.

A portable transistor immunosensor for fast identification of porcine epidemic diarrhea virus

Widespread distribution of porcine epidemic diarrhea virus (PEDV) has led to catastrophic losses to the global pig farming industry. As a result, there is an urgent need for rapid, sensitive and accurate tests for PEDV to enable timely and effective interventions. In the present study, we develop and validate a floating gate carbon nanotubes field-effect transistor (FG CNT-FET)-based portable immunosensor for rapid identification of PEDV in a sensitive and accurate manner. To improve the affinity, a unique PEDV spike protein-specific monoclonal antibody is prepared by purification, and subsequently modified on FG CNT-FET sensor to recognize PEDV. The developed FET biosensor enables highly sensitive detection (LoD: 8.1 fg/mL and 100.14 TCID50/mL for recombinant spike proteins and PEDV, respectively), as well as satisfactory specificity. Notably, an integrated portable platform consisting of a pluggable FG CNT-FET chip and a portable device can discriminate PEDV positive from negative samples and even identify PEDV and porcine deltacoronavirus within 1 min with 100% accuracy. The portable sensing platform offers the capability to quickly, sensitively and accurately identify PEDV, which further points to a possibility of point of care (POC) applications of large-scale surveillance in pig breeding facilities.

Assessing selection potential for Pacific oyster (Crassostrea gigas) to a North American OsHV-1 μvar: Comparing two experimental assay methods

Ostreid Herpesvirus - 1 is a global threat to Pacific oyster aquaculture as catastrophic farm losses have occurred where OsHV-1 microvariants (μvars) become endemic. Genetic selection has proven effective for increasing survival to OsHV-1 exposure in multiple Pacific oyster breeding programs, but a novel μvar detected in San Diego Bay California has the potential to severely impact commercial aquaculture along the North American Pacific coast as no OsHV-1 μvar selected population is available commercially. Using a Molluscan Broodstock Program population descending from imports originating in the Miyagi prefecture (Japan), we evaluated the potential for genetic selection to improve survival against the San Diego Bay OsHV-1 μvar using a laboratory trial. Two experimental methods were evaluated, a common garden method where all 15 full-sibling families were pooled, and a petri dish method, where each family was confounded within replicate. No striking differences were found between experimental methods, and both experiments displayed high heritability (0.69 and 0.64) of the tested trait, a broad range of estimated breeding values, and a moderate prediction accuracy for the survival trait. A significant genotype-by-environment interaction occurred, highlighting variability in laboratory-based disease challenges, but the two experiments yielded similar results with minor to moderate re-ranking of mean family survival. Overall, the Miyagi population of Pacific oysters curated by the Molluscan Broodstock Program is a promising population to use for artificial selection to increase survival against San Diego Bay OsHV-1 μvar in attempts to obviate farm losses if the virus expands its currently isolated range.

An aquaculture risk model to understand the causes and consequences of Atlantic Salmon mass mortality events: A review

AbstractMass mortality events (MMEs) are defined as the death of large numbers of fish over a short period of time. These events can result in catastrophic losses to the Atlantic salmon aquaculture industry and the local economy. However, they are challenging to understand because of their relative infrequency and the high number of potential factors involved. As a result, the causes and consequences of MMEs in Atlantic salmon aquaculture are not well understood. In this study, we developed a structural network of causal risk factors for MMEs for aquaculture and the communities that depend on Atlantic salmon aquaculture. Using the Interpretive Structural Modeling (ISM) technique, we analysed the causes of Atlantic salmon mass mortalities due to environmental (abiotic), biological (biotic) and nutritional risk factors. The consequences of MMEs were also assessed for the occupational health and safety of aquaculture workers and their implications for the livelihoods of local communities. This structural network deepens our understanding of MMEs and points to management actions and interventions that can help mitigate mass mortalities. MMEs are typically not the result of a single risk factor but are caused by the systematic interaction of risk factors related to the environment, fish diseases, feeding/nutrition and cage‐site management. Results also indicate that considerations of health and safety risk, through pre‐ and post‐event risk assessments, may help to minimize workplace injuries and eliminate potential risks of human fatalities. Company and government‐assisted socio‐economic measures could help mitigate post‐mass mortality impacts. Appropriate and timely management actions may help reduce MMEs at Atlantic salmon cage sites and minimize the physical and social vulnerabilities of workers and local communities.

