The invention relates to a robust iterative learning model predictive control method which can be applied to control of an intermittent stirred tank reactor and solves the problem of high-precision tracking of reaction temperature of the intermittent stirred tank reactor under a variable reference track. The method mainly comprises the following steps: constructing a linear parameter varying (LPV)prediction model according to a nonlinear mechanism model of the intermittent stirred tank reactor; absorbing the control experience of the past batch through iterative learning control, and improving the tracking precision; processing real-time interference effectively by adopting model prediction control, and ensuring the time domain stability and robustness of the system; regarding the changeof the reference track among batches of the intermittent stirred tank reactor as the bounded interference of the iterative shaft, and constructing the H infinite constraint to ensure the control effect under the variable reference track. Compared with the traditional iterative learning model predictive control method, the robust iterative learning model predictive control method has higher adaptability and better tracking performance under the variable reference track of the intermittent stirred tank reactor.