This paper proposes a collaborative framework integrating blockchain encryption and decision tree optimization for the problems of data fragmentation, high security risk and low classification efficiency in unmanned equipment environment sensing system. First, a blockchain data security system based on homomorphic encryption is constructed to prevent data monopoly through distributed bookkeeping and key management; second, a decision tree model applicable to blockchain nodes is designed, and a buffer null node mechanism is introduced to solve the problem of local fragmentation and enhance the robustness of classification. Further, improve the PBFT consensus mechanism, use the C4.5 decision tree to realize node trust evaluation and differential voting weight allocation, and set the consensus threshold to 50% of the total weight. Finally, the environment-aware capability requirement is refined based on DoDAF multi-view. Experiments show that the data throughput of the optimized decision tree reaches 288.29 × 10⁵ byte at an execution time of 100s on the Nursery and Mushroom datasets, which is a 50.6% improvement over the optimal comparison algorithm. The number of threads is 0.979 at a data parallelism of 10, with a 32% reduction in volatility. Instruction data encryption security averages 98.56%, an improvement of 4.71 to 7.43% over DES and RSA. The encryption takes only 398.3ms (2.0GB data), which is 2.8 times more efficient than RSA.
