Tad Gonsalves

Multi objective particle swarm optimization algorithm for the design of efficient ATO speed profiles in metro lines

One of the strategies for the reduction of energy consumption in railways systems is to execute efficient drivings (eco-driving). This eco-driving is the speed profile that requires the minimum energy consumption without degrading commercial running times or passenger comfort. When the trains are equipped with Automatic Train Operation systems (ATO) additional difficulties are involved. Their particular features make it necessary to develop accurate models that optimize the combination of the ATO commands of each speed profile to be used by the traffic regulation system. These commands are transmitted to the train via encoded balises on the track with little channel capacity (bandwidth). Thus, only a few and discrete values of the commands can be sent and the solution space of every interstation is made up of a relatively small set of speed profiles. However, the new state-of-the-art of signalling technologies permit a better bandwidth resulting in an exponential solution space. This calls for new methods for the optimal design of the ATO speed profiles without an exhaustive simulation of all the combinations. A MOPSO algorithm (Multi Objective Particle Swarm Optimization) to obtain the consumption/time Pareto front based on the simulation of a train with a real ATO system is proposed. The algorithm is able even to take into account only the comfortable speed profiles of the solution space. The fitness of the Pareto front is verified by comparing it with a NSGA-II algorithm (non-dominated sorting genetic algorithm II) and with the real Pareto front. Further, it has been used to obtain the optimal speed profiles in a real line of the Madrid Underground.

Swarm Intelligence in the Optimization of Software Development Project Schedule

The Software Development Project Scheduling Problem is similar to the well-known Resource-Constrained Multi-Project Scheduling Problem (RCMPSP). It consists in determining a schedule of tasks taking into consideration resource availabilities and precedence constraints, while optimizing an objective. Like RCMPSP, it is an NP-hard problem. In this paper, a task segmentation scheme to schedule a software development project is proposed and the average duration of the multiple concurrent projects is minimized using the Particle Swarm Optimization (PSO) meta-heuristic. PSO is a recent meta-heuristic algorithm, known for its simplicity in programming and its rapid convergence. A series of experiments show optimum results for several software development schedule scenarios.

Parallel swarms oriented particle swarm optimization

The particle swarm optimization (PSO) is a recently invented evolutionary computation technique which is gaining popularity owing to its simplicity in implementation and rapid convergence. In the case of single-peak functions, PSO rapidly converges to the peak; however, in the case of multimodal functions, the PSO particles are known to get trapped in the local optima. In this paper, we propose a variation of the algorithm called parallel swarms oriented particle swarm optimization (PSO-PSO) which consists of a multistage and a single stage of evolution. In the multi-stage of evolution, individual subswarms evolve independently in parallel, and in the single stage of evolution, the sub-swarms exchange information to search for the global-best. The two interweaved stages of evolution demonstrate better performance on test functions, especially of higher dimensions. The attractive feature of the PSOPSO version of the algorithm is that it does not introduce any new parameters to improve its convergence performance. The strategy maintains the simple and intuitive structure as well as the implemental and computational advantages of the basic PSO.

GA optimization of Petri net-modeled concurrent service systems

This paper deals with the performance modeling and the optimization of concurrent service systems. In large and complex service systems, asynchronous and concurrently occurring activities are common. Petri nets are ideal tools for modeling concurrent systems. However, Petri nets are lacking in time duration concept, in data collecting mechanism and in conjunctive logic on the preconditions of an event. These inherent limitations along with the state explosion problem severely restrict their scope of application. In this paper, we introduce the Client Server Petri net, which overcomes all these limitations. The Client Server Petri net is an extension of the Generalized Stochastic Petri net that allows greater flexibility in modeling and simulating concurrent systems. The total operational cost of service systems consists of service cost and waiting cost. The former is due to the hiring of service personnel, while the latter is due to the fact that customers weary of waiting may take their business somewhere else. The problem, in principle, can be formulated as a multi-objective optimization problem and then solved to obtain the Pareto-front. In this study, however, we formulate it as a single-objective optimization problem because optimization (minimization) of the total cost (service cost+waiting cost) is of paramount importance in economic models. Finding the optimal operational cost becomes a combinatorial optimization problem which we seek to minimize using the genetic algorithm, known for its robustness and versatility as an optimization meta-heuristic. We demonstrate the effectiveness of the novel Client Server Petri net model editor-simulator-optimizer with the practical example of an automobile purchase concurrent service system.

