适用于海洋动态观测设备嵌入式系统的低功耗管理预测模型

能耗是制约海洋设备在水下工作时长的关键因素。为进一步降低海洋动态观测设备嵌入式系统的动态功耗， 本文针对前人提出的用于预测空闲时间的指数平滑算法所存在的权重系数基于凑数法确定、算法难以适应大波动等问题， 设计了一种新的权重系数， 并构建了一种自适应的低功耗管理预测模型。当模型检测到预测的空闲时间超过工作和休眠状态切换的时间阈值后， 中央处理器关停系统的部分外设， 使系统进入低功耗休眠状态。仿真和单片机实验均表明，本文提出的低功耗管理预测模型能够有效地提高对处理器工作的空闲时间预测的准确性， 且面对突发状况具有良好的自适应性， 可显著降低嵌入式系统的功耗， 有效延长设备的工作时间。

Low-Power Management Strategy Prediction Model for the Embedded System of Marine Dynamic Observation Equipment

Energy consumption is a key factor limiting the operating hours of marine equipment. In order to reduce the dynamic power consumption of the embedded system of marine dynamic observation equipment, this paper designed a new weight coefficient for the problems of the exponential smoothing algorithm used to predict idle time proposed by the predecessor that the weight coefficient is determined based on the sum method and the algorithm is difficult to adapt to large fluctuations, and constructs an adaptive low power management prediction model. When this model detects that the predicted idle time exceeds the time threshold for the working and sleep time switch, the central processing unit shuts down some peripherals of the system and enters the low-power sleep state. Simulation and microcontroller experiments show that the low-power management strategy proposed in this paper can effectively improve the accuracy of idle time prediction of the central processing unit, and has good adaptability in the face of unexpected situations, which can significantly reduce the power consumption of embedded system and effectively extend the working time of the equipment.