Search Results (140)

Traffic accidents pose a significant threat to public safety, resulting in numerous fatalities, injuries, and a substantial economic burden each year. The development of predictive models capable of the real-time forecasting of post-accident impact using readily available data can play a crucial role in preventing adverse outcomes and enhancing overall safety. However, existing accident predictive models encounter two main challenges: first, a reliance on either costly or non-real-time data, and second, the absence of a comprehensive metric to measure post-accident impact accurately. To address these limitations, this study proposes a deep neural network model known as the cascade model. It leverages readily available real-world data from Los Angeles County to predict post-accident impacts. The model consists of two components: Long Short-Term Memory (LSTM) and a Convolutional Neural Network (CNN). The LSTM model captures temporal patterns, while the CNN extracts patterns from the sparse accident dataset. Furthermore, an external traffic congestion dataset is incorporated to derive a new feature called the “accident impact” factor, which quantifies the influence of an accident on surrounding traffic flow. Extensive experiments were conducted to demonstrate the effectiveness of the proposed hybrid machine learning method in predicting the post-accident impact compared to state-of-the-art baselines. The results reveal a higher precision in predicting minimal impacts (i.e., cases with no reported accidents) and a higher recall in predicting more significant impacts (i.e., cases with reported accidents). Full article

This paper is concerned with the choice of data granularity for the application of the Mack chain ladder model to forecast a loss reserve. It is a sequel to a related paper by Taylor, which considers the same question for the EDF chain ladder model. As in the earlier paper, it considers the question as to whether a decrease in the time unit leads to an increase or decrease in the variance of the loss reserve estimate. The question of whether a Mack chain ladder that is valid for one time unit (here called mesh size) remains so for another is investigated. The conditions under which the model does remain valid are established. There are various ways in which the mesh size of a data triangle may be varied, two of them of particular interest. The paper examines one of these, namely that in which development periods are preserved. Two versions of this are investigated: 1. the aggregation of development periods without change to accident periods; 2. the aggregation of accident periods without change to development periods. Taylor found that, in the case of the Poisson chain ladder, an increase in mesh size always increases the variance of the loss reserve estimate (subject to mild technical conditions). The case of the Mack chain ladder is more nuanced in that an increase in variance is not always guaranteed. Whether or not an increase or decrease occurs depends on the numerical values of certain of the age-to-age factors actually observed. The threshold values of the age-to-age factors at which an increase transitions to a decrease in variance are calculated. In the case of a change in the mesh of development periods, but with no change to accident periods, these values are computed for one particular data set, where it is found that variance always increases. It is conjectured that data sets in which this does not happen would be relatively rare. The situation is somewhat different when changes in mesh size over accident periods are considered. Here, the question of an increase or decrease in variance is more complex, and, in general terms, the occurrence of an increase in variance with increased mesh size is less likely. Full article

Poisoning and suffocation accidents occurred frequently in the pre-treatment pool workshops of biogas plants, so this paper provided a multi-dimensional risk analysis model: Bow-Tie-Qualitative Comparative Analysis (QCA)-Bayesian Neural Network-Consequence Simulation. First, the reasons for biogas poisoning and suffocation accidents were clarified through Bow-Tie. Then, the QCA method explored the accident cause combination paths in management. Next, the frequency distribution of biogas poisoning and suffocation accidents in the pre-treatment pool workshop was predicted to be 0.61–0.66 using the Bayesian neural network model, and the uncertainty of the forecast outcome was given. Finally, the ANSYS Fluent 16.0 simulation of biogas diffusion in three different ventilation types and a grid-independent solution of the simulation were conducted. The simulation results showed the distribution of methane, carbon dioxide and hydrogen sulfide gases and the hazards of the three gases to workers were analyzed. In addition, according to the results, this paper discussed the importance and necessity of ventilation in pre-treatment pool workshops and specified the hazard factors in biogas poisoning and suffocation accidents in the pre-treatment pool workshops. Some suggestions on gas alarms were also proposed. Full article

The Beluga Reefer accident underscores the hidden risks associated with complex wave–current interactions along South Africa’s coastline, particularly in the Agulhas Current retroflection zone. This study utilized ERA5 reanalysis and CMEMS surface current data to analyze the sea state conditions at the time of the accident. While the wind speeds were moderate (5.42 m/s) and windsea heights were relatively low (0.99 m), the significant wave height (Hs) peaked at 3.24 m, with a strong opposing NE Agulhas Current (1.27 m/s) inducing wave steepening and group compression, creating transient hazardous conditions despite a low overall wave steepness (0.0209). Just before the accident, the directional disparity (Δθ) between the swell and windsea systems collapsed sharply from 167.45° to 8.98°, providing a false sense of stability. The synergy of these conditions at the accident site triggered the event, demonstrating that visually aligned wave conditions can mask dangerous underlying interactions. These findings highlight the critical need for integrated wave–current diagnostics in maritime forecasting to better predict complex hazards and enhance vessel safety. Full article

