基于极限学习机的模拟应用程序加载模式识别系统设计

Design of simulation application loading pattern recognition system based on extreme learning machine

针对传统识别系统受噪声影响较大而导致识别效果差的问题,基于极限学习机设计了新的模拟应用程序加载模式识别系统。将S3C2440处理器作为系统核心处理器便于实现寻址。然后利用SDRAM同步内部指令确保数据不会存在丢失的情况。基于此,通过参数控制、调度数据、USB数据通信等过程完成信号识别。在此基础上,抽取模拟应用程序加载模式信息,分析基于极限学习机的加载信号,再通过设置加载信道增益参数激活函数类内识别过程,从而抵抗外界干扰。依据遍历满足条件的候选集合得到最终识别结果。实验结果表明,该系统USB信号识别结果幅值与标准值基本一致,信号加载识别结果最高精准度为0.98,证明其具有良好的识别效果。

The traditional recognition system is greatly affected by noise,which results in poor recognition effect,so a new simulation application loading pattern recognition system is designed based on extreme learning machine(ELM).The processor S3C2440 is used as the core processor of the system for easy addressing.The SDRAM(synchronous dynamic random access memory)is then used to synchronize internal instructions to ensure that no data will be lost.On the basis of the above,signal recognition is completed by parameter control,data scheduling,USB data communication and other processes.On this basis,the information of simulation application loading mode is extracted to analyze the loading signal based on ELM.And then the activation function interclass identification process of the loading channel gain parameter is set to resist the external interference.The final recognition results are obtained by traversing the candidate set that satisfies the condition.The experimental results show that the amplitude value of the USB signal recognition result of the system is basically consistent with the standard value,and the highest accuracy of the signal loading recognition result is 0.98,which proves that the system has a good recognition effect.