Claudio Ciancio

Heuristic techniques to optimize neural network architecture in manufacturing applications

Abstract Nowadays application of neural networks in the manufacturing field is widely assessed even if this type of problem is typically characterized by an insufficient availability of data for a robust network training. Satisfactory results can be found in the literature, in both forming and machining operations, regarding the use of a neural network as a predictive tool. Nevertheless, the research of the optimal network configuration is still based on trial-and-error approaches, rather than on the application of specific techniques . As a consequence, the best method to determine the optimal neural network configuration is still a lack of knowledge in the literature overview. According to that, a comparative analysis is proposed in this work. More in detail four different approaches have been used to increase the generalization abilities of a neural network. These methods are based, respectively, on the use of genetic algorithms, Taguchi, tabu search and decision trees. The parameters taken into account in this work are the training algorithm, the number of hidden layers, the number of neurons and the activation function of each hidden layer. These techniques have been firstly tested on three different datasets, generated through numerical simulations in the Deform2D environment, in an attempt to map the input–output relationship for an extrusion, a rolling and a shearing process. Subsequently, the same approach has been validated on a fourth dataset derived from the literature review for a complex industrial process to widely generalize and asses the proposed methodology in the whole manufacturing field. Four tests were carried out for each dataset modifying the original data with a random noise with zero mean and standard deviation of one, two and five per cent. The results show that the use of a suitable technique for determining the architecture of a neural network can generate a significant performance improvement compared to a trial-and-error approach.

A Stochastic Maximal Covering Formulation for a Bike Sharing System

This paper discusses a maximal covering approach for bike sharing systems under deterministic and stochastic demand. Bike sharing is constantly becoming a more popular and sustainable alternative transportation system. One of the most important elements for the design of a successful bike sharing system is given by the location of stations and bikes. The demands in each zone for each period is however uncertain and can only be estimated. Therefore, it is necessary to address this problem by taking into account the stochastic features of the problem. The proposed model determines the optimal location of bike stations, and the number of bikes located initially in each station, considering an initial investment lower than a given predetermined budget. The objective of the model is to maximize the percentage of covered demand. Moreover, during the time horizon, it is possible to relocate a certain amount of bikes in different stations with a cost proportional to the traveled distance. Both deterministic and stochastic models are formulated as mixed integer linear programs.

Branch-price-and-cut for the Mixed Capacitated General Routing Problem with Time Windows

The Mixed Capacitated General Routing Problem with Time Windows (MCGRPTW) is defined over a mixed graph, for which some nodes, arcs and edges have strictly positive demand and must be serviced. The problem consists of determining a set of least-cost vehicle routes that satisfy this requirement, while respecting pre-specified time windows and without exceeding the vehicle capacity. In this work, we transform the MCGRPTW into an equivalent node routing problem over a directed graph. Thus, we solve the equivalent problem by using a branch-price-and-cut algorithm which relies on some effective techniques introduced in the field. Computational experiments over instances derived from the Capacitated Arc Routing Problem with Time Windows and from the Mixed Capacitated General Routing Problem are presented. The article also describes experiments over benchmark instances.

A Multi-Objective Optimization of a Porthole Die Extrusion for Quality and Sustainability Issues

Nowadays manufacturing companies have to face conflicting issues continuously. Solving this type of problem means finding solutions that ensure a fair compromise between different objectives. In this work, a porthole die extrusion is considered as a specific case study. Usually, the main objective of this process is to find the combination of input parameters that allow the product quality to be maximized. However, product quality is not the only variable that companies have to take into account. In fact, it is also necessary to design the process in an efficient and sustainable way in order to reduce process cost and environmental impact. To this purpose, in this study the conflicting aims of product quality maximization and energy assumption minimization are considered and optimized. To pursue this aim an experimental investigation was executed, in order to build a preliminary database. The decision variables are the profile thickness and the process velocity. During the tests, the punch was measured in order to quantify the absorbed power along with the environmental impact of the process for changing conditions. In the same way, the mechanical properties of the extruded profile were measured by means of a tensile test, in order to assess the product quality. To solve this kind of problem the use of multi-objective optimization techniques is required in order to find the set of Pareto optimal solutions from which a single configuration will be selected according to specific business needs.