Blockchain systems face major load balancing challenges in multi-modal data scenarios. To address resource imbalance and performance degradation, this paper proposes a Multi-Modal Priority Quantification (MMPQ) algorithm that dynamically assigns priority based on transaction size, complexity, and queuing time, enhancing allocation fairness. A Dual-Mode Adaptive Packaging (DMAP) algorithm is also introduced, switching between throughput-maximization and latency-sensitive modes based on real-time workload to balance efficiency and responsiveness. Additionally, a timeout compensation rule with exponential priority escalation mitigates starvation and boosts robustness. Experiments show latency reductions of 38.2% (CPU-bound), 33.7% (I/O-bound), and 20.7% (mixed workloads) over FIFO. Overall, the scheme significantly improves blockchain execution efficiency and fairness.