Você está na página 1de 17

INTRODUCTION:

In this chapter the background and motivation for the need of fault diagnosis and fault-tolerant control of wind turbines are described. A brief overview of fault diagnosis and fault-tolerant control and the application to wind turbines is then given. This is followed by presenting the scope of the project and outlining the content of the this.

BACKGROUND:
Evolution of technology has increased power demands to operate the modern electrical equipment. This has increased the demand for fossil fuels and has made electrical energy more expensive. Because of such high demands for electric power, it is necessary to focus on renewable energy sources, as fossil fuel resources are limited. Furthermore, to protect the environment the emissions of greenhouse gases and undesired particles into the atmosphere have to be reduced. Among the renewable energy sources available today, wind power is the world's fastest growing [Wind Energy News, 2007]. With an annual growth rate in installed wind energy capacity of 30% on average throughout the past 10 years, wind turbines are de_nitely up and coming [GWEC, 2009,p. 15]. For several reasons wind energy is growing fast: it is cheap,

inexhaustible, widely distributed,clean, and climate friendly [Wind Energy News, 2007].

As many wind turbines are installed o_shore, a non-planned service can be highly costly, so it would be bene_cial if fault-tolerant control schemes could help the turbines produce some energy from the time a fault is detected to the next planned service. Furthermore, the implementation of fault diagnosis schemes entails operational bene_ts due to its feature of early detection of faults, which can make the wind turbine operate safer and reduce costs as a result of possible improved maintenance procedures [Hameed et al., 2009, p. 3]. Therefore, fault diagnosis and fault-tolerant control of wind turbines may over several benefits. Prevent catastrophic failures and faults deteriorating other parts of the wind turbine by early fault detection and accommodation. Reduce maintenance costs by avoiding replacement of functional parts, by applying conditionbased maintenance instead of timebased maintenance. Provide diagnostic details to the maintenance the state by remote diagnosis.

Increase energy production when a fault has occurred by means of fault-tolerant control. This section has addressed benefits of exploiting wind power and improving the reliability of

FAULT DIAGNOSIS AND FAULT TOLERENT CONTROL:


The purpose of this section is to give an introduction to fault diagnosis and fault-tolerant control,since these topics are addressed in this thesis. This is accomplished by providing a brief overviewof the terminology and available methods in these fields.Finally, the available fault diagnosis a faulttolerant control algorithms for wind turbines are discussed. Fault-tolerant control system is a system, which prevents component failures from becoming failures on the system level. The control system is though allowed to have degraded performance in some cases when exposed to a fault. A fault is a change in the characteristics of a component, while a failure makes a component completely dysfunctional. Overall, there are two di_erent types of fault-tolerant control systems; these are called passive fault-tolerant control systems (PFTCS) and active fault-tolerant control systems (AFTCS) [Zhang and Jiang, 2003]. Passive fault-tolerant control systems are designed to be resilient to a speci_ed set of faults.This

implies that the same controller is utilized both for the fault-free as well as the faulty system.In the design of passive fault-tolerant control systems, di_erent performance requirements are set up for the normal system and for the faulty system [Niemann and Stoustrup, 2005b].Therefore,these systems are not referred to as robust systems, but as reliable systems. Active fault-tolerant control systems have, in contrast to passive fault-tolerant control systems, different controllers for the normal system and for the faulty system. This implies that the state of the system has to be determined by fault diagnosis algorithms. The information from the fault diagnosis algorithms is utilized in a supervisor, to reconfigure the control system for accommodating faults.Fault diagnosis used in active fault-tolerant control systems consists of multiple parts, since faults both have to be detected, isolated, and in some cases estimated. Fault detection should detect that a fault has occurred and can rely on either an active or a passive approach. Passive fault detection should detect faults by comparing the expected system behavior with the observed system behavior; hence, it does not a_ect the system. In contrast to this, active fault detection uses injection of auxiliary signals into a system to improve the fault detection capabilities or in some cases make fault detection possible.Fault isolation should point out faulty components in the system. This is important information when faults should be accommodated, since the control system cannot rely on a faulty component. Some faults do not turn a component on or off, but have an intermediate state. This implies that fault estimation has to

determine the fault sizes in order to accommodate these. There are generally two types of faults: abrupt faults and incipient faults. An abrupt fault is generally easier to detect than an incipient fault, but it might have severe consequences for the system, since it happens instantaneously.

