［1］Altunay H C, Albayrak Z. A hybrid CNN+ LSTMbased intrusion detection system for industrial IoT networks［EBOL］.  ［20250901］. https:www.sciencedirect.comsciencearticlepiiS2215098622002312［2］Moussavou B M K, Park D J. Attentionbased 1D CNNBiLSTM hybrid model enhanced with fasttext word embedding for Korean voice phishing detection［EBOL］.  ［20250901］. https:webofscience.clarivate.cnwosalldbfullrecordWOS:001038898600001［3］Peng H, Wu C, Xiao Y，et al. A BiLSTMbased iot intrusion detection system with mutual information and focal loss［C］ Proc of the 6th Int Conf on Frontier Technologies of Information and Computer (ICFTIC). Piscataway, NJ: IEEE, 2024: 16［4］Zhao J, Liu Y, Zhang Q, et al. CNNAttBiLSTM mechanism: A DDoS attack detection method based on attention mechanism and CNNBiLSTM［J］. IEEE Access, 2023, 11(7): 136308136317［5］Sangeetha J, Kumaran U，et al. Using BiLSTM structure with cascaded attention fusion model for sentiment analysis［EBOL］.  ［20250901］. https:xueshu.baidu.comuser centerpapershow?paperid=1q1c0ve06y0p0m20783w0vs0 mc345254&site=xueshu_se［6］Bamber S S, Katkuri A V R, Sharma S, et al. A hybrid CNNLSTM approach for intelligent cyber intrusion detection system［EBOL］.  ［20250901］. https:www.sciencedirect.comsciencearticleabspiiS0167404824004516［7］Yin X, Fang W, Liu Z, et al. A novel multiscale CNN and BiLSTM arbitration dense network model for lowrate DDoS attack detection［EBOL］.  ［20250901］. https:webofscience.clarivate.cnwosalldbfullrecordWOS:0011 79043600053