Recently, I am busy with the research on distributioned sensor data mining. The number of sensor is about 60, and the same type. I need to collect all sensor values at the same time when robot moving at certian speed at time =1s, 2s, 3s. The data will be used to find pattern inside the data. All data will be Xi. and the velocity will be Yi category. There are about 8 categories. Whether the data can be collected at the same time is very important to the validity of the later analysis. So, could anyone give me some advice on this? Thank you.
This is a challenge in sensor monitoring systems. Depending on the channels of ADC in a CPU, you may use several CPU to construct an array, they can receive a broadcast command to sample data, but there is still difference between different CPUs, even the data of sensors that are collected by one CPU exist difference. You may use some strategy to compensate the difference. Also, you need to consider the problem to be solved, how high the frequency the problem is required? Hope it is helpful.
@Hongmei He
Thank you very much for answering my question and giving you guidance. Differences of received data can not be avoided, but generally what strategies can be used to solve this problem? Because the data are related to the movement of the vehicle, especially with the speed of the carrier. If the measured data can not be synchronized there will be greater data differences which can not measure the situation I want.
Best regards.
Nailong,
You could normalize your data to give the features an equal weigh.
HTH.
Samer
@Samer
Thank you for your advice. I will think over it and try the method you mentioned.
You can use several microcontrollers (for example STM32 has 16 analog inputs), and
write the data into microcontroller's memory together with the timestamp. Next interpolate.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining