论文_传感器的点01

数据的来源：

一般都是实测的：

Some data from sensors are used in real time for maneuvering and related actions and some sensors are placed in sensitive area like propeller blade to collect the data for future purpose such as system diagnosis.

数据的问题:

Considering the raw data may contains noisy, discontinuous and redundant information,

how to eliminate the noisy?

场景:

room. Because of external noise like cables and the coupling of electric and magnetic fields, the measured temperature is full of spikes.

Solution01

A natural way to eliminate them is to use median filtering technique,After three iterations of median filtering, the spikes are successfully removed and the series of temperature data becomes smooth,

Solution02

Another necessary data cleaning process is resampling. Because both the data modules in Table I with sampling frequency of 1.65Hz and 1Hz respectively are involved for ship motion prediction,

how to deal with discontinuity?

场景

In addition, raw data may contain jumping phenomenon due to data definition. The bottom panel of Fig. 2 is an example for heading recording. Because the definition of heading is within [0∘, 360∘], it is inevitable to occur jumping phenomenon when the heading changes near the border.

Solution01

To remove this type of discontinuity, the heading angle is defined unbounded and a variable for the number of laps is introduced:

参考文献01