The invention provides a multi-rate hierarchical learning control method for a dense medium
coal preparation process under unreliable communication. The method comprises the following steps: step 1, carrying out
system screening and
processing on historical operation data; 2, a basic loop controller is designed through
model predictive control; step 3, constructing a heavy medium
coal preparation whole process generalized controlled object containing two
layers of dynamic characteristics; 4, performing optimization solution on the set value
gain of the dense medium suspension
liquid density control loop; and 5, updating the weight parameter of the
performance index of the operation layer by using the operation data generated by the controlled
system online, substituting the updated weight parameter into the step 4 again, and circularly executing the step 4 and the step 5 until a set value meeting an optimization target is obtained. According to the method, a data-driven
performance index weight parameter setting and dense
medium density optimal set value optimization solving method is designed, so that the optimal dense medium separation effect is obtained, and the method has very important significance for improving the
economic benefits of a
coal preparation
plant.