The application discloses a hierarchical
system energy consumption prediction method and
system based on
machine learning, and particularly relates to the field of
processing energy consumption prediction, and is used for solving the problem of
powder processing production line grading
energy consumption prediction deviation, and is through constructing a multi-source
time sequence account book to draw the context from
drying to grading into a traceable line, to make the origin and destination of energy consumption fluctuation clear, and to combine prediction to no longer stop at static experience, but to follow the propagation path and screen surface work to depict the real load, to obtain the energy consumption
impact degree through the penetration
time difference density and the accumulated release work amount, and the operation side to generate the cost
impact label and the time slot priority according to the energy consumption
impact degree, the order energy consumption layering close to the field
rhythm, the prediction and accounting more consistent, the online checking to revise the alignment deviation with small steps, the short-time fluctuation convergence, the long-period cost curve tends to be smooth, and at the same time, the clear evidence chain is retained, and the review is based on evidence, and under the alternation of batches and beats, the stable response is still maintained, and the settlement deviation caused by misjudgment is reduced.