Data-driven nonlinear frequency response method for control-oriented design of forced periodic operation and operating conditions

The optimal design of forced periodic operation and associated operating conditions has received notable attention due to the potential to improve conversion rates through the intrinsic nonlinear dynamics of chemical reactions. However, the limited capacity of analyzing higher-order nonlinearities, especially under varying operating conditions, poses significant challenges. To tackle these difficulties, the Nonlinear Frequency Response (NFR) method and computer-aided approaches have been proposed and applied to different reaction processes. Nevertheless, due to reliance on the first principle model, current methods are difficult to be literally applied in practice to perform the designs. To address this limitation, the present study proposes an integrated data-driven design approach that can perform effective quantitative analysis and integrated design of both forced periodic operation parameters and operating conditions for the effective control and design of the underlying chemical reaction process. From the input–output data obtained from either simulations or experimental measurements, the proposed method builds a data-driven control-oriented model that can, for the first time, be used for both quantitative analysis and design of forced periodically operated reaction processes. This resolves the challenges with applying the existing NFR-based methods and provides an effective solution practically applicable to a wide range of complex chemical reaction processes. Simulation studies on a non-isothermal continuous stirred tank reactor (CSTR) and an experimental investigation on a small-scale CO2 absorption device have verified the effectiveness of the proposed method and demonstrated its potential in real-world applications.