改进自适应学习树电源管理预测策略

摘要

To reduce the power consumption of device, dynamic power management prediction strategies based on the structure of adaptive learning tree is researched Through added idle-time length node in the structure model of Probability-based adaptive learning tree, an improved Prediction Strategies on the strength of Probability statistics weighted idle-time is proposed, the idle-time length is used as projections and based on the probability statistics of actual historical state, the idle-time predicted result is updated. Results of the simulation indicated the proposed method can effectively lower power consumption and improve the prediction accuracy.%为了降低嵌入式设备的功耗,研究了基于自适应学习树结构模型的动态电源管理预测策略.通过在基于概率自适应学习树结构模型的基础上添加空闲时间长度结点,提出了概率统计加权空闲时间的改进自适应学习树电源管理预测策略,以空闲时间长度作为预测依据,同时采用实际状态历史概率统计的结果进行预测空闲时间长度的更新.仿真结果表明,该方法可以有效地降低设备功耗,并且提高了预测准确率.