Cost Minimization in Service Systems Using Particle Swarm Optimization

This paper deals with the optimization of the operational costs of service systems. The cost function consists of service costs and waiting costs. Service cost is associated with the employment of service-providing personnel, while the waiting cost is associated with the customers having to wait for the service. The cost function is minimized subject to the server utilization as well as to the customer satisfaction constraints, using the Particle Swarm Optimization (PSO) algorithm. PSO is a fairly recent swarm intelligence meta-heuristic algorithm known for its simplicity in programming and its rapid convergence. The optimization procedure is illustrated with the example of a practical service system. A series of experiments show optimum results for the operation of the service systems.

Generic Core Life Cycle and Conceptual Architecture for the Development of Collaborative Systems

In the conventional system development life cycle (SDLC), the system performance evaluation phase comes after the implementation phase. Our strategy is to project system performance estimate at the requirement analysis and design phase itself much before the implementation phase. To achieve this objective, we propose a technology-neutral integrated environment for the core life cycle of system development. This core life cycle consists of three phases: system modelling, performance evaluation and performance improvement.

White Lane Detection Using Semantic Segmentation

This paper deals with the application of machine learning techniques to the detection of white lanes for autonomous driving assistance using only a single visual camera. When performing white line detection, a method called semantic segmentation using fully convolutional network is used. This method is chosen to flexibly detect the shape of objects, since detection of white lanes cannot be done well with rectangular detection. The convolutional neural network which is characterized by the absence of fully connected layer outputs an image for a given input. FCN-8s are used for fully convolutional network. FCN-8s has an easy-to-understand structure and an advantage of being easy to create. In addition, we also created a dataset manually by extracting white lines from public roads and used it for training and testing the learning algorithm. Our segmentation algorithm is found to accurately detect the white lane markings from the dataset.

Autonomous Highway Car Following System Based on Fuzzy Control

The ability of controlling real-time vehicle status is extremely important for car-following systems. The status of the traffic stream on highways has impact on the vehicle control while in a high driving speed. In this paper, a feedback car-following system based on fuzzy control is proposed for autonomous vehicle status control to maneuver the vehicle in safe situation smoothly. The proposed system is based on simulated traffic network. The fuzzy-controller is able to adjust the time-varying features for controlling the status of the vehicle. Comparing with human control behavior, the fuzzy-controller has a better performance in analyzing all the information relating to the traffic situation. The simulation results show that the proposed system is efficient in controlling the vehicle status to a safe situation on highways.

Data Clustering through Particle Swarm Optimization Driven Self-Organizing Maps

The Self-Organizing Map (SOM) is a well-known method for unsupervised learning. It can project high-dimensional data onto a low-dimensional topology map which makes it an efficient tool for visualization and dimensionality reduction. The Particle Swarm Optimization (PSO) is a swarm-intelligence meta-heuristic optimization algorithm based on the feeding behavior of a swarm of birds. In this paper, we have combined these two diverse approaches to form a PSO-SOM which is applied to clustering problems. The advantage of this method is the reduction in computational complexity and increase in clustering accuracy as demonstrated by the experimental results.

Two Diverse Swarm Intelligence Techniques for Supervised Learning

Particle Swarm Optimization PSO and Enhanced Fireworks Algorithm EFWA are two diverse optimization techniques of the Swarm Intelligence paradigm. The inspiration of the former comes from animate swarms like those of birds and fish efficiently hunting for prey, while that of the latter comes from inanimate swarms like those of fireworks illuminating the night sky. This novel study, aimed at extending the application of these two Swarm Intelligence techniques to supervised learning, compares and contrasts their performance in training a neural network to perform the task of classification on datasets. Both the techniques are found to be speedy and successful in training the neural networks. Further, their prediction accuracy is also found to be high. Except in the case of two datasets, the training and prediction accuracies of the Enhanced Fireworks Algorithm driven neural net are found to be superior to those of the Particle Swarm Optimization driven neural net.