EXISTING FAULT DIAGNOSIS AND TOLERENT CONTROL METHODS:

FAULT

There exists several methods for designing fault diagnosis algorithms and fault-tolerant controllers, and the basics of these are outlined in this subsection.In the design of a passive fault-tolerant control system a good performance for the nominal control system has to be achieved while a graceful degradation is allowed in the case of a fault. In [Niemann and Stoustrup, 2005b] this is achieved by creating a controller structure relying on two separate controllers. One controller outputs nothing when the control system possesses nominal behavior, while the second controller equals the nominal controller. In the case of a fault, the fiest controller outputs a non-zero value; hence, changing the behavior of the control system. Other methods, as e.g. [Liao et al., 2003], rely on a multi-objective control system, which has a set of minimum requirements to the faulty system and are optimized to improve the performance of the normal system. In the design of an active fault-tolerant control system the first step is to design a fault diagnosis system. This essentially consists of designing a residual generator which is sensitive towards faults and insensitive towards other

exogenous inputs to the system. Methods for this include parity space approaches where, if possible, a perfect decoupling between disturbances and residual is designed.Another approach is to design a change detection algorithm, e.g. based on a CUSUM test, which is able to detect a change in the mean value of a signal. Finally, Kalman filter approaches can be utilized by making a description of the fault become part of the system model, allowing the fault to be estimated. These approaches are suitable for diagnosing incipient faults. When the fault has been diagnosed the active faulttolerant control system must be reconfigured.This could for example be to reconfigure the controller to rely on estimates instead of measurements.The active fault-tolerant control system is recon_gured by use of a supervisor, which chooses an appropriate controller from a family of possible controllers, designed for each fault state.

Fault Diagnosis and Fault-Tolerant Control of Wind Turbines:


In this subsection the current state of fault diagnosis and fault-tolerant control of wind turbines is outlined, by examining the available literature.Modern wind turbine control systems are equipped with condition monitoring systems and fault detection systems. These systems detect and isolate faults and determine the current operating conditions of the wind turbine. The available information can then be utilized for predictive maintenance,which basically predicts when maintenance should be performed to avoid failures.Most condition monitoring systems and fault detection systems in wind

turbines are signalbased and utilize e.g. vibration analysis to detect and isolate faults. This has enabled successful condition monitoring of bearings in the gearbox and the generator among others. Numerous other signal-based approaches utilized in wind turbines can be found in [Hameed et al., 2009].Only a few model-based fault diagnosis approaches exist for wind turbines; among these are fault diagnosis systems for pitch sensors and pitch actuators [Wei and Verhaegen, 2008] and [Donders,2002]. These diagnosis systems estimate some parameters in the pitch system, and determine if a fault has occurred based on these estimates. It has not been possible to find any fault-tolerant control systems for wind turbines in the literature review. The common approach is to deploy condition monitoring systems and shut down the wind turbine in case of a fault. However, in a few cases thoughts about fault accommodation have been presented, but have not been tested or simulated.In this section the terminology and available methods used in the fields of fault diagnosis and fault-tolerant control have been outlined. Additionally, fault diagnosis and fault-tolerant control applied to wind turbines have been investigated.The investigation has revealed that fault diagnosis algorithms exist for wind turbines, but mostly using signal-based methods. Additionally, only a few fault-tolerant control systems for wind turbines have been found.

CAN(CONTROLLER AREA NETWORK):


This paper presents a methodology to provide an advanced automation module. The automation such as carbon monoxide emission detection and real time control, seat belt control, cruise control has been

implemented using ARM7.The ARM7 kit and automobile hardware are interface and communicate directly using CAN bus. Rather than tradition scheme of CAN in automation will offer increase portability and efficiency compare with other possible protocols for car automation. The benefits of CAN in achieving automation, over other tradition schemes will offer increase flexibility and expandability for future technology.

CAN IN WIND TURBINE:


This paper is a CAN based architecture designed for the purpose of monitoring and fault diagnosis of wind turbine. CAN is a Message based protocol designed specifically for Automotive, later Aerospace, Industrial automation and Medical equipments. CAN interface module is used to communicate the monitored parameters between the wind turbine and the control center. During the transmission of the data from one node to another node disturbance occurs. To avoid these disturbances we propose CAN protocol.

Transistors:
Function Transistors amplify current, for example they can be used to amplify the small output current from a logic

chip so that it can operate a lamp, relay or other high current device. In many circuits a resistor is used to convert the changing current to a changing voltage, so the transistor is being used to amplify voltage. A transistor may be used as a switch (either fully on with maximum current, or fully off with no current) and as an amplifier (always partly on). The amount of current amplification is called the current gain.

Types of transistor: There are two types of standard transistors, NPN and PNP, with different circuit Transistor circuit symbols symbols. The letters refer to the layers of semiconductor material used to make the transistor. Most transistors used today are NPN because this is the easiest type to make from silicon. If you are new to electronics it is best to start by learning how to use NPN transistors.

The leads are labelled base (B), collector (C) and emitter(E). These terms refer to the internal operation of a transistor but they are not much help in understanding how a transistor is used, so just treat them as labels! A Darlington pair is two transistors connected together to give a very high current gain.

