【摘要】 为消除甲烷逃逸浓度软测量过程中测试参数之间延迟对软测量实时性和精度的影响，提出一种基于长短时记忆(long short-term memory, LSTM)神经网络和互信息(mutual information, MI)的船用双燃料低速机甲烷逃逸浓度软测量方法。首先，利用互信息进行辅助变量筛选和变量时间延迟的计算；将预处理后的数据导入LSTM模型来预测甲烷逃逸浓度；最后使用某低速双燃料机甲烷逃逸治理系统的历史数据进行模型性能的验证。结果表明：基于LSTM和互信息的软测量模型具有较好的预测能力，为船舶双燃料低速机甲烷逃逸浓度的监测提供了一种有效参考方法。更多还原

【Abstract】 In order to eliminate the impact of the delay between measured data on the real-time and soft measurement of methane slip concentration monitoring process for marine dual-fuel low-speed engines, a soft measurement method of methane slip concentrations for dual-fuel low-speed marine engines based on long shortterm memory(LSTM) neural network and mutual information(MI) was proposed. Auxiliary variable selection and variable time delay calculation were performed using mutual information. The preprocessed data was imported into the LSTM model to predict the methane slip concentration. Finally, the historical data from a methane slip processing system of a dual-fuel low-speed engine was used to test the performance of the proposed model. The experimental results show that the model based on LSTM and mutual information has good predictive ability. The method is proved to be effective by experiments, which can provide an effective method for monitoring the methane slip concentrations for dual-fuel low-speed marine engines.更多还原