| dc.description.abstract | Wheel loaders, one of the heavy equipment machinery are widely used in Construction and Mining industries. They typically transport loads over shorter distances. Due to their limited capacity, they have to perform repetitive loading and unloading operations which is called a Short Loading Cycle (SLC). Considering the total energy consumed by heavy machinery, there is a lot of scope to optimize the energy consumed by a wheel loader in an SLC. To analyze the same, a comprehensive control-oriented model of a wheel loader is developed which comprises complex mechanical subsystems such as Engine, Hydraulics, Steering, and Transmission, etc. For optimal fuel consumption, an SLC can be optimized using optimal control techniques such as Dynamic Programming (DP) and numerical optimal control approaches such as Indirect Methods (IMs) or Direct Methods (DMs).
DP is an optimal control technique that guarantees global optimum and yields a global control policy over state and control space. It is preferred for lower-order systems as it is computation-ally expensive. DMs are computationally efficient for higher-order systems. However, they don’t guarantee global minimum and convergence and only provide an optimal open-loop control policy which may not be useful for stochastic systems.
In this thesis, optimization of an SLC of a WL is performed using DP and DM, and their merits and demerits are compared using different metrics namely computational effort, optimal cost, convergence, etc. A customized DP method is proposed to solve the higher-order Optimal Control Problem (OCP) of a WL to address the issue of the curse of dimensionality (COD) and its result is compared to that of DM. | |
