Difference between revisions of "Team:HUST-China/Applied Design"
| Line 213: | Line 213: | ||
<div class="row"> | <div class="row"> | ||
<div class="col-lg-12"> | <div class="col-lg-12"> | ||
| − | <h2 class="pageTitle"> | + | <h2 class="pageTitle">Applied Design</h2> |
</div> | </div> | ||
Revision as of 22:57, 17 October 2018
Applied Design
1. design
With the completion of our molecular biology work. We will transfer from small-scale power generation environment to large-scale integrated power generation environment. Thus we have designed a software for large-scale integrated power generation environment so that we can estimate the internal conditions of the system based on sensors’ data and system modeling and decide how to operate the hardware according to these internal conditions in order to improve the efficiency of power generation.
Estimating the internal conditions of the system according to the sensor data and system modeling
The current device sensors cannot get enough data from the system. If we only use these data directly (for example, the absorbance only representing the concentration of the bacteria, the concentration of the specific chromogenic substance; the pH only indicating the acidity of the system), we will not be able to fully understand the internal conditions of the system. So our software is essential and we can use this data to understand further what we didn't understand in the system before. For instance, if a product of a bacterium cannot detect its concentration which means that we cannot know the speed of the rection, however, the growth of the bacterium is determined by the speed of this reaction and the other reaction. We can detect the growth speed of the bacterium and the rate of the other reaction. Then we can calculate the speed of this reaction. The speed is thus determined based on the integral. Therefore, our software can improve our utilization of sensor data.
Operating the hardware according to the internal conditions of the system
The operation method of the main fermenter, which produces electricity, relies more on personal calculations. However, the individual's ability is always limited, and may ignore some points. The introduction of computer learning can propose a more comprehensive and efficient fermentation operation method. We use the Q-learning algorithm for reinforcement learning, learnig from modeling data and experimental data, exploring in the modeling virtual environment and experimental real environment and knowing how to operate our fermentation experiment more efficiently.
Compared with the traditional large-scale integrated power generation environment
we design this software in order to know much about this environment and try to make full use of it so that we can improve power generation efficiency. In addition, our software can also be used widely in other integrated power generation environment so that they can be used to produce electricity efficiently.
Reference
[1] Cheng, Yizong (August 1995). "Mean Shift, Mode Seeking, and Clustering". IEEE Transactions on Pattern Analysis and Machine Intelligence. IEEE. 17 (8): 790–799. doi:10.1109/34.400568.
