Abstract: The traditional databases are inefficient for improving the performance of emulate and real-time databases.A feedback control approach with imprecise computation techniques addresses this problem.In this paper，an improved approach is presented.By setting an appropriate forgetting factor and introducing the average value of the maximum data error，the approach aims to reduce the sudden changes so that enhance the stabilization of the systems and improve the performance of the real-time systems under a varying system workload.

Key words: real-time database, Quality of Service（QoS）, imprecise computation, feedback control

摘要： 传统的数据库已经不能满足仿真数据库对性能优化的要求。因此，提出了非精确计算中由反馈环节控制的实时调度算法。在这些研究的基础上对该方法提出了改进，提出了使其在各种负载条件下，通过在反馈控制中设定遗忘因子以及引入最大限度数据错误的平均值来减少系统的突变，增强了系统的稳定性，提高了实时系统服务质量。

关键词: 实时数据库, 服务质量, 不精确计算, 反馈控制