Cell Status Diagnosis for the Aluminum Production on BP Neural Network with Genetic Algorithm - Advanced Research on Computer Education, Simulation and Modeling - page 151

Figure 4 is the components of the system. We use the 30 sets of data acquired on site to
test the well-trained BP neural network. As a result, 25 sets of them could diagnose the
status correctly. That is to say the accuracy rate could reach 83. The accuracy of this
cell diagnosis model is verified.
Fig. 4. System function module chart
5 Conclusion
1. According to the characteristics of the aluminum production cells status and the
production system at present, we accomplish an overall design for the system.
2. Neural network is used to establish the cell status diagnosis model. Also genetic
algorithm is applied to optimize the initial weights and threshold value of the BP neural
network.
3. After testing this system with the actual production data. We can find that the
diagnosis result is consistent with the actual situation.
Acknowledgment. Project supported by the National Natural Science Funds (Based on
the Six Sigma and Trend of Sequential Pattern of Aluminum Cell Condition
Forecasting Model and Algorithm for Research Funding: 51075423).
References
1. Liu, Y., Li, J.: Modern Aluminum Production. The Metallurgical Industry Press, Beijing
(2008)
2. Zeng, S., Li, J.: Model Predictive Control of Superheat for Prebake Aluminum Production
Cells. Light Metals (2008), TMS Annual Meeting (March 2008)