使用低解析度感測器之直流無刷電動機驅動器的新型可變取樣控制法

Abstract

由於直流無刷電動機的運轉效率等等優點，相關應用與日俱增，而通常使用霍爾元件來決定換相時間與量測運轉速度。由於原始設計目的在於換相，通常僅使用三個霍爾元件，因此速度回授時間因馬達轉速而異，亦即速度回授取樣時間是變動的，此一現象亦發生於一般直流無刷馬達的無位置感測器控制，吾人稱之為可變取樣系統。此一現象需要剛健性控制器來處理，本論文即針對此類使用低解析度的感測器的直流無刷馬達推導出兩種不同的剛健性控制器。 本論文首先提出一個剛健控制器的合成方法，藉由將系統的變取樣特性轉換為系統的不確定度，再利用相當普遍的Mu-合成程序計合成出控制器。正因為僅具低解析度感測器的直流無刷馬達回授的變取樣現象，可被視為系統不確定度，因而此現象可以線性分式轉換來描述，所以Mu-合成器可以提供控制器設計必要且足夠的條件。實驗結果驗證了原始設計規格並示範了此方法的成效。 不同於上述以系統不確定度發展控制器，本論文接著推導出在可變取樣率下的變動結構控制器設計方法。本研究改良了現有的離散時間控制器的逼近法則，並推針對直流無刷馬達導出可變取樣控制，完成準滑差控制法則。本控制方法業經模擬與實驗加以驗證。同時，本論文亦討論比較了數種控制器的各項優劣。 為了發展上述的控制器，可以支援變取樣模擬的工具不可或缺，本文設計了一個與速度相關的可變取樣擬工具。然後，由於實際上述霍爾元件安裝時位置誤差無可避免，因而引起速度量測誤差，數種抑制此類誤差的方法亦於本論文呈現。 最後，本論文提出針對三相四開關直流無刷電動機在無位置感測元件下新穎的控制方法，不同於傳統的六開關三相架構，此一方法之所以引人乃在於節省了兩個開關與霍爾感測器的成本。不像一般三相四開關結構，若非使用高階控制器來輔助只有四個向量的缺點，便是需要額外的電流感測來滿足六個向量控制所需，本論文的方法可以支援六個向量的無感測器控制，同時因為可以提供定電流所以轉矩亦可固定，實驗結果可以看出此方式效果非常好。

Brushless DC (BLDC) motors are gaining increased application because of their efficient operation, high-density characteristics, and easy maintenance. BLDC motors usually use Hall-effect sensors for commutation and rotor speed. Because the Hall sensor signals originally were designed for commutation purposes and the speed feedback is based on low-resolution sensors. Therefore, with or without a sensor, the time intervals between speed feedbacks depend on the motor speed. As a result, the speed feedback sampling time is variable, and is called a variable sampling system. Two robust variable sampling control methods for BLDC motors with low-resolution sensors were proposed and implemented in this dissertation. A method of synthesis of a robust controller was proposed. In this method, variable sampling rates are translated into system uncertainties, and then the popular Mu-synthesis procedure was used to synthesis the controller. The variable sampling feedback phenomenon of a BLDC motor with low-resolution sensors was treated as system uncertainty. Therefore, it is possible to describe the system with the standard linear fractional transformation. The Mu-synthesis can provide a necessary and sufficient condition for the controller design. Experimental results conformed to the original design specifications and have demonstrated the effectiveness of the proposed methods. The implementation skill of decomposition into fractional expansion from a complicated controller was also verified in experiments. As opposed to as the treatment of system uncertainty, another variable sampling VSC is also derived from the conventional discrete-time VSC in this dissertation. It uses a modification of the discrete-time variable-structure-control (VSC) reaching law and derives a variable sampling control law for a BLDC motor to achieve the quasi-sliding mode condition. Both simulation and experimental results have confirmed the effectiveness of the proposed control laws and estimator. A comparison of several controllers is also given in this dissertation. Further, a variable sampling simulation tool for BLDC motor drivers is proposed, it is very useful when developing a variable sampling controller. Then, mitigation methods of speed measurement error, which is caused by misalignment of Hall sensors, are also presented. Finally, a novel position sensorless control for four-switch three-phase (FSTP) BLDC motors is proposed. Different from the traditional six-switch three-phase (SSTP) scheme, an FSTP is a fancy solution because it eliminates the cost of two switches and Hall sensors. Unlike conventional FSTP structures, which are implemented in either four vectors with a high-performance MCU or six vectors with current sensors, the FSTP control method supports six vectors and position sensorless control. The control method provides constant current, which means constant torque. The experimental results show that the scheme works very well.