Models for predictive control of electric mining shovels

Wallis, Douglas (2007). Models for predictive control of electric mining shovels MPhil Thesis, School of Engineering, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
n40892395_mphil_abstract.pdf n40892395_mphil_abstract.pdf application/pdf 35.15KB 0
n40892395_mphil_content.pdf n40892395_mphil_content.pdf application/pdf 5.63MB 0
n40892395_mphil_front.pdf n40892395_mphil_front.pdf application/pdf 150.03KB 0
n40892395_mphil_totalthesis.pdf n40892395_mphil_totalthesis.pdf application/pdf 5.72MB 0
Author Wallis, Douglas
Thesis Title Models for predictive control of electric mining shovels
School, Centre or Institute School of Engineering
Institution The University of Queensland
Publication date 2007-06-22
Thesis type MPhil Thesis
Supervisor McAree, Peter R.
Mee, David J.
Subjects 290000 Engineering and Technology
Abstract/Summary Electric mining shovels are used in open pit mining. Their efficiency of operation directly effects mine productivity and profitability. Shovel technologies such as trajectory following are a first step towards autonomous operation and have perceived cost saving benefits through improved efficiency and collision avoidance strategies. The development of such technologies requires rigorous testing which carries a high risk of lost production time and the potential for damaging machinery. The cost of lost production, and the remote location and nature of the environments in which electric mining shovels operate, make on site testing expensive and often infeasible. This thesis develops and validates a Simulink model of a P&H-class electric mining shovel. The model provides a framework for developing and testing shovel technologies. Its usefulness is shown in simulations which test model predictive control (MPC) as a control strategy for point-to-point motion control of each of the three freedoms of a P&H-class electric mining shovel. The thesis first presents a Simulink model describing the rigid body dynamics of a P&H-class mining shovel. The model is based on work presented by Wauge [1] and relates torques applied by the three actuating drives to motions of the shovel dipper. The material of this thesis differs from that of Wauge [1] in two respects; (i) the model is presented from a systems (input/output) perspective for implementation in Simulink, and (ii) explicit expressions for various terms are given. The rigid body dynamics model is extended to include models of the DC motors, power electronics and the control loops. The input to the extended shovel model is a joystick position and outputs are drive position, speed, armature current, armature voltage, and field current. The model is validated against data collected from a P&H 2100BLE mining shovel and the predicted shovel response is shown to be in close agreement with measured values. MPC uses fit-for-purpose models to predict future plant behavior, from which a set of control inputs is computed that commands the plant to a desired output. Calculating good control inputs relies heavily on the ability to accurately predict future plant behavior. This thesis develops linear state space prediction models for each of the shovel’s three freedoms. The prediction models are verified against data collected from a P&H2100 BLE mining shovel and shown to provide “accurate enough predictions” for the purpose of predicting future plant behavior. A typical MPC optimization problem is formulated that uses the linear state space prediction models to optimize a cost function and command time-optimal point-topoint motion of each of the shovels three freedoms. Simulations which use the shovel model as a proxy for a real machine show that MPC is capable of commanding timeoptimal point-to-point motion control for the swing and crowd drives. Hoist drive non-linearities cause predictions to become inaccurate and the hoist drive is shown to become unstable when controlled using MPC.

 
Citation counts: Google Scholar Search Google Scholar
Created: Thu, 28 Feb 2008, 13:01:40 EST by Noela Stallard on behalf of Library - Information Access Service