Browse Prior Art Database

A TECHNIQUE FOR AUTOMATED IDENTIFICATION OF TRAIN CAR MODEL AND COMPUTATION OF TRAIN CAR SPEED IN ORE LOADING PROCESS

IP.com Disclosure Number: IPCOM000249056D
Publication Date: 2017-Jan-31
Document File: 8 page(s) / 258K

Publishing Venue

The IP.com Prior Art Database

Abstract

A technique for automated identification of train car model and computation of train car speed in the ore loading process is proposed. The technique benefits from the predictable rail track motion flow to enable the fast detection of train cars models and the computation of train cars speed. The technique includes steps of modeling, camera calibration, edge detection, train car model identification and computation of speed of the train car. Following components are considered for modeling: a train track, a train running over the train track, and a patch with known distance from the ground and from the track. A chessboard patch is used for camera calibration and as a reference for the world coordinate system. An important consideration of the proposed technique is that different train car models, with different sizes, have observable features at different but known spots in an image. By analyzing these spots over time, the technique detects presence of a train car, identifies the model and computes speed of the train in a fast and robust way.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 23% of the total text.

A TECHNIQUE FOR AUTOMATED IDENTIFICATION OF TRAIN CAR MODEL AND COMPUTATION OF TRAIN CAR SPEED IN ORE LOADING PROCESS

BACKGROUND

 

The present disclosure relates generally to iron ore loading process and more particularly to a technique for automated identification of train car model and computation of train car speed in the ore loading process.

Generally, the process of ore loading on ore load train terminals is time demanding and prone to error. The ore loading process requires a human operator to manually control the loading of ore in train cars passing by at slow speed. Both train car speed and train car model are key to optimizing the loading process. The train car model is important because amount of material that can be transported as well as how the material should be distributed within the train car depends on the model of the train car.

Nowadays, operators are required to pre-define a sequence of train cars that are used for ore loading in order to configure the loading supervisor system to execute the loading process. Operators are also required to communicate with the locomotive pilot to control the train car speed. Besides demanding additional efforts from the operator, such conventional method introduces risks in the loading process with respect to inconsistency between composition of material planned to be loaded and the actual composition, and communication failure between the operator and the locomotive pilot.  When such situations occur, the operator has to rely on her/his experience and intuition to solve them and minimize their impact on the loading process. Some hardware solutions, such as optical sensors are available to identify train car models and their speed, but they are expensive, thereby hindering their wide use.

One conventional technique uses a camera to determine train track characteristics, such as curvature and slope. Another conventional technique assesses in real time railway disaster vulnerability using video computational analytics based on neural networks. Yet another conventional technique identifies vehicles by feature detection and tracks them using a single video camera. However, none of the conventional techniques optimize iron ore loading process by automated train car model identification and computation of train car speed. Also, most algorithms for visual tracking depend on the motion pattern and the environment observed in the scene and loss of information caused by the projection of the three dimensional (3D) world on a two dimensional (2D) image, noise in images, complex object motion, occlusions, illumination changes, and other scene specific problems render such algorithms unsuitable for use in ore loading process.

It would be desirable to have an improved technique for automated identification of train car model and computation of train car speed in the ore loading process.

BRIEF DESCRIPTION OF DRAWINGS

Figure 1 depicts a model used by the proposed technique which includes a train track,...