Abstract:
Reducing natural resource consumption is essential in moving towards a sustainable society. One such application field is the mining sector, where electric locomotives haul tons of ore for many kilometres underground. Improving the efficiency of these vehicles will result in higher machine utilisation, increased profitability and reduced overall cost of ownership of these vehicles. To optimise the energy consumption of such a vehicle, it is essential to know its energy requirements.
The main contributing factors to energy consumption are the type of vehicle, the route being travelled, the gross combined mass and the velocity profile to complete the route in the logistical constraints imposed. Regenerative braking, various powertrain efficiencies and environmental conditions also significantly affect the overall energy consumed on a route.
The broad goal of this thesis is the development of a method that is capable of providing a vehicle controller with an optimal velocity profile to travel a route by, considering the route, loads and logistics in real-time. This project has three phases: route identification, mass estimation and velocity profile optimisation. The individual aspects are handled as generically as possible to broaden the fields of application.
A novel method of route identification is developed to identify an underground route being travelled in real-time from a database without the use of external communication. The route is identified by searching for patterns in heading and altitude data by comparing these to the routes available in a database, which is a computationally efficient method for real-time applications. If the route cannot be identified, the data is stored as a new route in the database and used for route identification in future traversing. The route topography from the recorded data is now available and used throughout the rest of the study. Extensive above-ground and limited underground tests confirm the usability of the strategy.
A patented real-time mass estimation strategy is developed that uses a simple torsional load cell on the driving axle of the vehicle. This method makes mass estimation simpler as it doesn't require multiple load cells and doesn't need trailers to be instrumented either. The driving torque, route topography and vehicle dynamics equations are applied to Newton's second law of motion for the vehicle, with the equations rearranged to solve for the unknown mass of the vehicle rig. Test results show that the vehicle's gross combined mass can be estimated to be within 5% of the actual value.
The only controllable variable in energy consumption for a specific vehicle on a specific route is the vehicle's velocity profile. A strategy is developed that can robustly optimise the velocity profile for the vehicle on a known route with a known mass such that the overall energy consumed is reduced, while still adhering to the logistical time requirements. The results show an energy reduction from 8.5 MJ to 5.7 MJ (33%) is obtained over a real-world velocity profile for a 10 km test route. The key to the optimiser's robust operation lies in how the initial guess to the local solver, fmincon, is executed using strategic inverting, scaling and shifting of the route's topographic profile in a low fidelity, quickly executable optimisation.
The initial guess strategy proved successful in yielding estimates of the optimal velocity profile close to the complete optimisation solution's values, typically within 10%, but at a fraction of the computational time, typically being around 20 seconds as compared to 3 hours for the full optimisation. The low-fidelity model facilitates an easily implemented lookup table to retrieve the scale and shift parameter values, thus estimating the optimal velocity profile within a fraction of a millisecond, rather than hours. This fast method facilitates real-time optimisation of a vehicle's velocity profile if the route is known.
This thesis set out to develop a system capable of reducing the energy consumption of a vehicle and proved successful in performing this optimisation in real-time due to the advancements made in real-time route identification, real-time mass estimation and real-time optimisation of the velocity profile for the vehicle.