ARM 7(LPC2148):
The NXP (founded by Philips) LPC2148 is an ARM7TDMI-S based high-performance 32-bit RISC Microcontroller with Thumb extensions 512KB onchip Flash ROM with In-System Programming (ISP) and In-Application Programming (IAP), 32KB RAM, Vectored Interrupt Controller, Two 10bit ADCs with 14 channels, USB 2.0 Full Speed Device Controller, Two UARTs, one with full modem interface. Two I2C serial interfaces, Two SPI serial interfaces Two 32-bit timers, Watchdog Timer, PWM unit, Real Time Clock with optional battery backup, Brown out detect circuit General purpose I/O pins. CPU clock up to 60 MHz, On-chip crystal oscillator and On-chip PLL.

MODELLING:
To facilitate a model-based approach in the design of the fault diagnosis and fault-tolerant control algorithms, a non-linear model of a variable-speed, variable-pitch wind turbine is set up. Additionally, the model acts as a simulation model for testing the designed algorithms. The model is based on a static model of the aerodynamics, a two-mass model of the drive train, an electromechanical model of the generator, dynamic actuator models, and zero-mean Gaussian distributed measurement noises. The parameters for the wind turbine model are provided by kkelectronic a/s, similar applies for the variances of the additive measurement noises. The input to the model is generated by a wind model, which includes wind shear, tower shadow, and turbulence.

The aerodynamics of the wind turbine is non-linear and is described in form of a lookup table, where the e_ciency of the aerodynamics is determined from the pitch angles of the blades and the tip-speed ratio. This part of the model is non-linear and introduces parameters that vary dependent on the operating conditions.

FAULT ANALYSIS:
A fault analysis is performed in order to determine the faults which should be considered in this project. First, a number of possible component faults are chosen and their propagations through the system are determined by describing their effects on the surrounding components of the system.Subsequently, the severity of the end-e_ects and the occurrence rates of the faults are estimated to select the faults of highest priority.The frequency of the faults are approximated based on statistics reported in the literature,whereas the severities of their end-e_ects are determined based on simulations. For conducting these simulations, a reference controller without faulttolerant capabilities is designed based on information about an existing control system. To limit the number of faults to be handled during this project, it is decided to focus on the faults related to the pitch sensors, pitch actuators, and generator speed sensor. The motivation behind selecting faults related to the pitch system, which e.g. cause rotor unbalance,is that these faults increase fatigue loads on the wind turbine structure. It is further seen that changed dynamics of the pitch system, caused by low pressure or high air content in the hydraulic oil, may result in an unstable closed-loop system. Finally, the main controllers in

the entire operating range of the wind turbine depend solely on the measured generator speed. Hence, it is essential to diagnose and accommodate any troubles regarding this particular measurement.

FAULT -TOLERANT CONTROL:


In the design of the fault-tolerant control system the faults are divided into two categories: faults that do not a_ect the dynamics of the system and faults that a_ect the dynamics of the system.The faults that do not affect the dynamics of the system are accommodated by correcting the measurement and reference signals, based on information provided by the fault diagnosis algorithms.This enables the faulttolerant control system to be designed independent of the controller structure and without affecting the nominal performance of the control system.Faults that affect the dynamics of the system are accommodated using both active and passive fault-tolerant control, to enable a comparison of the two methods. The main difference between these methods is that the active fault-tolerant controller depends on the fault diagnosis algorithms,while the passive fault-tolerant controller is independent of these algorithms. Both fault-tolerant controllers are LPV controllers, which are based on a common LPV system description, accounting for the parameter-varying nature of the wind turbine.

CONCLUSION:
In this project fault diagnosis and fault-tolerant control algorithms are developed for improving the reliability of wind turbines. The study is based on a model of a variable-speed, variablepitch 4:8 MW wind turbine, which represents a realistic but _ctitious wind turbine,

to which the collaborator kk-electronic a/s has provided the parameters. The faults considered in the project are chosen based on a severity and occurrence analysis, in which the most frequent and severe faults are identified. The analysis primarily focuses on sensor and actuator faults, which are included in the model of the wind turbine. In the diagnosis of the faults, model-based fault diagnosis algorithms are primarily developed, due to their improved resilience towards making incorrect decisions compared to signal-based approaches.Additionally, only the already available sensor information is utilized in the diagnosis of the faults. To obtain a fault-tolerant control system, di_erent approaches are utilized dependent on the nature of the faults. For faults that affect the dynamics of the system, active and passive fault-tolerant controllers are designed and compared. To access the performance of the designed algorithms, Monte Carlo simulations are performed to evaluate the robustness of the algorithms, where this is considered necessary.

DOCUMENTATION

CAN BASED WIND TURBINE WITH ONLINE MONITORING AND CONTROLLING

Name AJAY M JEYAVIJAY N AZARUDEEN A

Register Number 31610106002 31610106041 31610106303

INTERNAL

: Mr. S.OHMSHANKAR

Você também pode gostar