INTERNATIONAL JOURNAL OF ELECTRICAL ENGINEERING & TECHNOLOGY (IJEET)

Transcription

1 INTERNATIONAL JOURNAL OF ELECTRICAL ENGINEERING & TECHNOLOGY (IJEET) International Journal of Electrical Engineering and Technology (IJEET), ISSN (Print), ISSN (Print) ISSN (Online) Volume 3, Issue 2, July September (2012), pp IAEME: Journal Impact Factor (2012): (Calculated by GISI) IJEET I A E M E A NOVEL APPROACH IN DESIGNING A FILTER FOR A SOLID ROTOR ALTERNATOR TO MINIMISE HARMONICS G.Suresh babu U.K.Choudhury G.Tulasi ram das ABSTRACT Solid rotor alternator is of great importance at high ratings for generation of electric power as they can with stand huge centrifugal forces developed in the alternator. When a non liner load is connected to such an alternator network it invariably injects harmonics into the system which becomes a power quality issue. In order to compensate for the harmonic content a shunt active filter is connected across the load. A single-phase inverter with four controller switches, a standard H-bridge inverter is chosen as active filter. The AC side of the inverter through a filter inductance is connected in parallel with the other nonlinear loads. The DC side of the inverter is connected to a filter capacitor. The shape of the current through the filter inductor is controlled by the neural network based controller such that the line current is of the same shape and in phase with the input voltage. The supply current spectral analysis shows the successful reduction of harmonics, which are produced by the load, by the neural network active filter using Matlab Simulink to model the system. The reduction in harmonics injected by multiple nonlinear loads is shown in simulation results. An attempt is made to perform an experiment to validate the results. Key words: Solid rotor, Shunt active filter, Neuro controller, Harmonics INTRODUCTION The solid rotor alternator is robust and simple in construction and finds applications when the requirement of large unit size and high speeds. The solid rotor is made of solid piece of alloy of mild steel chromium-molybdenum steel and slotted on its periphery to insert the field windings as shown in Fig 16 will generate sinusoidal EMF. The pictorial view of the solid rotor inserted in the stator can be observed on Fig 18.The use of solid state switching devices is increasing in many applications such as furnaces, adjustable speed drives, energy efficient lighting and switched mode power supply. The solid state switching devices behave as non-linear loads causing harmonic current injection which results in degraded power quality of the distribution and transmission system. 328

2 As more of these loads enter the power system, this phenomenon is predicted to get worse, causing increased power system losses and interference in the communication circuits and sensitive loads. Conventionally, Passive filters have been used to eliminate current harmonics in power systems, but not without their disadvantages. The system impedance strongly affects the filtering characteristic of the passive filter that may create series or parallel resonance, which causes amplification of harmonic current or voltage at a specific frequency. The variability of non-linear load operating conditions also affects the filter performance. The ageing of the filter capacitors will also affect the performance of passive filters. As both the harmonic and the fundamental current components flow into the filter, the passive filters are required to be designed with a high current rating. To improve the power quality without the disadvantages of passive filters, shunt active filters (SAF) [1] are developed. Shunt Active filters consists of power switching devices and passive energy storage elements such as capacitors and inductors. A specific control strategy is used to drive the power switching devices in order to produce current that is able to compensate for the harmonics injected by the load. The active filter used to compensate for these two nonlinear loads [2] is a single-phase inverter. The shunt active filter controlled by a Neural Network (NN) controller forces the line current to be in phase and of the same shape as the supply voltage. A single phase shunt active filter configuration is simulated in this work.. The effect of shunt active filter on harmonic reduction is also presented. An experimental setup is made as shown in Fig.1 FIig 1 : Solid-rotor alternator DESIGN OF NEURAL NETWORK CONTROLLED SHUNT ACTIVE FILTER INVERTER MODEL Assume vc > vs during the positive half cycle of the source voltage, i L can be made more positive by making v x =0 and i L can be driven towards zero by making v x = v c. During the negative half cycle of the source voltage, i L il can be made more negative by making v,= 0; and i L il can be driven towards zero by making v =-v c. From this heuristic understanding of the 329

3 circuit operation, it is concluded that the two switches in each leg of the inverter can be used for different tasks. Specifically switches S 3 and S 4 are used to force v x 0 and v x 0 respectively, while switches S I and S 2 actively shape, i L as shown in Fig.2 Fig.2 :single phase inverter used as an active filter For modeling and control purposes, the state of each inverter switch is the fined by a switching function such that U x = 1 when S x is in conduction state 0 when S x is open where x denotes the switch number [4]. The two switches in each inverter leg must operate in a complementary fashion. u 1 + u 2 = 1 (1) and u 3 + u 4 = 1 (2) Given the definition of u for each switch, v, can be written in terms of vc as V x = [ u 1 u 4 u 2 u 3 ] V c (3) or, by combining (1) and (2) V x can be written as V x = [ u 1 + u 4-1] V c (4) Further, i c = [ u 1 u 4-1] i 1 (5) 330

