31-12-2009, 07:47 PM
The main objective of the current research effort is to explore the applicability of using principles from the field of Intelligent Adaptive Control for the on-line management and control of transportation systems. The main objective of the current study is to explore the applicability of using principles from the field of Intelligent Adaptive Control for the on-line management and control of transportation systems.Intelligent adaptive control is a field that encompasses the computational procedures of Fuzzy Logic,Artificial Neural Networks, and Genetic Algorithms. The aim of intelligent control is to design robust and learning controllers operating in an uncertain environment with very limited mathematical knowledge of the controlled process principles, which makes it very much applicable to the control of transportation systems.
a neuro-fuzzy controller is developed for predictive feedback control.The controller is developed for a test network, and its effectiveness will be evaluated using simulation.The study is also expected to result in the development of a prototype neuro-fuzzy controller for the on-line management of transportation systems. focus will be on exploring the feasibility of implementing a neuro-fuzzy controller for predictive feedback control. In this approach, an ANN will be used to predict future travel time, and a fuzzy controller will then try to balance the predicted travel times (as obtained from the NN) on the alternate routes between a given O-D pair.
Select Test Network:
The main criteria will be would be the availability of a calibrated transportation model for the network, since this would ensure that Origin-Destination, along with basic geometric and traffic data, are available.
Develop CORSIM Model for the Test Network
a CORSIM model will be developed for the test network which will be used in the study to represent the real-world and also for evaluating the fuzzy controller.
Develop the Predictive NN Model
A NN model is developed for predicting network traffic conditions.use of a predictive NN model might be quite advantageous as the transport systems are quite complex. given the ability of NNs to approximate nonlinear functions
Develop the Fuzzy Logic Controller
There are three methods:
1)the controller is designed to emulate a human controller
2)based upon designing a heuristic fuzzy controller
3)the traffic network is modeled as a fuzzy system, and the controller is designed based upon the analysis of that system.
Evaluate the Effectiveness of the Developed Controller
effectiveness of the developed controller is evaluated by computing the time savings resulting from the use of the controller.
[attachment=973]
[attachment=974]
a neuro-fuzzy controller is developed for predictive feedback control.The controller is developed for a test network, and its effectiveness will be evaluated using simulation.The study is also expected to result in the development of a prototype neuro-fuzzy controller for the on-line management of transportation systems. focus will be on exploring the feasibility of implementing a neuro-fuzzy controller for predictive feedback control. In this approach, an ANN will be used to predict future travel time, and a fuzzy controller will then try to balance the predicted travel times (as obtained from the NN) on the alternate routes between a given O-D pair.
Select Test Network:
The main criteria will be would be the availability of a calibrated transportation model for the network, since this would ensure that Origin-Destination, along with basic geometric and traffic data, are available.
Develop CORSIM Model for the Test Network
a CORSIM model will be developed for the test network which will be used in the study to represent the real-world and also for evaluating the fuzzy controller.
Develop the Predictive NN Model
A NN model is developed for predicting network traffic conditions.use of a predictive NN model might be quite advantageous as the transport systems are quite complex. given the ability of NNs to approximate nonlinear functions
Develop the Fuzzy Logic Controller
There are three methods:
1)the controller is designed to emulate a human controller
2)based upon designing a heuristic fuzzy controller
3)the traffic network is modeled as a fuzzy system, and the controller is designed based upon the analysis of that system.
Evaluate the Effectiveness of the Developed Controller
effectiveness of the developed controller is evaluated by computing the time savings resulting from the use of the controller.
[attachment=973]
[attachment=974]