The exploration of the statistical characteristics and distribution patterns of workplace accidents can help to reveal the intrinsic features and general laws of safety issues, which is essential for forecasting and decision making in safe production. Here, we conduct the detailed analysis of the distribution characteristics between the fatality number and the frequency of workplace accidents based on the in-depth data mining of various industries. The results show that the distribution between the fatality number and the frequency of workplace accidents follows a power-law distribution. Moreover, the exponents of such power-law distributions in different industries exhibit significant industry dependence, with the characteristic values of the power-law exponents in the coal mining industry, the hazardous chemicals industry, the transportation industry, and the construction industry being 1.55, 2.16, 2.15, and 2.92, respectively. Meanwhile, the temporal variation in the power-law distribution exponent in each industry can be used for the short-term prediction and evaluation of safe production, which will inform the decision making of the safety management department. Last, but not the least, the results of this study provide the theoretical basis for Heinrich’s Law and confirm that a substantial reduction in the number of small-scale accidents can effectively help control the frequency of large-scale fatal accidents. Full article

Predicting pit deformation to prevent safety accidents is the primary objective of pit deformation forecasting. A reliable predictive model enhances the ability to accurately monitor future deformation trends in pits. To enhance the prediction of pit deformation and improve accuracy and precision, an Improved Crown Porcupine Optimization Algorithm (ICPO) based on a Bidirectional Time Convolution Network–Bidirectional Gated Recirculation Unit (BITCN-BIGRU) is developed. This model is utilized to forecast the future deformation trends of the pit. Utilizing site data from a metro station pit project in Chengdu, the accuracy of the predicted values from Historical Average (HA), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) models is evaluated against the six models developed in this study, including the ICPO-BITCN-BIGRU model. Comparison of the test results indicates that the ICPO-BITCN-BIGRU prediction model exhibits superior predictive performance. The predicted values from the ICPO-BITCN-BIGRU model demonstrate R² values of 0.9768, 0.9238, and 0.9943, respectively, indicating strong concordance with the actual values. Consequently, the ICPO-BITCN-BIGRU prediction model developed in this study exhibits high prediction accuracy and robust stability, making it suitable for practical engineering applications. Full article

Vehicle owners often use certified service centers throughout the warranty period, which usually extends for five years after buying. Nonetheless, after this timeframe concludes, a large number of owners turn to unapproved service providers, mainly motivated by financial factors. This change signifies a significant drop in income for automakers and their certified service networks. To tackle this issue, manufacturers utilize customer relationship management (CRM) strategies to enhance customer loyalty, usually depending on segmentation methods to pinpoint potential clients. However, conventional approaches frequently do not successfully forecast which clients are most likely to need or utilize maintenance services. This research introduces a machine learning-driven framework aimed at forecasting the probability of monthly maintenance attendance for customers by utilizing an extensive historical dataset that includes information about both customers and vehicles. Additionally, this predictive approach supports workforce planning and scheduling within after-sales service centers, aligning with AI-driven labor optimization frameworks such as those explored in the AI4LABOUR project. Four algorithms in machine learning—Decision Tree, Random Forest, LightGBM (LGBM), and Extreme Gradient Boosting (XGBoost)—were assessed for their forecasting capabilities. Of these, XGBoost showed greater accuracy and reliability in recognizing high-probability customers. In this study, we propose a machine learning framework to predict vehicle maintenance visits for after-sales services, leading to significant operational improvements. Furthermore, the integration of AI-driven workforce allocation strategies, as studied within the AI4LABOUR (reshaping labor force participation with artificial intelligence) project, has contributed to more efficient service personnel deployment, reducing idle time and improving customer experience. By implementing this approach, we achieved a 20% reduction in information delivery times during service operations. Additionally, survey completion times were reduced from 5 min to 4 min per survey, resulting in total time savings of approximately 5906 h by May 2024. The enhanced service appointment scheduling, combined with timely vehicle maintenance, also contributed to reducing potential accident risks. Moreover, the transition from a rule-based maintenance prediction system to a machine learning approach improved efficiency and accuracy. As a result of this transition, individual customer service visit rates increased by 30%, while corporate customer visits rose by 37%. This study contributes to ongoing research on AI-driven workforce planning and service optimization, particularly within the scope of the AI4LABOUR project. Full article

