Abstract: Ensuring and strengthening the reliability，security and stability of computer systems are advanced requirement to system service.Based on autonomic computing，a self-tuning model on system dependability is proposed.By dynamically analyzing the change trend of object system dependability，and making reference to the grads fluctuation of accient dependability curve，it implements the on-line dependability evaluation，dependability dynamic prediction and self-tuning scheme selection in turn，which accomplishes the self-tuning function of system dependability.Confidence field grads calculation method is adopted to solve the optimization problem of self-tuning.

Key words: autonomic computing, self-tuning, dependability, confidence field

摘要： 保障和增强系统的可靠性、安全性和稳定性等可信性质是对系统服务的高级要求。提出了一个基于自律计算理念的系统可信性自调节模型，该模型动态分析目标系统的可信度变化趋势，参考先验可信度变化曲线的梯度升降，依序实现可信度在线评估、可信度动态预测和自调节方案选择，完成系统可信度非降的自主调节过程，采用信赖域梯度计算策略解决自调节的最优化问题。

关键词: 自律计算, 自调节, 可信性, 信赖域