Modelling additive extremile regression by iteratively penalized least asymmetric weighted squares and gradient descent boosting

Quantile regression has emerged as one of the standard tools for regression analysis that enables a proper assessment of the complete conditional distribution of responses. This article considers a valuable alternative class to quantiles, called extremiles. Extremiles bear much better than quantiles the burden of representing an alert risk measure to the magnitude of infrequent catastrophic losses. Additive regression model has concentrated on making the regression structure more flexible by including nonlinear effects of continuous covariates and interaction effects. As a consequence, additive extremile regression based on minimizing an asymmetrically weighted sum of squared residuals is introduced. Different estimation procedures are presented including iteratively penalized least asymmetric weighted squares and gradient descent boosting. The properties of these procedures are investigated in a simulation study and an analysis of tropical cyclone intensity of the North Atlantic which reveals the function of variable selection by modelling additive extremile regression simultaneously.

Nerambum: A Virtual Exhibition Platform

In the wake of the global pandemic, industries worldwide faced a disruptive challenge, particularly those thriving on physical exhibitions essential for expanding market share, unveiling new products, and forging valuable connections. The health crisis and consequent strict guidelines rendered traditional exhibitions untenable, causing a catastrophic loss of audience and opportunities for businesses. This unprecedented situation accelerated the adoption of virtual alternatives, but existing platforms failed to capture the essence and dynamism of live, in-person interactions, leading to a palpable gap in user experience. "Nerambum" emerges as a transformative solution in this landscape, offering a sophisticated online exhibition platform that recreates the authentic feel of physical events. It seamlessly integrates a 3D environment for attendees with a multifunctional web application, supporting various user roles including admin, exhibition owners, stall owners, and general participants. Beyond its immersive virtual space, "Nerambum" stands out with its comprehensive service management—from secure authentication, customizable avatars, and stalls to real-time video streaming and a practical payment gateway mimicking physical ticket purchases. The platform not only adapts to the current necessity for remote interaction but also anticipates the enduring relevance of virtual engagements in the exhibition industry's future.

What Drives Property Insurance Demand in Croatia?

Abstract Property insurance is an important part of risk management, insurance coverage and vehicle of catastrophic losses financing. In developed countries, property insurance plays an important role in the structure of non-life premium, while in developing markets, such as Croatian, property insurance penetration is rather low. Unfortunately, the recent earthquakes in Zagreb (Croatia) and the surrounding area in 2020 have once again highlighted a large protection gap, resulting in a dominant role of government in financing catastrophic losses. The aim of this paper is to discuss and investigate the factors that determinate property insurance demand in Croatia. Using the VAR (Vector AutoRegression) methodology extended via spillover index approach, we analyse a comprehensive data set ranging from the economic, financial, social to insurance indicators. The conclusions of this research point that some of the expected drivers of insurance demand such as mortgage loans, consumer confidence and loss experience, are not important in Croatia, and that property insurance is dominantly driven by economic development.

Spatiotemporal Prediction of Rainfall-induced Landslides Using Machine Learning Techniques

Landslides cause significant damage to infrastructure and loss of life. Landslide susceptibility map (LSM), as an important reference for landslide hazard assessment, indicates prone areas of landslides. However, conventional LSM only predicts the spatial distribution of potential landslides, which cannot fully explain the occurrence of landslides at different times. In the present study, a spatiotemporal LSM is conducted to predict landslides both in space and time. A landslide database containing 223 recorded landslide events in southwestern Pennsylvania is used. Fourteen topographic spatial factors and eight rainfall temporal factors are used for machine learning (ML). Four ML models are applied in the study, including Logistic regression (LR), Support vector machine (SVM), Random forest (RF), and Gradient boosting machine (GBM). The results show that through cross-validation, RF outperforms the other algorithms with a value 0.86 of the area under the receiver operating characteristic curve (AUC score). The optimal model is used to generate spatiotemporal LSMs. It is concluded that by introducing spatial and temporal information simultaneously, ML models have the capability of learning the pattern of landslide occurrence both in space and time, providing an effective assessment tool to reduce catastrophic loss of landslide hazards.

