應用邊緣計算模組於智能化工具機參數鑑別之研究;Intelligent CNC Parameter Identification Using Edge Computing Box

[[abstract]]工業4.0、智慧機械與人工智慧等議題，其基礎皆為大量資料收集。於工具機領域資訊主要分為兩類:加工訊號以及感測器訊號。加工訊號主要可由數值控制器獲得，獲得位置命令、位置回授、扭矩命令、伺服速度、主軸轉速、主軸扭矩等…機台運動狀態。感測器訊號則非常多元，也具有多家量測儀器公司提供感測器及訊號擷取卡，以振動抑制為例:振動訊號常用美國國家儀器推出之DAQ訊號擷取卡搭配加速規，多個加速規黏貼於機台各個位置，機台激振、走特定行程方式了解機台動態特性。此應用情境即須個別使用各家儀器廠商所提供之軟體搭配控制器訊號取得量測資料，最後將感測器與控制器資料彙整於科學計算軟體中如:MATLAB。本論文將水平整合類比訊號感測器、熱電偶以及FANUC控制器加工訊號於單一軟體中。使用邊緣計算模組串聯周邊硬體設備，由底層硬體、通訊、作業系統、作業系統核心至應用程式高度垂直整合。針對機械領域高精高解析度的採樣需求，修改並調整作業系統核心達成一定程度軟實時特性。軟體撰寫多執行緒程式，使軟體效能最佳化，提供使用者即時監控以及高精高解析度兩種模式。邊緣計算模組中具備整合型人機操作介面軟體，提供完善的使用者體驗，以此軟體與高精高解析度採樣技術為基底，整合實驗室先前研究技術成果「工具機參數鑑別」作為加值應用程式，此應用程式運行特定加工路徑，運行的同時採樣機台運動資訊，分析收集資料，提供使用者機台物理特性，乃至於機台調機基礎。此加值應用程式更為智慧機械應用範例，邊緣計算模組可作為資料彙整核心，以此平台為基礎載具，平台平價親民，且富有彈性與高度可擴充性，結合多項智能化技術，促使工具機各機台皆具備智能化應用，並為下一世代機械具備人工智慧作為前瞻性之研究。 關鍵字：智慧機械、系統整合、FANUC、邊緣計算模組 Industry 4.0, Smart Machine and AI lot of issues are based on data collection. Those issues have to collect the correct data as much as possible. The information in machine tools can separate into two categories: the signal of machining and the signal of sensor. First, the signal of machining can sample from CNC Controller. The signals what we can get are machine’s position command, position feedback, torque command, servo speed, spindle speed, and spindle torque lots of machine kinematic status. Second, many kinds of sensors provided by the companies can achieve high sensitivity with DAQ. For example, suppression of mechanical vibration needs both accelerometers’ signal and the signal of machining. Normally, one can use National Instruments’ DAQ and accelerometers in pairs. Multiple accelerometers stick on machine. One can run specific path or using machine dither, with simultaneously sampling the accelerometers’ signal and controller information. Finally, the sampled data is then imported to MATLAB to calculate and identify machine kinematic characteristic. In suppressing mechanical vibration case, one has to use lots of software include MATLAB. In this paper, the sensor and controller are integrated together. The analog sensors, thermocouples, and FANUC CNC controller are measured by a single software. Using “Edge computing Box” can connect both sensors and controller. The hardware, communication, operating system to OS kernel are highly integrated. Because the sampling frequency in mechanical system needs high resolution, the OS kernel should be rewritten to make it has basic “soft-realtime” property. Reduce the lost of samples. In software development, using multi-thread improve the software efficacy. Provide two modes for user: online watching and offline high resolution sample modes.Edge computing box has human machine interface (HMI) software which provides a perfect user experience. Based on offline high resolution sample mode, this paper also integrate some technologies of our LAB’s which includes the inertia and friction estimation. Users can use the information to determine better controller parameters such that the error of machining are reduced. This is one of the examples for smart machine. Edge computing box can also be a center of database. This platform has highly cost-effective and highly expandability. It is suitable to integrate smart machine technologies. Let all of the CNC have “smart” ability. This is a forward-looking study for next generation of machinery – “artificial intelligence machine”.Keywords：Smart Machine、System Integrate、FANUC、Edge Computing Bo