Gully pots are part of the infrastructure of a storm drain system, designed to drain surface water from streets. Any broken or blocked gully pot represents a potential cause of flooding, for instance during periods of intense or prolonged rainfall. Regular cleaning is necessary for gully pots to function effectively. We model gully pot maintenance as a risk-driven problem, and evaluate the maintenance quality of maintenance strategies by considering the risk impact of gully pot failure and failure behaviour. The results suggest investment directions and management policies that potentially improve the efficiency of maintenance. We find that the current maintenance quality is significantly affected by untimely system status information. We propose that low-cost sensor techniques might be able to improve timeliness of status information, and use simulation results to show the behaviour and advantages that might arise in a range of real-world scenarios.

Gully pots or storm drains are located at the side of roads to provide drainage for surface water. We consider gully pot maintenance as a risk-driven maintenance problem. We explore policies for preventative and corrective maintenance actions, and build optimised routes for maintenance vehicles. Our solutions take the risk impact of gully pot failure and its failure behaviour into account, in the presence of factors such as location, season and current status. The aim is to determine a maintenance policy that can automatically adjust its scheduling strategy in line with changes in the local environment, to minimise the surface flooding risk due to clogged gully pots. We introduce a rolling planning strategy, solved by a hyper-heuristic method. Results show the behaviour and strength of the automated adjustment in a range of real-world scenarios. (C) 2016 Elsevier B.V. All rights reserved.

Local search based meta-heuristics such as variable neighbourhood search have achieved remarkable success in solving complex combinatorial problems. Local search techniques are becoming increasingly popular and are used in a wide variety of meta-heuristics, such as genetic algorithms. Typically, local search iteratively improves a solution by making a series of small moves. Traditionally these methods do not employ any learning mechanism. We treat the selection of a local search neighbourhood as a dynamic multi- armed bandit (D-MAB) problem where learning techniques for solving the D-MAB can be used to guide the local search process. We present a D-MAB neighbourhood search (D-MABNS) which can be embedded within any meta- heuristic or hyperheuristic framework. Given a set of neighbourhoods, the aim of D-MABNS is to adapt the search sequence, testing promising solutions rst. We demonstrate the eectiveness of D-MABNS on two vehicle routing and scheduling problems, the real-world geographically distributed mainte- nance problem (GDMP) and the periodic vehicle routing problem (PVRP). We present comparisons to benchmark instances and give a detailed analysis of parameters, performance and behaviour. Keywords Meta-heuristic Local search Vehicle routing

Gaist Solutions Ltd. carries out national-scale road inspection surveys in the UK. Visual inspection is used to identify the need for road maintenance. An inspection vehicle that monitors one side of the road needs two traversals to monitor a typical road, whereas a vehicle with cameras that record both sides of the road only requires a one-pass approach. To determine whether the one-pass approach affords any real cost advantage, we analyse road networks of six typical UK cities and the county of Norfolk using a range of exact and heuristic methods, and extrapolate from our results to estimate the cost-effectiveness of these two approaches for the road network of the UK. Our analysis approach is based on the Chinese Postman Problem (CPP), using graph reduction to allow effective computation over very large data sets.

Gaist Solutions Ltd. carries out large scale surveys for UK road inspection. To estimate the total distance that vehicles travel, we model routing as a Chinese Postman Problem. We propose a novel graph reduction approach that dramatically speeds up the calculation of the Chinese Postman Tour for large-scale road networks. Because the analysis of large road-network graphs is now possible, planners can explore the effects of changes to traditional inspection techniques and scheduling. Case studies of road networks from six UK cities and the county of Norfolk are tested. The graph reduction process is also analysed on ten randomly generated road networks with different characteristics, to show its ability and give advice for suitable use.

As software systems become more pervasive and increase in both size and complexity, the requirement for systems to be able to selfadapt to changing environments becomes more important. This paper describes a model-based approach for adapting to dynamic runtime environments using metaheuristic optimisation techniques. The metaheuristics exploit metamodels that capture the important components in the adaptation process. Firstly, a model of the environment’s behaviour is extracted using a combination of inference and search. The model of the environment is then used in the discovery of a model of optimal system behaviour – i.e. how the system should best behave in response to the environment. The system is then updated based on this model. This paper focuses on investigating the extraction of a behaviour model of the environment and describes how our previous work can be utilised for the adaptation stage. We contextualise the approach using an example and analyse different ways of applying the metaheuristic algorithms for discovering an optimal model of the case study’s environment.

The human immune system has characteristics such as self-organisation, robustness and adaptivity that may be useful in the development of adaptive systems. One suitable application area for adaptive systems is Information Filtering (IF). Within the context of IF, learning and adapting user profiles is an important research area. In this paper, we proposed an architecture for adaptive IF incorporating ideas from Artificial Immune Systems (AIS). We then extract general features of potential immune inspired profile adaptation based on scenario-based modeling. The richness of scenario usage which we have encountered in this work indicates the importance of comprehensive process guidance.

