No abstract

In machine work, the productivity, energy efficiency, and the quality of the work depend strongly on the skills of the human operator. This paper proposes a hierarchical method for skill evaluation of human operators in machine work during their normal work. The method refines skill metrics obtained from work cycle recognition -based evaluation system proposed earlier by the authors. The proposed skill components are: machine controlling skills, control parameter tuning skills, knowledge of the work technique and strategy, and planning and decision making skills. The skill components in each task are evaluated by a dedicated fuzzy inference system, whose rule base is generated automatically. The method is utilized to evaluate skills of nine operators of a cut-to-length forest harvester.

Real-time performance assessment and condition monitoring are potential new features in mobile working machines that have to run in a wide range of operating conditions. Condition monitoring and performance assessment are needed to be able to proactively correct impending faults before severe failures or machine stoppage occur. This paper presents a data-driven approach for machine performance assessment and condition monitoring based on indices representing the performance of a subsystem. Instead of adding new sensors, the indices are computed using existing data from the machine control system. Metrics for machine performance follow-up are derived from these multidimensional data, which have strong nonlinear correlations in certain measurement variables. Although the indices describe primarily the technical performance of the machine, they have proven to be valuable also in terms of condition monitoring of various machine functions. The indices summarize in a concise and easily comprehensible manner changes in performance.

No abstract