在更精细地的分类基础上，蛋白质二级结构预测是预测和分析蛋白质功能的关键环节。交叉视皮层模型（ICM）作为模拟哺乳动物视皮层神经元局部连接与脉冲耦合特性的新型人工神经网络模型，可利用时间脉冲编码形式，从时空角度有效地提取输入的鉴别特征。本项目拟以蛋白质序列的生物特性为出发点，结合ICM的脉冲编码特性，围绕蛋白质序列的ICM脉冲时间编码、多特征提取与融合、基于强化学习等理论的模式分类学习、神经元参数的智能优化等方面开展研究，提出有效可行的蛋白质二级结构预测技术方案。本项目的研究能有效规避传统生物实验测定方法存在的不足和困难，高效地提取蛋白质序列特征信息并剔除冗余信息，强化数据模式的可鉴别性，为蛋白质的结构预测提供更为有效的特征信息，进而为蛋白质的结构分析和预测提供新的技术和方法。

Protein secondary structure prediction (PSSP) is the key issue for the prediction and analysis of protein function. Intersecting cortical model (ICM) is a new artificial neural network model which simulates the local connection and pulse coupling nature of neurons in mammalian visual cortex, and the model can effectively extract the classified features of protein sequence with its pulse coding. This study combines the biological characteristics of protein with the pulse coding characteristic of ICM for PSSP, and it includes the ICM pulse coding of protein sequence, multi-feature extraction and fusion, pattern classification based on reinforcement learning theory or other models, and parameters intelligent optimization of ICM. This study can effectively overcome the deficiencies and difficulties of traditional experimental biology, provide effective information for PSSP by extracting features and eliminating redundant information, strengthen the identifiability of data patterns, which can further provide effective features for PSSP. This study can provide the new technology and method for protein structure analysis and prediction, ulteriorly.