| 000 | 05753nas|a2200397 i 4500 | ||
|---|---|---|---|
| 005 | 20260817163304.0 | ||
| 008 | 180200s1999 th uu m rtt 00| a1eng d | ||
| 035 | _a.b11777618 | ||
| 099 | 9 | _aAIT Diss. no. ET-99-01 | |
| 100 | 0 | _aBoonserm Changaroon | |
| 245 | 1 | 3 | _aAn artificial neural network based power system damping controller (PSDC) |
| 260 |
_aBangkok : _bAsian Institute of Technology, _c1999 |
||
| 300 | _a156 leaves | ||
| 490 | 1 |
_aDissertation ; _vno. ET-99-01 |
|
| 500 | _aA dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering, School of Environment, Resources and Development | ||
| 502 | _aThesis (Ph.D.) - Asian Institute of Technology, 1999 | ||
| 520 | _aAn interconnected power system, depending on its size, has hundreds to thousands of modes of oscillations. A conventional power system stabilizer (PSS) designed for enhancing the damping of these low frequency oscillations is totally acceptable in a power system. Many large utilities have successfully tuned PSS for both interarea and local modes of oscillations using the conventional approach that has been well proven over the years. However, the values of parameter of PSS are directly affected by power system loading conditions. In order to avoid generating the high frequency self-sustained oscillations due to a high PSS gain, the stabilizer has to be tuned under a paiticular condition of the power system. The performance of PSS at this condition is normally better than that of other conditions. Thus, a set of PSS parameters, which provide a good dynamic performance under one condition, may no longer yield satisfactory results for another condition. Furthermore, the parameters of model representing the dynamics of the power system are required for the conventional tuning technique. These parameters sometimes are not matched to the actual system. The difference may result in poor tuning of PSS. Many adaptive control schemes have been applied to overcome this problem. Attificial Neural Network (ANN) is an approach that can provide better performance than the conventional PSS over a wide range of operating conditions. The Functional Link Network (FLN) model of a neural network, a single layer with the enhanced inputs, has been selected and applied for developing the power system damping controller (PSDC) in this thesis. Both the single layer and multilayer models have been tested and utilized to identify the dynamics of power output of a generator. Test results indicate that the FLN based functional expansion model provides the best performance for identifying the dynamic characteristics of a power system. A hybrid stabilizer that consisted of an identifier based on the tested FLN model and an optimization based predictive has been developed for damping out the oscillations in the simulated system taken from a practical system of the Electricity Generating Authority of Thailand (EGAT) system. The proposed FLN based identifier can work well with the predictive controller. The hybrid stabilizer can enhance the system damping over a wide range of operating conditions, but it consumes a long CPU time for processing the predictive control algorithm. A Neuro-PSS that consists of a neuro-identifier and a neuro-controller has been developed to show its performance and fast algorithm. The proposed neuro-PSS overcomes a drawback of the hybrid stabilizer and provides better performance over a wide range of operating conditions compared with that of the conventional PSS. Besides the proposed neuro-PSS, the FLN model has also been developed for Static Var Compensator (SVC) applications. A SVC stabilizer based on the FLN model has been trained and tested to show its performance on a ten-bus test system and also on a practical system of the EGA T equivalent system. Though both the neuro-PSS and the FLN based stabilizer for SVC can provide satisfactory output by utilizing the on-line training, it is difficult to measure the CPU time required for the training algorithm of the two stabilizers in the software simulation tests. To show the capability of the neuro-PSS for an on-line application, the PC-based hardware prototype has lll been developed for testing the training algorithm of the proposed neuro-PSS under real-time conditions. A three-phase alternator-network real-time simulator is used for the study system to provide practical signal quantities for the neuro-PSS via the interface circuit of a hardware prototype. Results obtained from the real time tests indicate that the training algorithm of the neuro-PSS is fast enough for on-line applications. | ||
| 650 | 0 | _aDamping (Mechanics) | |
| 650 | 0 | _aNeural networks (Computer science) | |
| 700 | 1 |
_aDhadbanjan, Thukaram, _eChairperson |
|
| 700 | 0 |
_aSurapong Chiraratananon, _eExamination Committee |
|
| 700 | 1 |
_aSadananda, Ramakoti, _eExamination Committee |
|
| 710 | 2 |
_aElectricity Generating Authority of Thailand, _eScholarship donor |
|
| 810 | 2 |
_aAsian Institute of Technology. _tDissertation. _vno. ET-99-01 |
|
| 856 |
_3Full-Text _uhttp://203.159.5.9/ait-thesis/detail.php?q=B11870 |
||
| 907 |
_a.b11777618 _bmnait _cu |
||
| 902 | _a240421 | ||
| 998 |
_b0 _c000207 _dm _ea _fu _g3 |
||
| 945 | _lmnait | ||
| 945 | _lmnarc | ||
| 942 | _c22 | ||
| 942 | _c40 | ||
| 909 |
_aBarcode : 30050120557359 _bCREATED : 2000-09-02 _cRECORD # : i12174877 _dLPATRON : 1016633 _eLCHKIN : 2005-05-08 _f# RENEWALS : 0 _g# OVERDUE : 0 _hIUSE3 : 0 _iTOT CHKOUT : 2 _jTOT RENEW : 0 |
||
| 909 |
_aBarcode : 30050160037049 _bCREATED : 2016-01-26 _cRECORD # : i1287405x _dLPATRON : 0 _eLCHKIN : - _f# RENEWALS : 0 _g# OVERDUE : 0 _hIUSE3 : 0 _iTOT CHKOUT : 0 _jTOT RENEW : 0 |
||
| 999 |
_c8180 _d8180 |
||