中文摘要：

分析了异构多处理机系统中的负载均衡和任务调度参数，讨论了异构集群任务调度模型，提出了一种负载均衡启发式优化算法（LBHOA）。LBHOA采用启发式搜索策略，每次分配一个任务时，从不完全分配的结点中选择估计值最小的结点进行扩展搜索，直到找到完全分配的目标结点，且目标结点的时间开销估计值是所有完全分配结点中最小的。实验结果表明，与算法WLCA和LTGA相比，算法LBHOA的平均应答延迟时间的开销减少了10％，任务完成时间的开销减少了15％。LBHOA降低了资源最优分配中的计算复杂度，能够满足异构集群系统中的负载平衡和优化调度的需要，使异构多处理机系统在系统资源均衡分配的同时使系统资源利用率最优。

英文摘要：

In order to share various resources proportionally and to optimize the utility of system resource in a heterogeneous multiprocessor, a Load Balance Heuristic Optimization Algorithm（LB- HOA） is proposed after the analysis of load balance and learning of task scheduling model. LBHOA uses heuristic search technique to choose the node of least estimation value and to spread search until finding a target note that has a least value of time spending in all complete distributive notes. The experimental results indicate that LBHOA can reduce 10% of average response delay and 15% of working time comparing with WLCA and LTGA and also can reduce computing complexity in resource allocating, which meets the requirements of load balance and optimization scheduling in heterogeneous multiprocessor system.