Abstract

The increasing complexity of hybrid quantum-classical computing systems requires new methods for effective workload distribution to guarantee computational efficiency and strong security. To overcome them, this paper suggests a Hybrid Graph-Based Workflow Evolutionary Optimization Model based on SparseEA-AGDS (Evolutionary Algorithm with Adaptive Genetic Operators and a Dynamic Scoring mechanism). The working loads are represented as dependency-conscious task graphs annotated with computational costs and security properties, such as quantum-specific vulnerabilities and classical security constraints. SparseEA-AGDS allocates tasks to quantum or classical computers by managing performance-critical, security-sensitive workloads using a dominance-based multi-objective optimization algorithm. The dynamic scoring mechanism ensures that tasks requiring quantum-safe encryption or sensitive data processing are always assigned to the right resources. Experimental results also show that the proposed model can improve execution efficiency (up to 21.8 %), partitioning overhead (18.4 %), and security compliance measures (25.6 %) compared to the baseline graph-based and heuristic partitioning schemes. Also, the scalability analysis indicates that as workload size increases, optimization performance remains consistent, with a degradation of less than 6%. These findings support the idea that the proposed framework can provide scalable, reliable workload management in hybrid quantum-classical systems, making it an appropriate choice for a range of sensitive fields, including cryptography, healthcare, and financial systems, where performance and data security are of significant importance.