Font Size: a A A

Power Rectifier Controller Based On Can Bus Research And Design

Posted on:2009-09-08Degree:MasterType:Thesis
Country:ChinaCandidate:X J GuFull Text:PDF
GTID:2208360245982511Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Conventional PID Controller is adopted for parallel multi-unit high power rectifier system in traditional control project, and each controller is individual. In the fact operation, it usually appears the phenomena that other rectifier unit current is affected when one unit current is altered. Sometimes the current of the whole system fluctuates strongly, and even bring the whole system vibrate. Dynamic capability of the system is not favorable.First, math model of individual unit high-power rectifier system is analyzed in the article, and the result is that individual unit high-power rectifier system is non-linear system. Math model of parallel two-unit high-power rectifier system is analyzed based on it, and the conclusion is that coupling relation and non-linearity exist between two units. Thereby, the above analyse draws the conclusion that parallel multi-unit high power rectifier system is multivariate and non-linear. In order to achieve favorable dynamic and static state performance, decoupling algorithm is need to decouple and control.Secondly, the paper puts forward decoupling methods of multivariate system which mainly include traditional decoupling method, self-adapting decoupling method and intelligent decoupling method. Methods above are not adopted because they need accurate math model of controlled object and it is very difficult to establish accurate math model of parallel multi-unit high-power rectifier system. The method of MPIDNN(multivariable PID neural network) is adopted to decouple and control in broad with self-study and self-adaptive ability of neural cell, on the premise that the math model of controlled object is unknown. The paper puts forward MPIDNN forward algorithm and the anti-propagation algorithm of connected weight values, and analyzes the convergence and stability for MPIDNN multivariate control system, and brings forward selecting method of initial connected weight values in the end.Simulation model of two-unit high-power rectifier system is established and traditional PID control and MPIDNN control are both adopted. With compared the two, MPIDNN algorithm can realize decoupling in Parallel Multi-Unit high-power rectifier system, and features favorable dynamic and static performance.Generally speaking, each rectifier unit needs individual controller, and each unit controller is designed to multi input single output system, in order to fulfill harmonious control for multi unit by MPIDNN algorithm in each controller. Input of each unit not only included its current set value and feedback value signal, but also knows current set value and feedback value signal of other units which are communicated by network method between each unit. CAN bus, as a favorable real-time, high reliability and simply structure network, which support multi-master stations, is adopted to transmit the current information between each rectifier unit to realize decoupling and control by adopting CAN bus.The last, this study puts forward hardware and software design of rectifier for on the base of CAN bus. Hardware design is that 16-bit MCU 80C196KC is adopted as rectifier controller and CAN protocol controller SJA1000 is connected 80C196KC to realize CAN bus communication. Software design includes MPIDNN algorithm design and CAN bus software design.
Keywords/Search Tags:parallel multi-unit high power rectifier system, multivariable and nonlinear, multivariable PID Neural Network, decouple and control, CAN bus
PDF Full Text Request
Related items