Detecting and Diagnosing Process Nonlinearity- Induced Unit-Wide Oscillations Based on an Optimized Multivariate Variational Mode Decomposition Method

In process control system, nonlinearity-induced unit-wide oscillations are a common fault, which degrades the control performance and threaten the stability. It is important to detect and diagnose the nonlinearity-induced unit-wide oscillations to improve the process control performance. In this paper, a novel method, termed as SSA-MVMD, is proposed by combining the sparrow search algorithm (SSA) and multivariate variational mode decomposition (MVMD) to detect and diagnose the nonlinearity-induced unit-wide oscillations. MVMD is an advanced signal decomposition and time-frequency method. However, its performance is affected by the mode number $K$ and penalty coefficient $\alpha $ . SSA is adopted to optimize the parameters of MVMD. Then, a novel SSA-MVMD-based detector is presented to detect and diagnose the nonlinearity-induced unit-wide oscillations. The proposed method is model-free and data-driven thus requiring no prior knowledge about the process dynamics. Compared with the latest related works, the proposed method can better decompose the multivariate nonstationary signals and adaptively analyze the unit-wide oscillations. In the end, the effectiveness and advantages are demonstrated by simulations as well as industrial cases.