Abstract: The operating conditions can be large-scale varying in terms of the Organic rankine cycle （ORC）system for waste heat recovery, therefore it is necessary to achieve its nonlinear control in the case of large-scale fluctuations in operating conditions. We propose a gain scheduling control strategy to control the superheat of the ORC system. We establish the dynamic mechanism model of the ORC system using the moving boundary method, simulate the dynamic characteristics of the ORC, and then construct the input-output datasets within the full operating range. We use the first order plus dead time (FOPDT) transfer function to characterize the local dynamics, therefore we can identify the local linear model under different operating conditions. In addition, in order to realize the dynamic mapping of the working conditions, we propose an ARMA dynamic mapping regression method based on XGBoost. The experiments of the control effect under the wide range of operating conditions show that the proposed data-driven operating condition mapping gain scheduling strategy can acquire better control performance than the traditional PI control scheme.