Insight - Centre for Data Analytics
http://hdl.handle.net/10468/2481
2019-01-18T01:35:34ZAn ontology-based approach to knowledge representation for Computer-Aided Control System Design
http://hdl.handle.net/10468/7145
An ontology-based approach to knowledge representation for Computer-Aided Control System Design
Benavides, Carmen; Garcia, Isaias; Alaiz, Hector; Quesada, Luis
Different approaches have been used in order to represent and build control engineering concepts for the computer. Software applications for these fields are becoming more and more demanding each day, and new representation schemas are continuously being developed. This paper describes a study of the use of knowledge models represented in ontologies for building Computer Aided Control Systems Design (CACSD) tools. The use of this approach allows the construction of formal conceptual structures that can be stated independently of any software application and be used in many different ones. In order to show the advantages of this approach, an ontology and an application have been built for the domain of design of lead/lag controllers with the root locus method, presenting the results and benefits found.
2018-10-30T00:00:00ZPushing the frontier of minimality
http://hdl.handle.net/10468/6287
Pushing the frontier of minimality
Escamocher, Guillaume; O'Sullivan, Barry
The Minimal Constraint Satisfaction Problem, or Minimal CSP for short, arises in a number of real-world applications, most notably in constraint-based product configuration. It is composed of the set of CSP problems where every allowed tuple can be extended to a solution. Despite the very restrictive structure, computing a solution to a Minimal CSP instance is NP-hard in the general case. In this paper, we look at three independent ways to add further restrictions to the problem. First, we bound the size of the domains. Second, we define the arity as a function on the number of variables. Finally we study the complexity of computing a solution to a Minimal CSP instance when not just every allowed tuple, but every partial solution smaller than a given size, can be extended to a solution. In all three cases, we show that finding a solution remains NP-hard. All these results reveal that the hardness of minimality is very robust.
2018-06-07T00:00:00ZSemi-online task assignment policies for workload consolidation in cloud computing systems
http://hdl.handle.net/10468/5268
Semi-online task assignment policies for workload consolidation in cloud computing systems
Armant, Vincent; De Cauwer, Milan; Brown, Kenneth N.; O'Sullivan, Barry
Satisfying on-demand access to cloud computing infrastructures under quality-of-service constraints while minimising the wastage of resources is an important challenge in data centre resource management. In this paper we tackle this challenge in a semi-online workload management system allocating tasks with uncertain duration to physical servers. Our semi-online framework, based on a bin packing approach, allows us to gather information on incoming tasks during a short time window before deciding on their assignments. Our contributions are as follows: (i) we propose a formal framework capturing the semi-online consolidation problem; (ii) we propose a new dynamic and real-time allocation algorithm based on the incremental merging of bins; and (iii) an adaptation of standard bin packing heuristics with a local search algorithm for the semi-online context considered here. We provide a systematic study of the impact of varying time-period size and varying the degrees of uncertainty on the duration of incoming tasks. The policies are compared in terms of solution quality and solving time on a data-set extracted from a real-world cluster trace. Our results show that, around periods of high demand, our best policy saves up to 40% of the resources compared to the other polices, and is robust to uncertainty in the task durations. Finally, we show that small increases in the allowable time window allows a significant improvement, but that larger time windows do not necessarily improve resource usage for real world data sets.
2018-01-03T00:00:00ZAn analytics-based decomposition approach to large-scale bilevel optimisation
http://hdl.handle.net/10468/6568
An analytics-based decomposition approach to large-scale bilevel optimisation
Fajemisin, Adejuyigbe
Bilevel optimisation problems contain several decision makers, each with different objectives and constraints, arranged in a hierarchical structure. One type of bilevel problem is the single-leader, multiple-follower problem, which has been used in applications like toll-setting, resource management, conflict resolution, and many others. This hierarchical structure allows for the reformulation of the forest harvesting problem as a multiple-follower bilevel problem. In the forest harvesting problem of Chapter 4, trees are cut into different log types, some of which are more valuable than others. Due to the fact that harvesting machines are designed to prioritise the production of these higher-value log types, over-production and waste of the high-value logs, as well as unfulfilled demand for the low-value logs is seen. Additionally, the discrepancy between amounts of log types expected pre-harvest and the actual amounts seen post-harvest leads to the inefficient harvesting of the forest. Despite the many approaches for solving multiple-follower problems, they are either not applicable in cases in which the follower problems are not traditional optimisation problems, or do not scale up appropriately. An example of this case occurs with the forest harvesting problem, where the follower problems are dynamic programming problems. Another example is the case where the follower problems are black-box functions. In such cases, replacing the follower problems with reformulations or optimality conditions are not applicable. Evolutionary algorithms can be used, but these are computationally-intensive schemes which do not scale up effectively. For this reason, an analytics-based approach, which is better able to sample the solution space is needed. The thesis defended throughout this dissertation is that an analytics-based decomposition approach can be used to solve largescale multiple-follower bilevel problems more efficiently than the other approaches. Specifically, the contributions of this thesis are: (i) a new class of multiple-follower bilevel problems is proposed; (ii) a novel analytics-based decomposition approach for solving this class of large-scale bilevel multiple-follower problems is given; (iii) the forest harvesting problem is reformulated as a bilevel optimisation problem to take into the account operation of harvester, and (iv) a reactive harvesting approach is developed to mitigate the effects of the uncertainty in the data .
2018-01-01T00:00:00Z