عنوان مقاله
مدل شبکه عصبی نویز سیگنال چند بعدی
فهرست مطالب
چکیده
توصیف کار
شبکه عصبی نویز – سیگنال پیکره بندی و بایاس متعدد
مثالهای کار شده و نتایج
بخشی از مقاله
توصیف کار
در کار حاضر، شبکه عصبی نویز- سیگنال ، شرایط بایاس (VDs, IDS) و تیپ پیکره بندی (CT) را در برمی گیرد به گونه ای که پارامترهای عملکردی دستگاه نه تنها در شرایط فرکانس عملیاتی واحد حول شرایط بایاس آموزش دیده که تعمیم فرکانس واحد (SFG) نامیده شده است، بلکه در کل باند فرکانس عملیاتی حول شرایط بایاس آموزش ندیده تعمیم داده می شود، این یکی تعمیم کل باند فرکانس (WFBG) نامیده می شود.
کلمات کلیدی:
Multidimensional signal-noise neural network model F.GUneg H.Torpi F. G U rgen Indexing terms: Neural network model, Microwave transistors, Signal parameters, Noise parameters Abstract: Signal and noise behaviours of microwave transistors are modelled through the neural network approach for the whole operating ranges including frequency, bias and configuration types. Here, the device is modelled by a black box whose small-signal and noise parameters are evaluated through a neural network based upon the fitting of both of these parameters for multiple bias and configuration. The concurrent modelling procedure does not require the solving of device physics equations repeatedly during optimisation, and by this type of modelling the signal (S) and noise (N) parameters can be predicted not only at a single operation frequency around the chosen bias condition for a configuration, but at the same time for the whole operation frequency band for the same operating conditions, with good agreement compared to the measurements. 1 Description of the work In this work the signal-noise neural network in [l] is extended to include bias condition (VDs, IDS) and configuration type (CT) so that performance parameters of the device can be generalised not only at a single operation frequency around the trained bias condition, which may be named single frequency generalisation (SFG), but at the same time in the whole operation frequency band around an untrained bias condition, which may be named as whole frequency band generalisation (WFBG). The same performance measures as [l] are also utilised for this model. The literature for the transistor modelling is given extensively in [l]. Applications of the neural networks in the microwave circuits reported in literature include automatic impedance matching [2], microstrip circuit design [3], microwave circuit analysis and optimisation [4] and, most recently, modelling of monolithic microwave integrated circuit (MMIC) passive elements [5] and simulation and optimisation of interconnects in high-speed VLSI circuits. The multide bias-configuration signal-noise 0 IEE, 1998 IEE Proceedings online no. 19981712 Paper first received 28th February and in revised form 3rd October 1997 F. GuneS and H. Torpi are with the Yildiz Technical University, Electrics and Electronics Faculty, Electronics and Communication Engineering Department, 80750 Be$ktaS-Istanbul, Turkey F. Giirgen is with BogaziGi University, Computer Engineering Depdrtment, 80815, Bebek-Istabul, Turkey neural network is described in the second Section in detail, and worked examples and conclusion are given in the last Section.