The relation between individuality and aggregation is an important topic in complex systems sciences, both aspects being facets of emergence. This topic has frequently been addressed by adopting a classical, individual versus population level perspective. Here, however, the frontiers that emerge in segregated communities are the focus; segregation is synonymous with the existence of frontiers that delineate and interface aggregates. A generic agent-based model is defined, with which we simulate communities located on grid and scale-free networked environments. Emerging frontiers are analyzed in terms of their relative occupancy, porosity, and permeability. Results emphasize that the frontier is highly sensitive to the topology of the environment, not only to the agent tolerance. These relations are clarified while addressing the topics of frontier robustness and the trade-off between its capacity to separate and allow exchange.

This paper explores approaches to Adaptive Information Filtering (AIF) in the context of changing user interests. Based on the existing artificial immune system for email classification (AISEC), we demonstrate an effective extension to classification based on the body of emails. Widening this to the problem of AIF on dynamic web content, we propose to explore dynamic clonal selection algorithms (DCSAs) that include dynamically changing thresholds.

Model comparison is an essential prerequisite for a number of model management tasks in Model-Driven Engineering, including model differencing, merging and transformation testing. Previous work highlighted some of the shortcomings of existing approaches to model comparison, and proposed a new language that addressed some of the key limitations. In this paper we present additional requirements that we have identified as a result of applying this particular language, the Epsilon Comparison Language, as well as possible solutions to the requirements. We aim to stimulate discussion of requirements and solutions for model comparison, both at a conceptual level and in terms of tool support for the particular model comparison language in question.

Research based on computer simulations, especially that conducted through agent-based experimentation, is often criticised for not being a reliable source of information - the simulation software can hide errors or flawed designs that inherently bias results. Consequently, the academic community shows both enthusiasm and lack of trust for such approaches. In order to gain confidence is using engineered systems, domains such as Safety Critical Systems employ structured argumentation techniques as means of explicitly relating claims to evidence - in other words, requirements to deliverables. We argue here that structured argumentation should be used in the development and validation process of simulation-driven research. Making use of the Goal Structuring Notation, we provide insights into how more trustworthy outcomes can be obtained through argumentation-driven validation.

Progress in marine ecosystem modeling has seen a proliferation of the number of state variables and processes represented, in order to realistically describe system dynamics and feedbacks associated with, for example, changing climate. Assigning realistic and robust values to the many associated model parameters has become increasingly difficult due to underdetermination through lack of data and sensitivity to chosen parameterizations. Complexity science is becoming ever more relevant in this regard, with novel approaches coming to the fore based on traits, trade-offs and the theory of complex adaptive systems. We describe one such approach in which a global ocean circulation model was seeded with many tens of plankton functional types (PFTs) whose physiological characteristics were assigned stochastically at the outset. After the simulation was set in motion, competition eliminated unfavorable PFTs, giving rise to a robust self-organizing model architecture as an emergent property of the system.

EMF and GMF are powerful frameworks for implementing tool support for modelling languages in Eclipse. However, with power comes complexity; implementing a graphical editor for a modelling language using EMF and GMF requires developers to hand craft and maintain several low-level interconnected models through a loosely-guided, labour-intensive and error-prone process. In this paper we demonstrate how the application of model transformation techniques can help with taming the complexity of GMF and EMF and deliver significant productivity, quality, and maintainability benefits. We also present EuGENia, an open-source tool that implements the proposed approach, illustrate its functionality through an example, and report on the community's response to the tool.

The study of Complex Systems is growing rapidly, and modelling and simulation tools are an important part of the process. This volume brings together work from a multidisciplinary group of scientists, from biology and computer science, who are studying a variety of techniques and applications for modelling and simulating complex systems. A common theme emerging from much of this work is an emphasis on validation: how one can have confidence that a computer simulation is saying something sensible about the complex real-world domain of interest.

As part of our work on the formal analysis of object-oriented models, we turn to systems where many autonomous individuals interact to give rise to complex collective behaviour. We adapt our ZOO [1,2] structuring and apply it to a case study based on a published model of part of the immune system [3]. The formalisation calls for a bottom-up solution with no central control over individual units, and includes an approach to represent feedback channelsenabling broadcast communication between individuals and across levels.

We extend coarse graining of cellular automata to investigate aspects of emergence. From the total coarse graining approach introduced by Israeli and Goldenfeld, Coarsegraining of cellular automata, emergence, and the predictability of complex systems, Phys. Rev. E, 2006, we devise partial coarse graining, and show qualitative differences in the results of total and partial coarse graining. Mutual information is used to show objectively how coarse grainings are related to the identification of emergent structure. We show that some valid coarse grainings have high mutual information, and are thus good at identifying and predicting emergent structures. We also show that the mapping from lower to emergent levels crucially affects the quality emergence.