More and more organizations are using optimization, simulation and big data techniques to make better decisions and reduce risks. Big companies are improving their economic activities by investing in business analytics or operation research. However, small companies are not able to afford this kind of inversions. This article proposes a new software as a service (SAAS) to bring the huge potential and benefits of linear optimization to daily activity of small companies and common people. The service described in this article allows solving optimization models through a simple step-by-step user interface. Thus, the only requirement to execute a model and analyze its results is an electronic device connected to the net. The results obtained show the usability and the competitive advantages of using the proposed service by decisions makers in whatever real life activity.View Service
Optimization models are being more used to tackle agro-business prob- lems. This way, cloud computing, machine learning, big data, internet of things and optimization are key factors in the innovation of the sector. This paper proposes a cloud architecture to suggest a ranking of candidate sows to be replaced and a set of tools to help the farmers to make better strategic, tactical and operational decisions related with the structure of the sow farms. The cloud architecture automatizes the process to obtain the results, making the process transparent to the farmer. This work extends the advantages of optimization models with the potential of cloud computing. The results shows that the cloud architecture proposed helps decisions makers in real pig farms to make a better planning by obtaining the competitive advantages of using the proposed model in a usable, exible and simple way.View work
This project consists in two parts due to the double degree in Computer Engineering and Administration and Business Management requirements. The first one presents and approaches a stochastic mixed integer linear programming model. The model conceived for pig production planning has the objective to optimize the entire pig supply chain according to the number of farms operating for the same company or cooperative. The model maximizes the revenue calculated from the income of sales to the abattoir and the production costs. Production costs depends on each type of farm involved in the process. Factors like farm capacity, trucks available and transportation costs are considered, among others. The model considers a medium-term planning horizon and specifically provides optimal transport planning in terms of number of animals to be transported from one facility to another and the number of trucks to do so. The algebraic modelling software OPL studio, in combination with the solver CPLEX both from IBM ILOG have been used to solve the model and study its properties and behaviour. In addition, a potential improvement in the execution time has been sought considering that a real application of the model is expected. This has been done through the generation of instances variating the number of farms that take part in each execution. The instances are required to apply a genetic algorithm on the search space defined by the cartesian product of the solver domains. Once a given time limit elapses, it returns the value of the parameters for which it has had a superior performance on average over the group of instances. Then, the results are analysed and the variations of the model execution time are studied. The second part of the project consists in developing a decision support system (DSS), which takes form of an application that approaches and facilitates the use of the model in order to provide each user a personalized solution. In addition, a friendly interface displays a google maps personalized instance presenting all the facilities that users, companies and cooperatives or even farmers, have introduced to the system. The information introduced will be used to run the model and generate the specific solution for the user. In addition, the interface also allows users to introduce new facilities to the system and visualize their current information and characteristics.View work
Power system expansion specifically for distribution networks is gaining more importance due to the integration of distributed energy systems. Optimization models are more used to tackle generation and transmission expansion problems (GTEP). This way, cloud computing, machine learning, big data, internet of things, simulation and optimisation are critical factors in the innovation of the power sector. In the recent past, a new inter-disciplinary subject focusing on energy and information system called energy informatics has emerged. This paper proposes a novel mathematical model to deal with GTEP regarding the collaboration and competition between all the nodes and actors in the power network.
Additionally, this work proposes the usage of a parallel algorithm to solve the GTEP problem using the model efficiently. This way, to assist power network companies to make better strategic, tactical and operational decisions related to the investments, maintenance or evaluation of the power network a cloud-based Decision Support System (DSS) is proposed.
Mainly, focused at: (i) integrate the data in the system, (ii) integrate the model, (iii) automate the resolution process, and finally present the results in an interactive way to the end-users. This work extends the advantages of optimisation and simulation models with the potential of parallel and cloud computing to automate and offer the knowledge and the analytics. The results show that the decision support system proposed helps decisions makers in real situations to do better planning by obtaining the competitive advantages of using the proposed model in a usable, flexible and straightforward way.View work
This project consists in developing a decision support system and its integration in a cloud platform to offer optimization models as a service integrated in a sole application accessible from different devices. It has two parts due to the double degree in Computer Engineering and Administration and Business Management requirement.
The part related to the Administration and Business Management degree is based on a mixed integer linear-programming (MILP) model to support marketing decisions on fattening farms without individual weight control. It includes the development of the running program to solve the MILP model and the assessment for specific factors. The objective is to determine the optimal delivery time and quality bonus of the fattened pigs to the slaughterhouse and thus, getting the best value for the producer. This mathematical model maximizes the revenue and reduces the risk of neglecting opportunity costs by taking into account important key factors such as: homogeneity of the batch, price-grid system, transportation cost, among others. The analysis performed on this study demonstrated the importance of these different factors that affect significantly on the optimal marketing policy of fattening farms, and also in a vertically integrated company.
The second part is related to the Computer Engineering degree and concerns the development of a cloud decision support system (cDSS). The cDSS takes form of a client application that allows users to run the mathematical model introduced previously and through it, depict their best marketing decisions for a pig farmer. Through this application, the farmer can store data and update information of the farms, make requests from different devices or visualize an output display of the results obtained from the model.View work
This work presents a general parallelisation of the Progressive Hedging algorithm to coordinate the resolution of two-stage and multi-stage stochastic mixed-integer problems without (binary or integer) variables in the first stage. We report a benchmark study between the computational improvements using our proposal and the parallel version (using pyro) of the Pyomo integrated Progressive Hedging. Moreover, we study the influence of a quadratic term to accelerate the convergence, different scenario-cluster formation and several step update policies by solving different instances using our proposal.View work
Microservices and decoupled applications are rising in popularity. These architectures, based on containers, have facilitated the efficient development of complex SaaS applications. A big challenge in eHealth is to manage and design microservices with a massive range of different facilities, from processing and data storage to computing predictive and prescriptive analytics. Moreover, these systems require the capacity to integrate into current medical systems while meeting the Quality of Service (QoS) constraints. The primary purpose of this work is to present a cloud architecture based on containers aimed towards guaranteeing a determined level of QoS regarding cost, resource usage, and service level agreement policies focused on assisting medical staff in their daily tasksView work View Code