The typical 1 000 MW power unit coordinate system has several widely-known practical problems, such as large fluctuations of key parameters, significant energy consumption under load variation, poor adaptation of coal species, and so on. However, the Infit system, which combine the predictive control technology, the neural network learning technology and the self-adaptive control technique, was used in Huaneng Haimen Power plant. Therefore, the optimal control system, includingthe optimized parameter such as the main stream pressure, mid-point temperature, are achieved. In this paper, key parameters are compared, the research results can be referenced for the operation of similar 1 000 MW power units.