Machine Learning for Sensor Transducer Conversion Routines
Machine Learning for Sensor Transducer Conversion Routines
复制标题
DOI：
10.1109/les.2021.3129892
复制
发表时间：
2021-08
影响因子：
1.6
通讯作者：
T. Newton;James Timothy Meech;Phillip Stanley-Marbell
中科院分区：
文献类型：
--
作者：
T. Newton;James Timothy Meech;Phillip Stanley-Marbell
具有数字输出的传感器需要软件转换例程将无单位模数转换器样本转换为具有正确单位的物理量。由于低功耗嵌入式系统的计算资源有限，这些转换例程在计算上是复杂的。这封信介绍了一组机器学习方法，用于学习新的、不太复杂的转换例程，这些例程不会牺牲BME 680环境传感器的精度。我们提出了一个帕累托分析的模型和模型，减少现有的行业标准的温度，压力和湿度的转换例程的计算开销分别为62%，71%和18%，准确性和计算开销之间的权衡。相应的RMS误差为0.0114 °C、0.0280 KPa和0.0337%。这些结果表明，用于学习转换例程的机器学习方法可以产生计算开销减少的转换例程，同时保持良好的准确性。
Sensors with digital outputs require software conversion routines to transform the unitless analog-to-digital converter samples to physical quantities with correct units. These conversion routines are computationally complex given the limited computational resources of low-power embedded systems. This letter presents a set of machine learning methods to learn new, less-complex conversion routines that do not sacrifice accuracy for the BME680 environmental sensor. We present a Pareto analysis of the tradeoff between accuracy and computational overhead for the models and models that reduce the computational overhead of the existing industry-standard conversion routines for temperature, pressure, and humidity by 62%, 71%, and 18%, respectively. The corresponding RMS errors are 0.0114 °C, 0.0280 KPa, and 0.0337%. These results show that machine learning methods for learning conversion routines can produce conversion routines with a reduced computational overhead which maintain good accuracy.