An event is an occurrence that takes place at a specific time and location that can be either weather-related (snowfall), social (crime), natural (earthquake), political (political unrest), or medical (pandemic) in nature. These events do not belong to the “normal” or “usual” spectrum and result in a change in a given situation; thus, their prediction would be very beneficial, both in terms of timely response to them and for their prevention, for example, the prevention of traffic accidents. However, this is currently challenging for researchers, who are called upon to manage and analyze a huge volume of data in order to design applications for predicting events using artificial intelligence and high computing power. Although significant progress has been made in this area, the heterogeneity in the input data that a forecasting application needs to process—in terms of their nature (spatial, temporal, and semantic)—and the corresponding complex dependencies between them constitute the greatest challenge for researchers. For this reason, the initial forecasting applications process data for specific situations, in terms of number and characteristics, while, at the same time, having the possibility to respond to different situations, e.g., an application that predicts a pandemic can also predict a central phenomenon, simply by using different data types. In this work, we present the forecasting applications that have been designed to date. We also present a model for predicting traffic accidents using categorical logic, creating a Knowledge Base using the Resolution algorithm as a proof of concept. We study and analyze all possible scenarios that arise under different conditions. Finally, we implement the traffic accident prediction model using the Prolog language with the corresponding Queries in JPL. Full article

The collapse of Baltimore’s Francis Scott Key Bridge in March 2024 has stressed the need for urban traffic network optimization within smart city initiatives. This paper utilizes the ARIMA model to forecast what traffic would have been like if the bridge had not collapsed, giving us a benchmark to assess the impact. It then identifies the roads most affected by comparing these forecasts with the actual post-collapse traffic data. To address the increased demand for efficient public transport, we propose an intelligent bus network model. This model uses principal component analysis and grid segmentation to inform decisions on increasing bus stations and adjusting bus frequencies on key routes. It aims to satisfy stakeholders by enhancing service coverage and reliability. The research also presents a comprehensive traffic model that leverages principal component analysis, genetic algorithms, and KD-tree to evaluate overall and directional traffic flow, providing strategic insights into congestion mitigation. Furthermore, it examines traffic safety issues, including accident-prone areas and traffic signal intersections, to offer recommendations. Finally, the study evaluates the effectiveness, stability, and benefits of the proposed intelligent traffic network model, aiming to improve the city’s traffic infrastructure and safety. Full article

Water transportation is a critical component of the overall transportation system. However, the gradual increase in traffic density has led to a corresponding rise in accident occurrences. This study proposes a quantitative framework for analyzing the evolutionary paths of maritime traffic accident risks by integrating complex network theory and link prediction methods. First, 371 maritime accident investigation reports were analyzed to identify the underlying risk factors associated with such incidents. A risk evolution network model was then constructed, within which the importance of each risk factor node was evaluated. Subsequently, several node similarity indices based on node importance were proposed. The performance of these indices was compared, and the optimal indicator was selected. This indicator was then integrated into the risk evolution network model to assess the interdependence between risk factors and accident types, ultimately identifying the most probable evolution paths from various risk factors to specific accident outcomes. The results show that the risk evolution path shows obvious characteristics: “lookout negligence” is highly correlated with collision accidents; “improper route selection” plays a critical role in the risk evolution of grounding and stranding incidents; “improper on-duty” is closely linked to sinking accidents; and “illegal operation” show a strong association with fire and explosion events. Additionally, the average risk evolution paths for collisions, groundings, and sinking accidents are relatively short, suggesting higher frequencies of occurrence for these accident types. This research provides crucial insights for managing water transportation systems and offers practical guidance for accident prevention and mitigation. Full article

In this study, we developed an interpretable machine learning (ML) framework to predict marine pollution and economic losses from accident risk factors. A triple-feature selection process identified key predictors, followed by a comparative analysis of eight ML models. Random forest outperformed other models in forecasting environmental and property damage. The interpretable model was established based on the SHAP value framework, which revealed that onboard personnel count, vessel dimensions (length), and accident/ship types account for the risk factors with the most severe consequences, with environmental conditions like wind speed and air temperature contributing secondary effects. The methodology enables two critical applications: (1) environmental agencies can proactively assess accident impact through the identified risk triggers, optimizing emergency response planning, and (2) insurance providers gain data-driven risk evaluation metrics to refine premium calculations. By quantifying how human/technical factors, including crew members and vessel specifications, dominate over natural variables in accident effects, this data-driven approach provides actionable insights for maritime safety management and financial risk mitigation, achieving high prediction accuracy while maintaining model transparency through Shapley value explanations. Full article

