Recent advancements in intelligent transportation systems and data analytics within transportation systems present a significant opportunity to enhance operational efficiency. In this context, the pivotal role of intelligent agents in achieving real-time optimisation for traffic management is highlighted. Such agents can predict and decide autonomously and can be trained to understand the underlying complexities of the traffic in real-time. In this paper, an innovative framework to perform real-time traffic optimal management decisions is proposed. Its rationale uses a fusion of data observations and simulation to enable an autonomous agent capable of accurate adaptive traffic management. A Case Study of application is developed using the M50 motorway in Dublin, where the speed limits are applied as adaptive parameters for optimal traffic management. Results show that the intelligent agent can autonomously predict travel times and decide in real-time the optimal speed limits to impose on a motorway when signs of congestion are found. The agent can reduce the mean travel time of a time interval by up to 55 % and the mean waiting time by up to 69 % in a situation of congestion. The average travel times of the studied M50 junction have significantly improved, showing the potential of autonomous agents in enhancing real-time optimal traffic management.

In traffic networks, some elements are more prone to suffer or to create disruptive situations, and the identification of these elements becomes a challenge due to the large number of possible threats. The following paper presents two new methodologies to identify and rank vulnerable and critical links of traffic networks. These methodologies use the Fisher Information Matrix, and the analysis of eigenvalues and eigenvectors, to systematically rank the links of a network. The identification is done by using traffic variables, such as the demand, the travel time, and the network's flow. For the ranking of the links, disruptions are considered in all the possible locations of the network, and the effects are systematically evaluated. In addition, the evaluation of traffic resilience is included in the process to validate the results. Finally, both methodologies are tested in a real network to infer on the validity of the results.

Different emerging threats highlighted the relevance of recovery and adaptation modelling in the functioning of societal systems. However, as modelling of systems becomes more complex, its effort increases challenging the practicality of the engineering analyses required for efficient recovery and adaptation. In the present work, metamodels are researched as a tool to enable these analyses in traffic networks. One of the main advantages of metamodeling is their synergy with the short decision times required in recovery and adaptation. A sequential global metamodeling technique is proposed and applied to three macroscopic day-to-day user-equilibrium models. Two reference contexts of application are researched: optimal recovery to a perturbation (with response times reduced by 98% with loss of accuracy lower than 1%) and adaptation under uncertainty with perturbation-dependent optimality. Results show that metamodeling-based metaheuristics enable fast resource-intensive engineering analyses of traffic recovery and adaptation, which may change the paradigm of decision-making in this field.

Significant research has been conducted in identifying optimal recovery and adaptation decisions in disruptive scenarios using engineering models. In this context, an aspect that has been target of limited research is that of response times. Modelling is expected to grow progressively more complex as it becomes more accurate. Such complexity increases modelling efforts, and the promise of optimal adaptation and recovery may become hindered. The present work discusses the role of modelling fidelities in adaptation and recovery of systems, and in particular that of using a lower fidelity model that enables zero-time analyses of a system. A framework is proposed for using different fidelities in adaptation and recovery, considering system's decision time requirements. The relevance of this analysis is researched in two traffic networks and results show that multi-fidelity models should be expected to play a key role in increasing the efficiency of optimal adaptation and recovery decisions.

The present work discusses an innovative approach to metamodeling in reliability that uses a field-transversal rationale. Adaptive metamodeling in reliability is characterized by its large spectra of models and techniques with different assumptions. As a result, the reliability engineer is frequently faced with the highly challenging task of selecting an appropriate model or technique with limited a priori knowledge about the performance function that defines the problem of reliability. To tackle this challenge, a complement-basis is proposed for adaptive metamodeling. It consists in using a batch of multiple metamodels or techniques that, accordingly to an activation criterion, are selected to solve the reliability analysis. This activation is set to depend on the model synergy with the problem in-hand. In the present work the leave-one-out loss is applied as evaluator of compatibility, and results show that the absolute loss successfully performs as an activator. A metamodel-independent learning approach and stopping criterion are implemented to study the proposed approach in five representative examples. Results show that the complement-basis allows to increase the efficiency of the reliability analysis through the selection of adequate metamodels, which is indicative of the untapped potential that further transversal research may add to metamodeling in reliability analysis.

Reliability assessment with adaptive Kriging has gained notoriety due to the Kriging capability of accurately replacing the performance function while performing as a self-improving function for learning procedures. Recent works on adaptive Kriging pursued to improve the efficiency of the active learning through the application of distinct learning functions, sampling methods, or frameworks to assess the learning space. Within this context, the present work exploits three innovative applications of density scanning to improve the efficiency of the adaptive Kriging. Density scanning has significant synergies with adaptive Kriging implementation. For most learning criteria, candidate points occur in dense clusters. This is due to the fact that the most efficient learning strategies pursue to improve predictions near the failure region, or when the prediction uncertainty is large. Identifying dense clusters of points, and fomenting exploitation of these, parallelizing computations, and limiting the generation of dense clusters in the design of experiments are examples of learning frameworks that can be achieved with density scanning. Three reference examples are researched in the present work, a complex function, a series system, and a relatively high dimension engineering problem. For all the cases, the application of density scanning is identified to improve the active learning efficiency.