Data recovery methods for DNA storage based on fountain codes

Today's digital data storage systems typically offer advanced data recovery solutions to address the problem of catastrophic data loss, such as software-based disk sector analysis or physical-level data retrieval methods for conventional hard disk drives. However, DNA-based data storage currently relies solely on the inherent error correction properties of the methods used to encode digital data into strands of DNA. Any error that cannot be corrected utilizing the redundancy added by DNA encoding methods results in permanent data loss. To provide data recovery for DNA storage systems, we present a method to automatically reconstruct corrupted or missing data stored in DNA using fountain codes. Our method exploits the relationships between packets encoded with fountain codes to identify and rectify corrupted or lost data. Furthermore, we present file type-specific and content-based data recovery methods for three file types, illustrating how a fusion of fountain encoding-specific redundancy and knowledge about the data can effectively recover information in a corrupted DNA storage system, both in an automatic and in a guided manual manner. To demonstrate our approach, we introduce DR4DNA, a software toolkit that contains all methods presented. We evaluate DR4DNA using both in-silico and in-vitro experiments.

A complex network perspective on spatiotemporal propagations of extreme precipitation events in China

Extreme precipitation events (EPEs) cause catastrophic losses to human lives and social economy. Focusing on the wet season from May to October, this paper formulates the complex network to characterize the spatiotemporal propagations of EPEs over 218 river basins in China. Specifically, based on the daily Multi-Source Weighted-Ensemble Precipitation version 2, the event synchronization along with significance test facilitates the complex network of EPEs from 1979 to 2020. By characterizing the topology of EPEs, the results show that synchronous EPEs generally occur over the Hetao Area of Yellow River Basin, Haihe River Basin and Northeast China. The complex network of EPEs classifies the 218 river basins into six communities. In May and June, more than 50 % of EPEs generally occur in South China. From July to September, over 60 % of EPEs tend to be observed in the Huang-Huai-Hai Plain, upper-middle reaches of Yellow River Basin, Northeast China and Southwest China. In October, about 65 % of EPEs happen in South and Southwest China. In the meantime, less than 10 % of EPEs occur in Northwest China. Among the synchronous EPEs starting in late June or early July and peaking in late July to early August, there exist propagations along the Yellow River, Haihe, Liaohe and Songhua River Basin within one to eight days. Eastward moisture transport along this pathway within four to six days drives the propagation of synchronous EPEs. The complex network analysis of this paper provides useful information for early warning and forecasting of EPEs and flood hazards.

Catastrophic Talar Bone Loss from High Velocity Trauma Treated with Structural Tricorticate Fibula Autograft, Compression Frame and Midfoot Distraction Arthroplasty

Osteonecrosis is caused by the interruption of subchondral blood supply and can affect various bones in the human body. This case study details the treatment of a 37-year-old male with comminuted foot fractures, post-traumatic osteonecrosis, and previous surgeries who regained full weight-bearing ability and returned to work duties through a variety of techniques. Following the diagnosis of the disease through MRI scans, a surgical plan was devised for the patient: Pan-Talar Fusion Anterior Plate Reflect Anterior Flap Fibula Structural Autograft, lateral approach Complex frame compress ankle and distract mid-foot Distraction arthroplasty of midfoot The patient successfully achieved full weight-bearing and returned to work 11 months post-surgery. X-rays confirmed progressive bone healing and no hardware complications. This case demonstrates a successful treatment approach for complex hindfoot fractures, enabling full weight-bearing and return to work. Further research is needed to explore long-term outcomes and potential improvements in the surgical technique.

