The invention relates to the technical field of grain sorting and cleaning, in particular to a multi-
modal sensor fusion-based
grain quality dynamic grading and
stone removal cooperative control method, which comprises the following steps of: 1, synchronously collecting by a multi-
modal sensor, and generating a synchronous pulse through a high-speed photoelectric trigger when grains fall, a
linear array camera is triggered to collect grain surface morphology sequence images, a near-
infrared spectrometer collects an
absorbance curve and
moisture content data, and a piezoelectric weighing platform collects
impact pressure waveforms; step 2, dynamic grading
decision making; step 3, generation of collaborative parameters; 4, executing and feedback optimization, and issuing the generated parameters to a vibrating screen, a fan and an inclination angle adjusting mechanism to execute control; and detecting the proportion of residual impurities at a
stone removal outlet, and triggering the
cascade decision-making model to retrain when the proportion exceeds the limit continuously. Through the design, the grain grading and stone removing efficiency and accuracy are remarkably improved, and high-quality product experience and value-added benefits are provided for users.