4 ISSN (Online) Volume 3, Issue 2, July- September (2012), IAEME From the analytical expressionss for v x and i c The state equations for the inductor current and capacitor voltage are written as (6) (7) DESIGN OF THE NEURAL CONTROLLER The advantages of neural controller over PID controller are: Adaptive learning: An ability to learn how to do tasks based on the data given for training or initial experience. Self-Organization: An ANN can create its own organization or representation of the information it receives during learning time. Real Time Operation: ANN computations may be carried out in parallel, and special hardware devices are being designed and manufactured which take advantage of this capability. Fault Tolerance via Redundant Information Coding: Partial destruction of a network leads to the corresponding degradation of performance. However, some network capabilities may be retained even with major network damage. The neural controller uses back propagation algorithm with a sigmoid function since it provides flexibility and is differentiable.the inputs to the neural controller are the capacitor voltage and the inductor current and output is the conduction time of the switches. The output of switching function block- I together with the decision frequency (fd) decides the pulse pattern for the switches SI and S2 and the output of the switching function block-2 gives the pulse pattern for switches S3 and S4.The block diagram of the controller with the active filter is shown in Fig.3 switching function block-1 switching function block-2 331

5 Fig.3: block diagram of the controller with the active filter The capacitor voltage and inductor current are given as input to the comparator first, if Vc and Il are in phase (both are positive) then the output is 1 otherwise 0. These inputs and outputs obtained are given neural network tool box where the training process takes place using back propagation algorithm (training of the data until minimum error is obtained) and the outputs are given as pulses to IGBT s in the shunt active filter. In the shunt active filter which acts as a voltage source converter the capacitor voltage is maintained constant and the inductor current is made equal in magnitude but opposite in phase to that of the load current consisting of harmonics. So, the harmonics injected [3] by the load current in the supply current are cancelled by the negative harmonics injected bye the filter output current thereby, resulting in the harmonic reduction. SIMULATION RESULTS The simulation diagrams of the filter without and with NN controlled ANN are shown in fig.4 and fig.5 respectively. The line current and commands to switches S1 and S2 are shown in Fig.11,whereas the source voltage and commands to switches S3 and S4 are given in Fig.12.The output current of uncontrolled load is as shown in Fig.9. The current spectrum without and with APF compensation are depicted in Fig.6 and Fig.7 respectively. The sum of active filter current and load current gives the supply current which is illustrated in Fig 14. The supply voltage, supply current and load voltage are shown in Fig.17.The supply current harmonics before and after the implementation of NN based SAF is shown in the table.1. The wave forms generated by simulation of the network by using ANN tool box of MATLAB/SIMULINK for source current, uncontrolled load current, active filter current, controlled load current and their sum is shown in the Fig 15. For an instance the input current of AC controller load [4] at a firing angle of 45 degrees is depicted in Fig 8. In order to appreciate the usage of APF compensation, the supply current wave form is captured without APF compensation as shown in Fig 10. The hard ware results recorded by power meter for odd harmonics of supply current are shown in Fig

6 Fig 4 : Simulated block Without SAF with Nonlinear load and AC controlled load Fig 5 : Simulink block diagram of single phase shunt active filter with neuro controller 333

7 Fig.6 current spectrum without APF compensation Fig.7 current spectrum with APF Fig 8:.Input current of ac controller load with firing angle

8 Current (A) Time (s) Fig 9: Output current of uncontrolled rectifier Fig10 : Supply Current without APF compensation (ie composite non linear load current) 335

9 Fig 11 : Line current and commands to S 1 and S 2 336

10 Fig 12 : Supply voltage and commands to S 3 and S 4 337

11 Fig 13: Odd harmonics of supply current recorded by power meter Fig 14: Waveform of Supply current, Active filter current and Load current of single phase Shunt Active Filter 338

12 Fig 15: Wave forms at different stages Fig 16: Designed Solid rotor 339

13 Fig 17: Waveform of supply current, Supply voltage and Load voltage Fig 18: Rotor Inserted In the stator 340

14 TABLE 1 HARMONIC CURRENT IN SUPPLY CURRENT Harmonic order Without APF With APF THD CONCLUSIONS The design procedure of a 230V, 1.79A/KVA load single phase neural based shunt active power filter is presented. The filter is intended to control harmonic current in supply line to the allowable value defined by IEEE Std-519. The filter actively shapes the supply current and compensate for the harmonics in the non linear load currents. The modeling of the system in Matlab simulink is presented and simulation results obtained show that the neural based active power filter controller is designed to minimize the harmonics in the supply current. The % THD measured in the presence of a controlled shunt active power filter are within the IEEE-519 harmonic standards. The designed circuit compensates up to the 27 th harmonics. REFERENCES 341

15 [ 1] N.B. Muthuselvan, Subhransu Shekar Dash and P. Somasundaram, Artificial Neural Network Controlled Shunt Active Power Filter, IEEE Indian conference Sep. 2006, D.O.I / INDCON [ 2] David A.Torrey and Adel M.A.M., AI-Zamel, "Single- Phase Active Power Filters For Multiple Non Linear Loads", IEEE Trans. on Power Electronics, Vol. 10, No.3, 1995, pp [ 3] Sharmeela.C., Mohan M.R., uma.g., "Line Harmonics Reduction Using Neural Based Controller For Shunt Active Filters", TENCON2003. conference on convergent Technologies for Asia-pacific Region, Volume 4, Oct Page(s): Vol.4 [ 4] Janko Nastran, Rafael cajhen, Matiia Seliger and Peter Jereb, "Active Power Filter For Non- Linear A.C loads", IEEE Trans. on Power Electronics, Vol.9. No. I, 1994, pp AUTHORS INFORMATION 1.G.Suresh babu, Associate professor, Dept of EEE, CBIT, Hyderabad , sbsreevatsaji@gmail.com 2. U.K.Choudhury, GM,( PEC&HR), BHEL corporate (R&D), Hyderabad G.Tulasi ramdas, professor, Dept EEE & Vice Chancellor JNTU,Kakinada,A.P. 342