According to the characteristics of medium-term thermal-unit commitments optimization， a mathematical model considering both the traditional mode and energy-saving principle is proposed， and an improved progressive optimality algorithm（IPOA） is adopted to solve the model. The algorithm takes the initial feasible solution from heuristics search as the basis， and the progressive optimality algorithm（POA） is adopted to improve it. At the same time， the strategies for system operating capacity equalization and single station operating peak combination are adopted to move the target solution from the boundary constraints along the target isopleths into the feasible solution's interior to provide search space for iterative loop and overcome the disadvantage of POA that is apt to fall into local optimum. In the process of algorithm implementation， the local verifying tactics are adopted to simply judge the constraints of peak and valley duration of unit commitment capacity to reduce the calculation cost. Case studies with Yunnan power grid are given to verify the proposed model and algorithm.