In order to solve the problem of automatic control of spindle rotation speed of straw picking compactor and facilitate the full mechanization of the compactor, the mathematical model and rotation speed prediction model of the electro-hydraulic control system were designed. The spindle rotation speed control method based on GPC-ILC for straw compactor was proposed. By collecting input and output data during the previous operation of the straw compactor, using the least square method with forgetting factors to identify the parameter model of generalized predictive control and calculate the predictive output value. By correcting the predictive output value based on historical model errors, and deriving an iterative learning control law, new control quantities were calculated online and real-time to achieve spindle speed control. Under the proposed algorithm, a field harvest experiment was conducted. When increasing the load, the maximum dynamic deviation of the rotational speed was 3.21r/min, the deviation from the target value was 2.6%, and the maximum margin was 1.23r/min. During load shedding, the maximum dynamic deviation was 2.23r/min, the deviation from the target value was 2.47%, and the maximum margin was 0.89r/min. The time for increasing or decreasing the load speed to reach stability was less than 5s, and the overshoot was less than 3%. Field experiments showed that the maximum dynamic deviation was 3.75r/min, the deviation from the target value was 3.47%, and the maximum margin was 1.79r/min, which met the needs of field operation of the molding machine. GPC-ILC algorithm can timely correct the uncertainty of rotational speed control caused by model mismatch and interference.