Feasibility study on measuring the chainsaw work and automated information extraction from sensor data.

Abstract

The 9-axis sensor of a smartphone measured the pseudo motion of chainsaw work to standardize the quantitative analysis of its work and extract information via clustering. A hidden Markov model could achieve high clustering accuracy between the felling and not-felling motion in engine-off and engine-on situations. New features (e.g., jerk, three-axis synthetic gyro acceleration, and the choice of the proper noise reduction parameter by moving average) were revealed. These features and parameters contribute to the robustness and high accuracy of clustering, revealing the hidden Markov model probabilities of automated information extraction from big raw data.