基于微流控芯片的脉冲神经型膜计算研究

发布时间：2018-06-26 05:41

本文选题：脉冲神经型膜计算 + 微流控芯片　；参考：《安徽理工大学》2017年硕士论文

【摘要】：在微电子技术逐渐地改进和完善以后,集成电路物理性质的限制最终将会被凸显出来。而在当今计算机科学领域关于传统芯片的替代品这一研究热点,就是为了打破这种限制而去提高计算机的性能。膜计算作为生物计算的一个重要分支,是当前生物计算研究的热点问题,其具有非确定性、极大并行性和分布式等特点。在计算机科学研究领域中膜计算模型的研究具有重要的实用价值,并且在生物系统建模中还具有新型的理论意义。但其研究大多都是在传统计算机上的,由于其局限性并不能充分体现出膜计算的计算特点。而微流控芯片用于制作集成到纳米级的操作环境中的网络化微通道、检测与控制模块,是微流控技术的载体。并且微流控芯片作为硬件平台,其非常高的并行能力有着不可替代的优势。本文尝试使用微流控芯片作为实现膜计算的硬件平台,结合其微型化、集成化和并行性等特点,实现微处理器最基本的功能结构。利用膜计算的特点,设计了完成逻辑运算和算术运算的模型,实现了带符号数的运算。其输入数以二进制补码的形式输入到模型中,在运算结束后以二进制补码的形式作为结果输出。并且详细说明了模型的计算步骤和MeCoSim软件仿真结果。在此基础上,使用聚苯乙烯代替脉冲神经型膜计算的脉冲,并且根据微流控芯片的Coulter计数器、介电泳分离、气动微阀和流体驱动等模块,设计了基于膜计算模型的逻辑和算术运算芯片。并且通过FPGA对检测信号进行采集、处理,通过返回信号对芯片电极、微阀等进行控制,实现了逻辑运算和算术运算模型的具体功能。在对膜计算研究的逐步深入以后,传统计算机的缺点也慢慢显现出来。那就是随着膜计算建立的模型越来越复杂,其神经元和脉冲数成倍的增长,从而导致膜计算处理信息的效率低下。并且膜计算最大并行性的特点在传统计算机上也不能很好的体现出来。而且长久以来,众多学者很少研究其硬件实现,大多都致力于研究膜计算的理论与建模。因此结合膜计算的思想,设计和实现出集成了各种结构功能,并且具有具体功能的微流控芯片就很有必要。这既能提高运算的准确性,又能解决物理限制的问题,进而推动生物计算机的研究。将膜计算的并行性与微流控芯片的微型化结合起来,很有可能在不久的将来使得基于膜计算的"CPU"得以实现。因此,基于微流控芯片的脉冲神经型膜计算的研究有着极其显著的学术和实用意义。
[Abstract]:With the gradual improvement and improvement of microelectronic technology, the limitations of physical properties of integrated circuits will eventually be highlighted. In the field of computer science, the research focus on the replacement of traditional chips is to improve the performance of computers in order to break this limitation. As an important branch of biological computing, membrane computing is a hot issue in biological computing. It has the characteristics of uncertainty, maximum parallelism and distribution. In the field of computer science, the study of membrane computing model has important practical value, and also has a new theoretical significance in biological system modeling. However, most of the research is on traditional computer, because of its limitations, it can not fully reflect the characteristics of membrane calculation. The microfluidic chip is used to fabricate the networked microchannel, detection and control module, which is the carrier of the microfluidic technology. And microfluidic chip as a hardware platform, its very high parallel capability has irreplaceable advantages. This paper attempts to use the microfluidic chip as the hardware platform to realize the film computing, combining its miniaturization, integration and parallelism, to realize the most basic functional structure of the microprocessor. Based on the characteristics of membrane computing, a model of logical and arithmetic operations is designed, and the symbolic operation is realized. The input number is input into the model in the form of binary complement, and the output is in the form of binary complement after the operation. The calculation steps of the model and the simulation results of MeCoSim software are described in detail. On this basis, polystyrene was used to replace the pulse calculated by the pulsed neural membrane, and according to the Coulter counter of the microfluidic chip, the dielectric electrophoresis separation, the pneumatic microvalve and the fluid driving module, etc. A logic and arithmetic chip based on membrane computing model is designed. The detection signal is collected and processed by FPGA, and the chip electrode and microvalve are controlled by the return signal. The specific functions of logic operation and arithmetic operation model are realized. After the study of membrane computing is further developed, the shortcomings of traditional computer appear. That is, as the model of membrane computing becomes more and more complex, the number of neurons and pulses increases exponentially, which leads to the inefficiency of membrane computation in processing information. Moreover, the characteristics of maximum parallelism of membrane computing are not well reflected on traditional computers. For a long time, many scholars seldom study its hardware implementation, most of them focus on the theory and modeling of membrane computing. Therefore, it is necessary to design and implement microfluidic chip which integrates all kinds of structure and functions with the idea of membrane calculation. This can not only improve the accuracy of calculation, but also solve the problem of physical limitation, and then promote the research of biological computer. Combining the parallelism of membrane computing with the miniaturization of microfluidic chips, it is very likely that the "CPU" based on membrane computing will be realized in the near future. Therefore, the study of pulse neural membrane calculation based on microfluidic chip is of great academic and practical significance.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN492

【相似文献】