Research On Parameter Identification Of Brushless Doubly-fed Stand-alone Power Generation System

Electromechanical energy conversion is one of the main ways to realize energy utilization.Applications such as wind power generation,shaft power generation,residual pressure power generation and flywheel energy storage all use electromechanical energy conversion to achieve efficient energy use.Due to the variable speed of the generator in above-mentioned applications,a converter is required to achieve constant frequency control.This type of power generation is called variable-speed-constant-frequency(VSCF)power generation,which can be further divided into grid-connected power generation and stand-alone power generation.Noticing that the stand-alone power generation can provide constant-frequency-constant-amplitude supply voltage,it has great potentials in the areas not covered by grid.The VSCF stand-alone generation is usually constructed based on permanent magnet synchronous generator or doubly-fed induction generator.However,in recent years,the VSCF stand-alone generation system based on brushless doubly-fed induction generator(BDFIG)has also received extensive attentions.BDFIG uses another set of stator winding to control the magnetic field,eliminating brushes and slip rings.Therefore,it combines the high reliability of permanent magnet synchronous generator and the partial power conversion of doubly-fed induction generator,and is especially suitable for applications under severe operational conditions with high maintenance costs.At present,the modeling,control,and speed sensorless researches of brushless double-fed stand-alone power generation system have been rich in research results,but all the abovementioned researches need to know BDFIG parameters first.Thus when machine parameters are unknown,they need to be identified first.However,the special structure of BDFIG brings more challenges to its parameter identification.The existing identification methods either base on theoretical calculations,requiring professional machine knowledge and detailed structure information;or calculate parameters based on open-circuited and short-circuited measurement data,resulting in a lot of setup wiring changes and cumbersome process.Besides,the existing parameter identification and controller design are relatively independent,and there is a lack of comprehensive evaluation of each parameter influence on system performance.In view of the above-mentioned shortcomings,this research focuses on brushless double-fed stand-alone power generation system,fully considering the practical application constraints and control system requirements,and studies parameter identification in depth.The contents include:(1)Based on the mathematical model of brushless doubly-fed stand-alone system,the influences of all resistances and inductances on steady-state and dynamic system characteristics are evaluated.The beneficial conclusion is obtained that the steady-state characteristics are mainly determined by inductances,while the dynamic characteristics are jointly determined by both resistances and inductances.This conclusion improves the pertinences of identification designs under different applications;(2)Aiming at applications where steady-state characteristics are concerned,an inductance identification method based on data fitting is proposed.This method first collects voltage and current data under no-load and loaded operation conditions,and then solves the required inductances based on semi-definite programming.Besides obtaining the required inductances,this method can also lay a foundation for the subsequent parameter identification methods;(3)Aiming at the applications where both steady-state and dynamic characteristics are concerned,a method for full parameter identification based on neural network technology is proposed.This method constructs a "neural network" based on brushless double-fed stand-alone system model,and only requires a small amount of voltage and current data under no-load and loaded conditions for network training.After the training is completed,all machine parameters can be extracted from the network,which effectively overcomes shortages of existing methods such as complicated calculation and weak ability for multiparameter identification;(4)Aiming at the adverse effects caused by identification errors accumulation,a method for online error correction based on the existing closed-loop control system and harmonic injection is proposed.This method utilizes the positive dependency between closed-loop control characteristics and parameter accuracy under no-load condition,and the parameter correction can be completed during the conventional start-up process of stand-alone system,which greatly simplifies the system installation and commissioning.