Multivariable control of an earthmoving vehicle powertrain experimentally validated in an emulated working cycle

The operation of an earthmoving vehicle involves the coordination of a multivariable powertrain and the execution of specific tasks in a repetitive fashion. The performance and efficiency depend heavily on human expertise. The purpose of this research is to automate the coordination of a multi-input multi-output (MIMO) nonlinear electro-hydraulic powertrain and to validate the performance and efficiency improvements in a human-machine interaction. Firstly, a robust gain scheduling method is developed to design a powertrain controller and to analyze its robust stability and robust performance. The gain scheduling is based on a Local Controller Network strategy and its satisfactory properties are analytically confirmed using robust control theories. Secondly, the improvement of performance and efficiency are validated through two experiments performed on an Earthmoving Vehicle Powertrain Simulator (EVPS). This testbed is a Hardware-In-the-Loop HIL environment representative of a class of electro-hydraulic systems with multiple loads. The two human-operated experiments include a reference tracking test and a working cycle test. In the second test, three types of loads are modeled for a typical wheel loader and emulated on EVPS for a 180-degree loading cycle. These models include the steering, the drive, and the implement. The load emulation technique ensures that the HIL working cycles are representative of real life cycles. The reference tracking and loading cycle results show the significant improvement in productivity in terms of performance, efficiency, and ease of operation.