Transportation system traffic controlling system using a neural network

A traffic controlling apparatus for a transportation system having traffic and traffic controllers, said traffic controlling apparatus comprising:

• a traffic volume detecting means for detecting traffic volumes in the transportation system;
• a traffic flow presuming means for presuming traffic flows from the traffic volumes detected by said traffic volume detecting means, the traffic flow presuming means including a neural network for determining relationships between traffic volumes and traffic flows, and a first filter means for filtering the traffic flows determined by the neural network;
• a presumption function constructing means for constructing and correcting the neural network of said traffic flow presuming means, wherein said presumption function constructing means contains a plurality of relationships between traffic flow patterns and traffic volumes, and constructs said neural network by using arbitrarily selected plural relationships among said relationships, and further corrects said neural network by using newly selected relationships between traffic flow patterns and traffic volumes on the basis of traffic flows presumed from actually measured traffic volumes and controlled results;
• a control parameter setting means for setting control parameters for controlling said traffic controllers on the basis of the traffic flow determined by the traffic flow presuming means.

A traffic controlling apparatus comprising:

• a traffic volume detecting means for detecting traffic volumes in a transportation system;
• a traffic flow presuming means for presuming traffic flows from the traffic volume detected by said traffic volume detecting means, the traffic flow presuming means including a neural network for determining relationships between traffic volumes and traffic flows, and a backup neural network which periodically determines relationships between traffic volumes and traffic flows;
• a presumption function constructing means for constructing and correcting the neural network of said traffic flow presuming means, wherein said presumption function construction means contains a plurality of relationships between traffic flow patterns and traffic volumes, and constructs said neural network by using arbitrarily selected plural relationships among said relationships, and further corrects said neural network by using newly selected relationships between traffic flow patterns and traffic volumes on the basis of traffic flows presumed from actually measured traffic volumes and controlled results, and said presumption function constructing means compares and evaluates results of said neural network and results of said backup neural network and replaces the contents of said neural network with the contents of said backup neural network when the results of said backup neural network are superior to the results of said neural network;
• and
• a control parameter setting means for setting control parameters for controlling said transportation system on the basis of the traffic flow determined by the traffic flow presuming means.

A method for controlling traffic in a transportation system comprising the steps of:

• a) detecting traffic volume in a transportation system;
• b) estimating traffic flow from the traffic volume using a presumption function;
• c) constructing and correcting the presumption function based on known traffic flow and traffic volume relationships;
• d) setting control parameters for controlling the transportation system based upon the estimated traffic flow;
• e) detecting control results and drive results of the transportation system;
• and
• f) updating the control parameters and the presumption function based upon the control results and the drive results.