本論文提出一非對稱歸屬函數之派翠機率模糊類神經網路對內藏式永磁同步馬達驅動系統即時慣量鑑別之技術，估測出的慣量將應用於內藏式永磁同步馬達驅動系統IP速度控制器之增益設計，並線上自動調變。本論文首先研究了具有IP速度控制器的磁場導向控制內藏式永磁同步馬達伺服驅動系統的動態分析，然後提出了非對稱歸屬函數之派翠機率模糊類神經網路用於即時鑑別內藏式永磁同步馬達伺服驅動系統的慣量。此外，還介紹了非對稱歸屬函數之派翠機率模糊類神經網路的網路結構和收斂性分析。根據實驗結果，可以在不同的操作條件下有效地線上調變IP速度控制器的增益。本論文亦發展一種基於小波濾波器的共振頻率偵測，以獲取轉子速度中共振頻率的特徵，並利用快速傅立葉轉換找出內藏式永磁同步馬達伺服驅動器的機械共振頻率。實驗所使用之硬體為應用德州儀器公司生產之浮點數數位訊號處理器TMS320F28075之內藏式永磁同步馬達驅動系統。;A real-time moment of inertia identification technique using Petri probabilistic fuzzy neural network with an asymmetric membership function (PPFNN-AMF) for an interior permanent magnet synchronous motor (IPMSM) servo drive is proposed in this thesis. The estimated moment of inertia will be used in the online design of an integral-proportional (IP) speed controller to achieve the gains auto-tuning of the IPMSM servo drive. In this thesis, first, the dynamic analysis of a field-oriented control (FOC) IPMSM servo drive system with an IP speed controller is studied. Then, a heuristic approach using the PPFNN-AMF is proposed for the real-time identification of the moment of inertia of the IPMSM servo drive system. Moreover, the network structure and the convergence analysis of the PPFNN-AMF are introduced. From the experimental results, the gains of the IP speed controller can be effectively tuned online at different operating conditions. Furthermore, a wavelet filter-based scheme for the detection of mechanical resonance frequency is also developed in this thesis. The wavelet Daubechies 4 (db4) is adopted to extract the features of the resonant frequencies embedded in the rotor speed. In addition, fast Fourier transform (FFT) is applied to perform the detection of the mechanical resonant frequencies for IPMSM servo drives. The experimentation using the IPMSM servo drive based on Texas Instruments′ floating point digital signal processor (DSP) TMS320F28075 carried out.
