Design And Implementation Of The Embedded Intelligent Node

This dissertation has conducted a systematic research and an attemptive study on the key technologies used to design an Embedded Intelligent Node (EIN), the application of the neural network in the EIN, and the design of the management software for the EIN.As for the key technologies used to develop an EIN, this dissertation proposes a complete system design of the EIN including the hardware implementation and the software modularization of the system functions and introduces the whole design ideas as well as the program flowcharts of the proposed solution in details. The TCP/IP–based communication ability and some network services, such as HTTP, FTP and SMTP, are provided in each proposed EIN, based on the WATTCP stack under a DOS environment. A series of accessorial functions have also been implemented herein. With the TCP/IP-based communication technology, the EIN is capable to share information resources over the whole Intranet/Internet.As for the applications of neural network in the EIN, this dissertation carries a lot of researches and experiments on the back-propagation algorithm (BP), the genetic algorithm (GA), optimization methods of BP and GA, as well as the usage of GA to optimize the initial weights of neural network. The study has overcome the defects of basic BP algorithm such as slow convergence speeds and great possibility of getting into local minimums. According to the various engineering circumstances, the neural-network-studying module of the EIN can dynamically retrain the networks by adaptively choosing the optimal methods. As a result, a general software module of neural network studying has been formed, and therefore, the enhancement of the intelligence of the EIN.As for the design of the management software for the EIN, many attentions have been taken on the techniques used to design and develop the application software using the visual c++ & MFC, including network communication, multitask mechanism, graphical user interface (GUI), and the realization of web-based telemonitoring using ActiveX component technologies, etc.