The invention discloses a load prediction method in coordinate control system of a supercritical coal fired unit. The method comprises following steps of according to a mechanical model of IEEE standard steam turbines and boilers, deriving a two-input two-output unit model; using the genetic algorithm to identify parameters of the unit model during operation with different coal properties; providing a constrained dynamic matrix control algorithm based on a variable time domain idea to solve the load of the unit and the main steam pressure so as to obtain the optimal solution to allow related operation parameters of the unit during change of the load to be able to meet requirements of safe production; and adding a supervision link, establishing a framework with an upper layer dynamic control layer and a lower layer dynamic control layer to solve the balance performance, and switching to the original PID loop to ensure the safety and reliability of the system when a primary frequency modulation action, a communication fault or a prediction output abnormal value occurs. According to the invention, the output load of the supercritical coal fired unit can be precisely predicted and operation parameters like the main steam pressure and the main steam temperature can be controlled; and stability and economical performance of the operation of the supercritical coal fired unit are improved.