Refinement of Automated Dragline Swing Control Algorithms

The control of the swing cycle of a dragline a key component of the dragline automation roadmap presented in the ACARP Open Cut Automation Scoping Study (C11054). The CSIRO team in partnership with ACARP and mining companies have been progressively developing a Dragline Swing Assist (DSA) system through projects, C3007, C5003, C9028, C10039.

In November 2002, a very successful field trial of the DSA system (C9028) was conducted on the Bucyrus 1350 dragline at Callide Coalfields, Boundary Hill Pit. The trial demonstrated a system that can out perform a skilled operator for simple two point paths but was slower that a skilled operator for more complex paths. A preliminary analysis of the field trial data identified two key reasons why the performance of the DSA system did not exceed the performance of an operator over the more complex paths:

• Non-optimised bucket path planning.
• Large safety margins for bucket hoisting.

This project addresses the first point through investigation and demonstration of improved swing control algorithms that allow the DSA system to match and potentially exceed the cycle time performance of a skilled dragline operator for a range of operating conditions. The key steps in achieving these objectives were to:

1. Analyse C9028 production trial data in relation to the development of new algorithms.
2. Develop alternative/improved swing control algorithms.
3. Commission the C9028 DSA System Swing Control on the Leslie Consulting model dragline.
4. Implementation and testing of optimised swing control algorithms on model dragline.
5. Industry open day.

The following list details the key achievements and outcomes from addressing the research objectives described above:

1. A detailed re-analysis of the productivity trial data collected in C9028 showed that:

• The original DSA swing controller cycle times were 3-7s slower than human operators for complex multi-point bucket trajectories.
• The original DSA swing controller had a significantly greater number of hoist limited cycles (41%) compared to human operators (15%).
• The original DSA swing controller on average hoisted 5.6m higher than an operator due to imposed safety constraints.

2. An optimised bucket path planning algorithm was developed which adaptively considers perfor­mance limitations of all drives (drag, hoist and swing) and maximises swing velocity through the via-points whereby decreasing the cycle time whilst maintaining critical timing events such that the bucket passes through all scheduled trajectory points to avoid bucket collisions.

3. A dragline simulation model was developed and used to evaluated different motion planning algo­rithms. The key findings were:

• Simulation of the optimised swing algorithm on the DSA field trial data sets showed a mean cycle time savings of greater than 1.6 seconds (approximately 3% on a 60 second cycle) was achievable on hoist-limited, three trajectory point cycles.
• Simulation showed that by lowering the bench via-point by up to two bucket heights (currently generally set at 3-4 bucket heights), further cycle time savings of up to 5 seconds is possible, therefore, bringing the DSA cycle times equal to or better than that of an operator.

4. The 1/7 scale-model dragline experimental platform operated by Leslie Consulting (LC) was made capable of autonomous control. The performance of the model dragline was made identical to the Callide Coalfields BE1350 dragline used for the productivity trial at the control software level.

5. An extensive experimental program was conducted which showed that:

• Measured cycle time reductions of approximately 2.3 seconds for the complete dig cycles over that of the original DSA motion planner which supported the simulation study.
• A preliminary experimental evaluation reduced via-point hoist conservatism showed cycle time reductions up to approximately 8 seconds recorded. These observed improved cycle times close the gap between operator and DSA performance to the extent that on the model dragline they would be equivalent.
• High-precision and repeatable dumping is possible with both the original and improved DSA systems.

The control and motion planning systems for the DSA system developed on the model dragline during this project, were made the same format as those implemented on the production dragline. Therefore, all algorithms/techniques developed on the model dragline can be seamlessly moved to a production dragline of both conventional or UDD configurations.

The experimental demonstration of the significant cycle time improvement that can be achieved through reducing hoist safety margin highlights the next logical step of incorporating the improved DSA system of this project, with the Digital Terrain Mapping algorithms of ACARP C10039 into the dragline automation. This would allow optimum bucket trajectories with obstacle avoidance to be determined taking into account terrain profile and growing spoil piles.