The problems faced by distributed intelligent operation decision under the condition of multi-domain cooperative engagement is analyzed in the paper, including the scarcity of training data samples, the decision making process is opaque and difficult to explain, and the credibility of the decision result is to be verified. The ideas of constructing digital twin battlefields, exploring the action mechanism of intelligent algorithms, perfecting the knowledge graph of intelligent decision making, and establishing the method of credibility and ability evaluation are put forward. The key technologies are studied, such as situation understanding and cognition under distributed processing, multi-objective intelligent threat assessment based on weight self-learning, and autonomous task assignment oriented game confrontation system.