Combine harvester operation speed control system and method based on multi-operation parameter reward

The invention provides a combined harvester operation speed control system and method based on multi-operation parameter reward. Comprising the following steps: inputting the actual operation speed, the super voxel number of harvested crops, the grass-grain ratio, the plant height, the stubble height and the cutting width information into a Gaussian process regression model of a universal gravitation search algorithm to obtain a predicted value of the feeding amount; a DQN neural network is established, the predicted value of the feeding amount is input into the DQN neural network, the impurity rate, the breakage rate, the loss rate and the rotating speed serve as reward functions to be input into the DQN neural network, and the DQN neural network outputs the theoretical operation speed; the difference value of the actual operation speed and the theoretical operation speed and the variable quantity of the difference value are input into a fuzzy neural network PID controller, and the forward speed of the combi