Helium in the Extended Atmosphere of the Warm Superpuff TOI-1420b

Superpuffs are planets with exceptionally low densities (ρ ≲ 0.1 g cm−3) and core masses (M c ≲ 5M ⊕). Many lower-mass (M p ≲ 10M ⊕) superpuffs are expected to be unstable to catastrophic mass loss via photoevaporation and/or boil-off, whereas the larger gravitational potentials of higher-mass (M p ≳ 10M ⊕) superpuffs should make them more stable to these processes. We test this expectation by studying atmospheric loss in the warm, higher-mass superpuff TOI-1420b (M = 25.1M ⊕, R = 11.9R ⊕, ρ = 0.08 g cm−3, T eq = 960 K). We observed one full transit and one partial transit of this planet using the metastable helium filter on Palomar/WIRC and found that the helium transits were 0.671% ± 0.079% (8.5σ) deeper than the TESS transits, indicating an outflowing atmosphere. We modeled the excess helium absorption using a self-consistent 1D hydrodynamics code to constrain the thermal structure of the outflow given different assumptions for the stellar XUV spectrum. These calculations then informed a 3D simulation, which provided a good match to the observations with a modest planetary mass-loss rate of 1010.82 g s−1 ( Mp/Ṁ≈70 Gyr). Superpuffs with M p ≳ 10M ⊕, like TOI-1420b and WASP-107b, appear perfectly capable of retaining atmospheres over long timescales; therefore, these planets may have formed with the unusually large envelope mass fractions they appear to possess today. Alternatively, tidal circularization could have plausibly heated and inflated these planets, which would bring their envelope mass fractions into better agreement with expectations from core-nucleated accretion.

A Reliable Framework for Detection of Smart Contract Vulnerabilities for Enhancing Operability in Inter-Organizational Systems

Information and communication technology based inter-organizational systems enable companies to integrate information and conduct business electronically across different parts of the organization. For organizations embracing blockchain, smart contracts provide automation and operational efficiency for inter-organizational systems. Initially utilised for financial transactions, smart contract are extended beyond banking and deployed in wide number of organizations. Smart contracts are regarded as self-executing type of contract consisting of agreement’s terms embedded directly into the code which plays a vital role in operability for inter-organizational systems, however, smart contract vulnerabilities can arise due to programming errors, leading to security issues. The effects of smart contract vulnerabilities can be significant, including loss of funds, unauthorized access to sensitive information, manipulation of data, and loss of trust in the application leading to catastrophic financial losses followed by legal implications for an organization based on blockchain technology. The goal of smart contracts exploiting vulnerabilities is to discover and eliminate potential security vulnerabilities in smart contract code prior to it being deployed. Detecting vulnerabilities in a timely manner helps to prevent financial losses, unauthorized access, and data manipulation. In order to provide a robust solution to detect vulnerabilities in smart contracts, the proposed methodology presents a novel approach for rapid detection of vulnerabilities by integrating genetic algorithm with isolation forest. Furthermore, enhancing smart contract vulnerability identification with higher accuracy and false-positive rate provides a reliable gateway for organizations to adopt blockchain.

A comprehensive survey of research towards AI-enabled unmanned aerial systems in pre-, active-, and post-wildfire management

Wildfires have emerged as one of the most destructive natural disasters worldwide, causing catastrophic losses. These losses have underscored the urgent need to improve public knowledge and advance existing techniques in wildfire management. Recently, the use of Artificial Intelligence (AI) in wildfires, propelled by the integration of Unmanned Aerial Vehicles (UAVs) and deep learning models, has created an unprecedented momentum to implement and develop more effective wildfire management. Although existing survey papers have explored learning-based approaches in wildfire, drone use in disaster management, and wildfire risk assessment, a comprehensive review emphasizing the application of AI-enabled UAV systems and investigating the role of learning-based methods throughout the overall workflow of multi-stage wildfire management, including pre-fire (e.g., vision-based vegetation fuel measurement), active-fire (e.g., fire growth modeling), and post-fire tasks (e.g., evacuation planning) is notably lacking. This survey synthesizes and integrates state-of-the-science reviews and research at the nexus of wildfire observations and modeling, AI, and UAVs — topics at the forefront of advances in wildfire management, elucidating the role of AI in performing monitoring and actuation tasks from pre-fire, through the active-fire stage, to post-fire management. To this aim, we provide an extensive analysis of the existing remote sensing systems with a particular focus on the UAV advancements, device specifications, and sensor technologies relevant to wildfire management. We also examine the pre-fire and post-fire management approaches, including fuel monitoring, prevention strategies, as well as evacuation planning, damage assessment, and operation strategies. Additionally, we review and summarize a wide range of computer vision techniques in active-fire management, with an emphasis on Machine Learning (ML), Reinforcement Learning (RL), and Deep Learning (DL) algorithms for wildfire classification, segmentation, detection, and monitoring tasks. Ultimately, we underscore the substantial advancement in wildfire modeling through the integration of cutting-edge AI techniques and UAV-based data, providing novel insights and enhanced predictive capabilities to understand dynamic wildfire behavior.

