To address the challenges of high cross-shard transaction ratios and delays in cross-shard transaction verification encountered in optimizing blockchain performance through sharding technology, this paper proposes a blockchain layered sharding algorithm. Based on the long-term accumulated transaction data in the blockchain, this method predicts and optimizes the shard distribution of transactions by analyzing the transaction characteristics between accounts, thereby reducing the number of transactions that need to be conducted between different shards. A blockchain transaction sharding algorithm (OSABTC) was designed to optimize conventional shards (i-shard) and bridge shards (b-shard) in layered sharding through transaction characteristics. By reducing the cross-shard transaction ratio and increasing the proportion of bridge shard transactions, the problem of high cross-shard transaction ratios in transaction sharding was effectively resolved. Experimental comparisons show that, compared with the Random Sharding Algorithm (RSA), the OSABTC algorithm can reduced the cross-shard transaction ratio and increased the bridge shard transaction ratio, while transaction latency was reduced. These results demonstrate the effectiveness of the OSABTC algorithm in significantly enhancing the performance and efficiency of blockchain systems.