Aiming at the problem that massive data in distribution Internet of Things is vulnerable to network attack,a data security protection method of distribution Internet of Things based on blockchain is proposed. Firstly,a collaborative security protection architecture is constructed,and the abnormal traffic identification mechanism is designed through adaptive traffic monitoring. Secondly,the improved practical Byzantine fault tolerance (PBFT) is used to establish a trust mechanism to share the attack detection model between the distribution cloud master station and each distribution edge agent. Then,based on the blockchain smart contract,the attack detection model is dynamically updated,and the deep reinforcement learning is used to train and fuse each attack detection model to obtain the attack detection fusion model. Finally,a simulation platform based on Mininet is built to demonstrate the proposed method. The results show that the comprehensive performance of the proposed attack detection model is better than that of centralized and distributed models.