Abstract: For the drawbacks of existing open-loop cognitive model and closed-loop feedback cognitive model with determinative cognition approach, an intelligent cognitive model with simulated feedback regulation mechanism and multi-cognition approach fusion is explored in this paper, to simulate human cognitive process of free regulating cognitive approaches in the face of various cognitive demand, with repeating intercomparison and deliberately from global to local cognitive characteristics. Firstly, an intelligent cognitive model with simulated feedback regulation mechanism and multi-cognition approach fusion is proposed, the structure and function of the model are proposed, and the operating mechanism of the model is designed. Secondly, the cognitive demanding of samples is analyzed to achieve the adaptive regulation from global to local of cognitive approaches to guide the establishment of optimized feature space and classified cognitive rules. Thirdly, the similarity index of cognitive result is defined to evaluate the credibility of cognitive outcomes to update the cognitive demanding of samples. Finally, the proposed model is applied to the offline handwritten Chinese character cognition. Based on the simulation experiment on GB 23122 - 80 handwritten Chinese sample library, the average cognitive accuracy of this method is 92.78 %. The experimental results show the superiority of the method.

Key words: Simulated feedback regulation mechanism, cognitive demanding, cognitive result similarity, fusion cognition 

摘要： 针对已有开环认知模型和基于固定认知方法的闭环反馈认知模型的缺陷，仿人面对不同认知需求时自由调整认知方法反复推敲比对从全局到局部认知的特点，探索了一种具有仿反馈调整机制的多认知方法融合智能认知模型。首先，提出了一种具有仿反馈调整机制的多认知方法融合智能认知模型，给出了认知模型的结构、功能和运行机制；第二，分析样本的认知需求，实现认知方法的自适应调整融合以指导不同样本优化特征空间和分类认知准则建立，从全局到局部认知；第三，定义了相似度评价指标，评估认知结果的可信度，以更新认知需求；第四，将该模型应用于脱机手写体汉字认知。通过对GB 23122 - 80手写体汉字样本库的仿真实验，平均认知精度达92.78 %，实验结果表明方法的优越性。

关键词: 仿反馈调整机制, 认知需求, 认知结果相似度, 融合认知