A neoteric antibacterial ceria-silver nanozyme for abiotic surfaces

Community-associated and hospital-acquired infections caused by bacteria continue to yield major global challenges to human health. Bacterial contamination on abiotic surfaces is largely spread via high-touch surfaces and contemporary standard disinfection practices show limited efficacy, resulting in unsatisfactory therapeutic outcomes. New strategies that offer non-specific and broad protection are urgently needed. Herein, we report our novel ceria-silver nanozyme engineered at a molar ratio of 5:1 and with a higher trivalent (Ce3+) surface fraction. Our results reveal potent levels of surface catalytic activity on both wet and dry surfaces, with rapid, and complete eradication of Pseudomonas aeruginosa, Staphylococcus aureus, and methicillin resistant S. aureus, in both planktonic and biofilm form. Preferential electrostatic adherence of anionic bacteria to the cationic nanozyme surface leads to a catastrophic loss in both aerobic and anaerobic respiration, DNA damage, osmodysregulation, and finally, programmed bacterial lysis. Our data reveal several unique mechanistic avenues of synergistic ceria-Ag efficacy. Ag potentially increases the presence of Ce3+ sites at the ceria-Ag interface, thereby facilitating the formation of harmful H2O2, followed by likely permeation across the cell wall. Further, a weakened Ag-induced Ce–O bond may drive electron transfer from the Ec band to O2, thereby further facilitating the selective reduction of O2 toward H2O2 formation. Ag destabilizes the surface adsorption of molecular H2O2, potentially leading to higher concentrations of free H2O2 adjacent to bacteria. To this end, our results show that H2O2 and/or NO/NO2−/NO3− are the key liberators of antibacterial activity, with a limited immediate role being offered by nanozyme-induced ROS including O2•- and OH•, and likely other light-activated radicals. A mini-pilot proof-of-concept study performed in a pediatric dental clinic setting confirms residual, and continual nanozyme antibacterial efficacy over a 28-day period. These findings open a new approach to alleviate infections caused by bacteria for use on high-touch hard surfaces.

Failure-mode scale transitions in RC and PC beams

Current design Standards for reinforced concrete beams prescribe to respect a minimum, ρmin, and a maximum, ρmax, reinforcement ratio in the design of structures. Below ρmin a brittle failure due to unstable crack propagation is expected. On the other hand, for ρ > ρmax a brittle failure due to concrete crushing is obtained. In this framework, a reinforced concrete element with ρmin < ρ < ρmax presents yielded steel at Ultimate Limite State (ULS) with a stable behaviour and no catastrophic loss of bearing capacity. Design Standards define ρmin and ρmax limits on the basis of the Bernoulli’s hypothesis of plane sections, and completely disregard size-scale effects. Within the present paper, Dimensional Analysis is used to determine the Brittleness Numbers that govern the behaviour of reinforced concrete (RC) as well as of prestressed reinforced concrete (PC) beams. Therefore, parametric analyses carried out by means of the Cohesive/Overlapping Crack Model (COCM) are used to study the ductile-to-brittle transitions in RC and PC beams, and to highlight the size-scale dependency of the two above-mentioned reinforcement limits.

Natural Infection with H5N1 Highly Pathogenic Influenza (HPAI) Virus in 5- and 10-Day-Old Commercial Pekin Ducklings (Anas platyrhynchos domesticus).

