Abstract: The efficient creation of examination timetables is a recurring and important problem for universities worldwide. Good timetables typically are characterized by balanced distances between consecutive exams for all students. In this contribution an approach for the examination timetabling problem as defined in the second International Timetabling Competition (http://www.cs.qub.ac.uk/itc2007/) is presented. The solution approach is managed on the top level by GRASP (Greedy Randomized Adaptive Search Procedure) and it involves several optimization algorithms, heuristics and metaheuristics. A construction phase is executed first producing a relatively high quality feasible solution and an improvement phase follows that further ameliorates the produced timetable. Each phase consists of stages that are consumed in a circular fashion. The procedure produces feasible solutions for each dataset provided under the runtime limit imposed by the rules of the ITC07 competition. Results are presented and analyzed.
Abstract: We present how a standard task in the clearance of financial brokers can find new creative, integer optimization, computerized solutions. The broker makes wholesome
massive transactions of Stock Exchange securities and then wants to distribute them in a fair way to the retail customers, so that each one has almost the same average price of buying or selling.
Abstract: We present an approach for the examination timetabling problem as defined in the second International Timetabling Competition (http://www.cs.qub.ac.uk/itc2007). The solution approach can be considered as an implementation of the GRASP (Greedy Randomized Adaptive Search Procedure) method with the combination of several other metaheuristics. Three stages are employed. The first stage is responsible for the construction of a relatively high quality feasible solution while the second stage improves it using simulated annealing local search. The final stage uses mathematical programming and analyzes each examination period in isolation proposing movements of exams to other rooms resulting in further improvement of the solution quality. The procedure produces feasible solutions for each dataset provided under the runtime limit imposed by the competition’s rules. Results are presented and analyzed
Abstract: In this paper the problem of minimizing the electricity cost required by a water storage and disposal system is analyzed. Three solution approaches are presented focused in the way that various pumps will be scheduled to operate and satisfy prospected water demand while at the same time respect availability of water and reservoirs capacity. The first solution uses a heuristic approach closely related to the way a human operator might have used to solve the problem. The second solution uses mathematical programming, formulates the problem as an Integer Programming problem and solve it by using a Branch and Bound commercial solver. Finally the third solution uses Genetic Algorithms and defines a fitness function describing the attractiveness of each solution for a set of solutions and through a number of evolution steps creates a population with desirable characteristics. A comparative study of the three approaches is presented. The actual formulas that compute the electricity cost is used for the comparisons. Those formulas state the fact that demand peaks especially during high demand periods result in high electricity cost. Data used in our experiments were provided by the municipal enterprise of water supplies and sewage of Chania (Crete).
Abstract: This paper describes a system that generates intelligent alarms using sensor data located at various geographical locations. Users are informed, using a rule enabled engine whenever certain conditions demanding attention occur, through an SMS messaging system. The user can easily define, through a graphical user interface, new rules or update existing ones. An agricultural case study of the system is presented.
Abstract: This paper describes the ERMIS system, specially designed to support the fleet and personnel scheduling and management issues, as well as to support various business collaboration issues for a helicopter taxi company. The implemented system offers a Web-based interface for the flying and stationary personnel and a software application running on a PDA with GSM and GPS capabilities for the pilots. The information flow and the business processes of the ERMIS system are based on XML Web services which makes the system open and interoperable with other company-owned information systems.
Abstract: In this paper a mathematical model and a solution for the pump scheduling problem based on genetic algorithms is presented. The main objective is the reduction of the electricity costs of the water department for the pumping effort. The constraints are such as to maintain strategic security and reliability limits for each water reservoir. The reduction of the peaks during a scheduling period is equivalent to the minimization of the electricity costs and this is in fact used in the solution process of our experiment. Actual results with the pump scheduling results of Chania, Greece are also presented.
