Abstract: Aiming at the dominant load distribution problem of early large heat exchange units caused by greedy characteristics when heuristic algorithm is used to optimize the heat exchange network, that is, the "preconceived" phenomenon, which will lead to the difficulty of structural variation and make the search fall into local optimization. In view of this, a penalty incentive bi-directional optimization strategy is proposed to optimize the heat exchanger network. Firstly, the block penalty is applied to the large heat exchange unit in the structure, which can delay the growth rate of the large heat exchange unit and promote the succession of the dominant heat exchange unit in the structure at the later stage of optimization; At the same time, in order to increase the competitiveness of the new unit, an evolutionary penalty is imposed on the existing heat exchange unit on the cold stream or hot stream, which is used to promote the development of the new unit and realize the renewal of the dominant heat exchange unit on the stream. Through the verification of two typical heat exchange network examples, the results show that the method can effectively avoid the local optimization and obtain a better heat exchange network structure, which proves its effectiveness in improving the global search ability of the heuristic algorithm.