This paper is concerned with the choice of data granularity for application of the EDF (Exponential Dispersion Family) chain ladder model to forecast a loss reserve. As the duration of individual accident and development periods is decreased, the number of data points increases, but the volatility of each point increases. This leads to a question as to whether a decrease in time unit leads to an increase or decrease in the variance of the loss reserve estimate. Is there an optimal granularity with respect to the variance of the loss reserve? A preliminary question is that of whether an EDF chain ladder that is valid for one duration (here called mesh size) remains so for another. The conditions under which this is so are established. There are various ways in which the mesh size of a data triangle may be varied. The paper identifies two of particular interest. For each of these two types of variation, the effect on variance of loss reserve is studied. Subject to some technical qualifications, the conclusion is that an increase in mesh size always increases the variance. It follows that one should choose a very high degree of granularity in order to maximize efficiency of loss reserve forecasting. Full article

Trend forecasting and early anomaly warnings are important for avoiding aircraft engine failures or accidents. This study proposes a trend forecasting method based on enhanced Slice-level Adaptive Normalization (SAN) using a Long Short-Term Memory (LSTM) neural network under multi-operating conditions. Firstly, a condition recognition technology is constructed to automatically identify the operating conditions based on the predetermined judgment conditions, and vibration signal features are adaptively divided into three typical operating conditions, namely, the idling operating condition, the starting operating condition and the utmost operating condition. The features of original signals are extracted to reduce the impacts of signal fluctuations and noise preliminarily. Secondly, enhanced SAN is used to normalize and denormalize the features to alleviate non-stationary factors. To improve prediction accuracy, an L₁ filter is adopted to extract the trend term of the features, which can effectively reduce the overfitting of SAN to local information. Moreover, the slice length is quantitatively estimated by the fixed points in L₁ filtering, and a tail amendment technology is added to expand the applicable range of enhanced SAN. Finally, an LSTM-based forecasting model is constructed to forecast the normalized data from enhanced SAN, serving as input during denormalization. The final results under different operating conditions are the output from denormalization. The validity of the proposed method is verified using the test data of an aircraft engine. The results show that the proposed method can achieve higher forecasting accuracy compared to other methods. Full article

Road safety is a critical concern, as accidents happen globally. Despite efforts to enhance roads and enforce stricter driving rules, the number of accidents remains high. These issues arise from distracted driving, speeding, and driving under the influence. In the United States, fatal accidents increased by 16% from 2018 to 2022. The number of deaths rose from 36,835 in 2018 to 42,795 in 2022. This trend reveals a critical need for new solutions to reduce traffic incidents and improve road safety. Machine learning (ML) can help make roads safer and reduce traffic-related deaths. This paper presents an ML-based real-time highway traffic incident forecasting system named “VehiCast”. VehiCast utilizes vehicular ad hoc networks (VANETs) and federated learning (FL) to collect real-time traffic data, such as average delay, average speed, and the total number of vehicles across several highway zones, to enhance traffic incident prediction accuracy in real-time. Our extensive experimental results showcase that VehiCast reaches an impressive prediction accuracy of 91%, highlighting the power of innovation and determination. Full article

We investigate maritime accidents near Bornholm Island in the Baltic Sea, focusing on one of the most recent vessel collisions and a way to improve maritime safety as a prevention strategy. By leveraging Long Short-Term Memory autoencoders, a class of deep recurrent neural networks, this research demonstrates a unique approach to forecasting vessel trajectories and assessing collision risks. The proposed method integrates trajectory predictions with statistical techniques to construct probabilistic boundaries, including confidence intervals, prediction intervals, ellipsoidal prediction regions, and conformal prediction regions. The study introduces a collision risk score, which evaluates the likelihood of boundary overlaps as a metric for collision detection. These methods are applied to simulated test scenarios and a real-world case study involving the 2021 collision between the Scot Carrier and Karin Hoej cargo ships. The results demonstrate that CPR, a non-parametric approach, reliably forecasts collision risks with 95% confidence. The findings underscore the importance of integrating statistical uncertainty quantification with deep learning models to improve navigational decision-making and encourage a shift towards more proactive, AI/ML-enhanced maritime risk management protocols. Full article