Highly pathogenic avian influenza (HPAI) has resulted in catastrophic economic losses globally in poultry. This case report describes the diagnostic detection and pathology of HPAI H5N1 in 5-day-old commercial ducklings, which is an atypical age for detection of natural infection of HPAI in poultry. The pathology observed at 5 days of age was also compared to lesions observed in ducklings from the same flock evaluated at 10 days of age before depopulation. The California Animal Health and Food Safety (CAHFS) Laboratory, Tulare, received ten 5-day-old Pekin duckling (Anas platyrhynchos domesticus) carcasses for diagnostic evaluation due to mortality that started increasing at 3 days of age. The most common gross findings included bilateral pulmonary edema with congestion and enlarged, mottled livers and spleens. Microscopically, cerebral neuronophagia, pancreatic necrosis, and interstitial pneumonia with pulmonary edema were observed in the 5-day-old ducklings. Oropharyngeal and cloacal swabs were positive for avian influenza virus (AIV) by real-time reverse transcriptase PCR. The AIV was typed as HPAI, EA/AM 2.3.4.4b H5N1 goose/Guangdong clade lineage by the National Veterinary Services Laboratory. Ducks at the affected premises were depopulated 4 days after the 5-day-old ducklings were submitted to the CAHFS lab, at which time additional tissue samples were collected for comparison to 10-day-old ducklings on the same premises. Differences in microscopic lesions and AIV tissue distribution were observed between the 5-day and 10-day tissues collected. Notably, microscopic lesions were more severe in the brain and pancreas at 10 days of age. Findings in 10-day-old ducklings included cerebral lymphoplasmacytic perivascular cuffing, gliosis, neuronal degeneration, and pancreatic necrosis. AIV antigen distribution and intensity was greatest in the cerebral tissue of the brains at 10 days and in the lungs at 5 days of age. To the authors' knowledge, published studies are limited on AIV natural infection in domestic ducks less than 9 days of age.

Application of back propagation neural network in complex diagnostics and forecasting loss of life of cellulose paper insulation in oil-immersed transformers

Oil-immersed transformers are expensive equipment in the electrical system, and their failure would lead to widespread blackouts and catastrophic economic losses. In this work, an elaborate diagnostic approach is proposed to evaluate twenty-six different transformers in-service to determine their operative status as per the IEC 60599:2022 standard and CIGRE brochure. The approach integrates dissolved gas analysis (DGA), transformer oil integrity analysis, visual inspections, and two Back Propagation Neural Network (BPNN) algorithms to predict the loss of life (LOL) of the transformers through condition monitoring of the cellulose paper. The first BPNN algorithm proposed is based on forecasting the degree of polymerization (DP) using 2-Furaldehyde (2FAL) concentration measured from oil samples using DGA, and the second BPNN algorithm proposed is based on forecasting transformer LOL using the 2FAL and DP data obtained from the first BPNN algorithm. The first algorithm produced a correlation coefficient of 0.970 when the DP was predicted using the 2FAL measured in oil and the second algorithm produced a correlation coefficient of 0.999 when the LOL was predicted using the 2FAL and DP output data obtained from the first algorithm. The results show that the BPNN can be utilized to forecast the DP and LOL of transformers in-service. Lastly, the results are used for hazard analysis and lifespan prediction based on the health index (HI) for each transformer to predict the expected years of service.

Intelligent fault diagnosis of ultrasonic motors based on graph-regularized CNN-BiLSTM

The ultrasonic motor (USM) is peculiarly prone to failure due to continuous high-frequency friction-related power transfer, whose failure mechanisms are remarkably different from traditional induction motors. Intelligent fault diagnosis provides a way to alarm and avoid catastrophic losses proactively. However, previous studies using deep learning usually ignore the inherent geometric structure of the signal distribution. This paper proposes an intelligent multi-signal fault diagnosis framework for USMs to restore the linear or nonlinear manifold structure by preserving the internal structure by integrating graph regularization with deep neural networks. Firstly, the one-dimensional CNN to learn spatial correlations and BiLSTM to exploit temporal dependencies are coalesced to build the deep neural network. Then, an improved k-nearest neighbor graph is proposed to protect the geometric structure information and force the latent features to be more concentrated within their classes. Moreover, the layer in the deep architecture to integrate graph regularization is designed to reduce computation cost, and an adaptive decay strategy is considered to adjust the coefficient of graph regularized automatically. A two-stage training algorithm is developed by considering the time to calculate the graph regularization term. Finally, the proposed multi-signal fault diagnosis framework is validated using datasets from the fault injection experiment of similar USMs in China’s Yutu rover of Chang’e lunar probe. Experimental results show that the proposed method can effectively discriminate different fault types.