【摘要】 在电解铝生产过程中，传统的人工控制决策方式已经难以适应现代铝电解生产要求，当下深度学习算法处理此类时间序列数据得到广泛应用，决策是否高效影响铝电解槽的稳定运行和高效产出铝。同时，数据隐私问题也不容忽视，隐私安全既影响电解铝生产过程又影响其正常使用，误用、滥用数据挖掘可能导致用户数据特别是敏感信息的泄露，而信息一旦丢失或泄漏将造成重大的损失。针对以上问题，提出一种利用改进LSTM模型结构结合优化ElGamal算法的电解铝决策方法：首先针对数据隐私问题提出了优化后的ElGamal算法；再针对电解铝数据特性改进LSTM模型结构与优化ElGamal算法的双结合。实验结果表明，本方法可以在保证决策隐私安全的情况下，性能优于传统方法，在实际情况中有参考的价值。更多还原

【Abstract】 In the process of electrolytic aluminum production, the traditional manual control decision-making method has been difficult to adapt to the requirements of modern aluminum electrolysis production.Deep learning algorithms have been widely used to process such time series data, and the efficiency of decision-making affects the stable operation of the aluminum electrolytic cell and the efficient output of aluminum. At the same time, data privacy issues can not be ignored.Privacy security not only affects the production process of electrolytic aluminum, but also affects its normal use.Misuse and abuse of data mining may lead to the leakage of user data, especially sensitive information.Once the information is lost or leaked, it will cause significant losses.In order to solve the above problems, this paper proposes an electrolytic aluminum decision-making method based on the improved LSTM model structure combined with the optimized ElGamal algorithm. Firstly, the optimized ElGamal algorithm is proposed to solve the problem of data privacy. Then according to the characteristics of electrolytic aluminum data, the LSTM model structure is improved and the ElGamal algorithm is optimized.Experimental results show that the performance of this method is better than that of traditional methods under the condition of ensuring the privacy and security of decision-making.It has reference value in actual situations.更多还原