As the core of intelligent power system construction，the rapid popularization of power-related chips has made its energy consumption a part of the energy consumption of power systems. For solving energy consumption issues for power terminal multi-core chip，firstly，based on the original task scheduling technology，this paper proposes an intra-task scheduling method that considers the probability distribution of task execution time； and uses mixed integer linear programming （MILP） to model the problem to obtain mathematical methods，to find the best scheduling solution. Finally，this paper verifies this method by establishing an experimental verification platform. The results show that the proposed method can save more than 30% energy consumption expectation than the traditional scheduling method.