Research On Predictive Current Control And Parameter Identification For DFIG

As one of the most effective renewable energy sources to response to theworld’s energy and environmental crisis, wind power has been concernedworldwide. In recent years, with the rapid growth of the world’s wind powerindustry, the development of China’s wind power market has also maderemarkable achievements, and demonstrated the enormous potential. However,while China’s wind power is developing rapidly, there are also some deficiencies,for example, the core device control technique is relatively backward anddependent on foreign countries. Therefore, this paper conducts research oncontrol strategy and parameters identification method of doubly-fed inductiongenerator (DFIG) in-depth.DFIG is the most mature and most widely used wind power generator.Because it’s mainly excited by rotor current, the control of rotor current is thekey of the DFIG control system. In this paper, by establishing the mathematicalmodel of DFIG, principle of variable speed constant frequency (VSCF) andtraditional PI control of the rotor current are introduced. In ad dition, in order toachieve flexible grid-connecting, a segmented control strategy for connectingDFIG to the grid is used.PI controller is usually used for the rotor current control of DFIG, but thedynamic response of PI controller is slow, which restricts the operationperformance of the system when wind speed changes rapidly. To solve thisproblem, a deadbeat control strategy for rotor current is proposed. On the basisof the discrete mathematical model of DFIG, the deadbeat rotor current controlequations are presented. To compensate the error caused by the system delay, acurrent prediction algorithm based on double sampling and Luenberger observeris combined to the deadbeat controller to improve the dynamic response speed ofthe rotor current. The effectiveness of this control method is verified bysimulation.The machine parameter is not constant while DFIG is actually running, aswell as the offline parameter identification is not appropriate since DFIG isconnected to the grid. Therefore, an online parameter identification based onleast square method is proposed. After obtaining the least square form of themathematical model, online identification of multiple parameters of DFIG isrealized by using recursive least squares estimation with forgetting factor. Thesimulation results can be seen that the accurate parameters can be identified with good stability and convergence.After programming the predictive deadbeat rotor current control and onlineidentification of machine parameters, the strategy is verified by experiments onthe experimental platform DFIG. The experimental result shows that the rotorexiting current can be regulated to follow the command of master controller androtor speed, and machine parameters can be estimated steadily and accurately bythe online parameter identification based on least square method even if theoutput power steps.