Random forest-based quality fluctuation stability analysis for early abnormal warning in spinning process

Aiming at the problem that it is difficult to forewarn the abnormal fluctuation of spinning process quality, a random forest-based quality fluctuation abnormal warning mechanism for spinning process stability is proposed. Firstly, the time series set of yarn quality process parameters is decomposed by the window, and the abnormal fluctuation rate is defined to characterize the fluctuation degree of the window sequence. Then the window according to whether there is abnormal fluctuation is marked, and random forest model is built to identify the window fluctuation pattern, so as to build an abnormal early warning strategy for spinning process quality fluctuation. Finally, a case study is conducted through a spinning process data set to evaluate the performance of proposed model. Results show that the proposed method can provide early warning of abnormal fluctuation in spinning quality, which provides guarantee for quality stability in spinning production process.