基于LSTM神经网络我国印染布产量预测

Prediction of China's printing and dyeing fabric production based on LSTM

采用性能较优的长短期记忆神经网络模型(LSTM)对我国印染布产量进行预测。结果显示:平均预测误差仅为0.2591%,比极限学习机的1.0181%减小76.0942%,比循环神经网的0.6020%减小56.9601%。运用LSTM模型预测2023—2025年我国印染布产量,通过分析,表明这一预测结果有较高的可信度。

LSTM neural network model introduces a gating mechanism to control the retention and forgetting of information,which has stronger memory ability and can handle longer data sequences,greatly improving fitting and prediction accuracy.Due to the irregular and complex distribution of the production data sequence of printed and dyed fabrics in China,the prediction performance of traditional recurrent neural networks is unsatisfactory.Therefore,The LSTM model was adopted to predict the production of printed and dyed fabrics in China.The results showed that the average prediction error was only 0.2591%,a decrease of 76.0942%compared to the extreme learning machine's 1.0181%,and a decrease of 56.9601%compared to the recurrent neural network's 0.6020%.The LSTM model was used to predict the production of printed and dyed fabrics in China from 2023 to 2025.The analysis shows that this prediction result has high credibility.