| Summary: | 碩士 === 國立高雄應用科技大學 === 電機工程系博碩士班 === 106 === The paper focuses on how to remove the Tilapia scales maximally and finds its best combination of parameter in the high pressure water jet machine.
The paper used uniform design to experiment and its uniform layout was U10*(10^8). It could realize the model and equation by regression analysis. The parameters were frequency of chain, frequency of water pressure and frequency of nozzle.
Each method in the uniform layout could experiment 3 times and then calculated its mean as removal rate for decreasing the error from experiment.
In the research, the paper used back-propagation neural network (BPNN) to train the 10 data from uniform layout to model the machine, and it would experiment extra 2 data as test data. It used 3, 6 and 9 nodes in the hidden layer to train the data and compare each other. It used the value R, root mean square error (RMSE) and mean absolute percentage error (MAPE) as standard to express the model was good or not.
I-Ching algorithm was a kind of evolutionary algorithm. It proposed by Chen et al. (2017). The paper used I-Ching algorithm in regression analysis and BPNN's models to find the best combination of parameter and its predicted value.
Eventually, the 2nd combination in the uniform layout had the best removal rate 85% in the 12 experiment. After modelled from regression analysis and BPNN, it found the predicted value was 100% from the best combination of parameter by using I-Ching algorithm. After the paper verified the best combination in the machine, it would find its removal rate was 96.5% and its value of MAPE was 3.585%.
|
|---|
