Adaptive network-based fuzzy inference systems for sensorless control of PMSG based wind turbine with power quality improvement features

Abstract

The wind power generation is rapidly becoming competitive with conventional fossil fuel sources and already today is at par with new coal or gas fired power stations. The wind turbine design objectives have changed over the past decade from being convention-driven to being optimized driven within the operating regime and market environment. The wind turbines are growing in size, designs are progressing from fixedspeed, stall-controlled having drive trains with gearboxes, to become pitch controlled, variable speed and with or without gearboxes. The advancement in power electronics devices further supports the trend toward variable speed turbines. Today, the wind turbines in the market have a variety of innovative concepts, with proven technology for both generators and power electronics interface. However, the increasing penetration of large wind farms into electrical power systems also poses different kind of challenges due to their intermittent nature. This inspires the designers to develop both custom generators and power electronics devices with sophisticated modern control system strategies. Recently, variable-speed permanent magnet synchronous generator (PMSG) based wind energy conversion systems (WECS) are becoming more attractive in comparison to fixed-speed WECS. In the variable-speed generation system, the wind turbine can be operated at maximum power operating points over a wide speed range by adjusting the shaft speed optimally. Moreover, the use of Permanent Magnet reduces size, and weight of overall WECS, as there is no need of field winding and its excitation system. The absence of rotor winding also reduces heat dissipation in the rotor and hence improves the overall efficiency. This kind of configuration also find special favor for off-shore wind application, where the geared doubly fed induction generator requires regular maintenance due to tearing-wearing in brushes and gear box. To perform maximum power point tracking at different wind speeds, the variable speed operation of PMSG is required. For the variable speed operation of PMSG, generally vector control is preferred as it allows the independent torque and field control just like a simple DC motor control. The vector control of PMSG essentially requires the rotor position and speed information. For this purpose, usually shaft mounted speed and position sensors are used, resulting into additional cost and complexity of the system. In order to eliminate the sensors and their associated problems, a novel adaptive networkbased fuzzy-inference system (ANFIS) architecture is proposed for rotor position and speed estimation over wide range of speed operation. The ANFIS architecture has well known advantages of modeling a highly non-linear system, as it combines the capability of fuzzy reasoning in handling the uncertainties and capability of artificial neural network (ANN) in learning from processes. Thus, the ANFIS is used to develop an adaptive model of variable speed PMSG under highly uncertain operating conditions, which also automatically compensates any variation in parameters such as inductance, resistance etc. An error gradient based dynamic back propagation method has been used for the on-line tuning of ANFIS architecture. In the proposed work a PMSG based WECS is modeled for both isolated and grid connected system. In the isolated WECS operation, a wind-battery hybrid system is presented. The battery energy storage system (BESS) in the isolated system is used to absorb the wind power fluctuations and varying load demand. In grid connected system, the fault ride through capability of WECS is demonstrated under grid voltage sag/swell conditions. Another objective is to develop an advance controller for grid side inverter. Since the inverter works under highly fluctuating operating conditions, it is not possible to set the optimal value of gains for the conventional proportional-integral (PI) regulator. This may lead to false operation of inverter. To alleviate this problem an adaptive neurofuzzy controller is developed, which has well known advantages in modeling and control of a highly non-linear system. The main objective is to achieve smooth operation of grid side inverter, where the conventional PI controller may fail due to the rapid change in the dynamics of the overall system. The combined capability of neuro-fuzzy controller in handling the uncertainties and learning from the processes is proved to be advantageous while controlling the inverter under fluctuating operating conditions. Moreover, in the proposed work, the grid side inverter rating is also optimally utilized by incorporating the power quality improvement features. Normally, the grid interfacing inverter has very low utilization factor 20-30 % with a possible peak of 60% of rated output due to the intermittent nature of wind. Therefore, if the same inverter is utilized for solving power quality problem at point of common coupling (PCC) in addition to its normal task, then the additional hardware cost for custom power devices like APF, STATCOM or VAR compensator can be saved. Thus, the author have proposed a very simple and cost effective solution by using the grid side inverter as a load harmonics, load reactive power and load unbalance compensator of a 3P4W non-linear unbalanced load at PCC in a distribution network, in addition to its normal task of wind power injection in to the grid. Similarly, it has also been shown that the grid side inverter can also be used to maintain constant voltage at PCC for a dedicated load despite of voltage sag/swell and unbalance in grid side voltage.

Date	16 Jul 2010
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Ambrish Chandra (Supervisor)