| Nowadays, Mechanical and electrical integration control systems are mainly constructed with foreign Programmable Logic Controller (PLC) and Field-bus Control System (FCS) which use industry computer and foreign PLC as core controller. But the cost is very expensive. For changing this circumstance, in this dissertation, we researched a new kind of PLC, which use embedded micro processor (ARM+FPGA) as its core processor. Inside the process, we can extend modules conveniently, such as: memory, I/O module of digital and simulation and CAN communication module and so on. The lattice LCD display module and touch-screen as program and system running status monitors.μClinux is embedded in it to realize multitask and friendly graphical user interfaces. The explaining programs of basic instructions and functional instructions of PLC are designed by C language independently. And the intelligent control arithmetics are realized in FPGA coprocessor, such as: PID, Fuzzy, heredety. These control arithmetics are transferred as function model in PLC ladder diagram.In this dissertation, BP neural network is realized in FPGA and be trained as PID controller. This intelligent PID controller is embedded in PLC to help ARM processing complex datas. Take advantage of repeat rub-write characteristic of FPGA, the capability of system will be consummation.This dissertation mainly included the following points: The realization of hardware model of BP neural network: This included floating point arithmetic organ, the storage and transfer of weight, the realization of activation function. The connection between the coprocessor (FPGA) and the main controller (ARM): That is the realization of double-bytes serial asynchronous transceiver. This Included baud rate generator, data sender and receiver. The temperature control system: This Include PID parameters ascertain and the simulation of the thermocouple temperature control system. |
