機器學習技術應用於轉速表辨識系統

機器學習是人工智慧發展中的一部分。近年來機器學習已發展成為是一門橫跨多領域的學科，亦即以機器學習來解決人工智慧中的一些問題。本篇論文提出「機械學習技術應用於轉速表辨識系統」。其中的轉速表，就是一個可以量測並顯示轉速的儀器，量測的結果是以指針或數位的方式顯示每分鐘的轉速(RPM)，指針偏離不同的角度就代表不同的轉速。汽車或機車上的轉速表，主要的用途是用來指示引擎曲軸的轉速。一般的轉速表都有標示引擎轉速的安全範圍，目的就是要幫助駕駛調整油門及排檔到較佳的駕駛狀態。假若引擎在轉速安全範圍外運轉過久，有可能會導致潤滑不足而出現過熱的現象，而產生額外的磨損與造成引擎永久的損壞。為了解決此困擾，本篇論文提出轉速表的辨識系統。辨識過程簡述如下：(1)首先是以影像辨識轉速表的引擎轉速；(2)接著是以辨識演算法判斷引擎轉速是否超出安全範圍；(3)若是超出安全範圍則發出警報。本篇論文所使用的硬體，包含：樹莓派PI4等。軟體是以樹莓派的內建作業系統IPython為機器學習的核心，經IPython寫入辨識程式至轉速表的辨識系統。經由上述機器學習方式的轉速表辨識系統，當引擎之轉速超出安全範圍時能立即的發出警報，並藉此以避免造成引擎的損壞及保護駕駛的安全。

關鍵字

機器學習；人工智慧；轉速表

並列摘要

Machine learning is part of artificial intelligence. In recent years, machine learning technology was widely used to solve some problems in artificial intelligence. This paper employs mechanical learning technology to identify tachometer. Tachometer is an instrument that can measure and display the speed. The measurement is with the speed per minute (RPM) in the form of a pointer or digits. The deviation of the pointer from different angles represents different speeds. The purpose of the tachometer on the cars or motorcycles is to detect the rotating speed of the engine. In general, tachometers have a safety range that indicates the engine speed. Its purpose is to help the driver adjust the throttle and shift to a better driving state. If the engine runs too long outside the safe range of speed, it may cause insufficient lubrication and overheating, which may cause additional wear and permanent damage to the engine. To overcome this problem, this paper proposes a tachometer identification system. The identification process is briefly described as follows: (1) First, the engine speed of the tachometer is identified by the image; (2) Then, the identification algorithm is used to determine whether the engine speed is outside the safe range; (3) If it exceeds the safe range, an alarm is issued. The hardware used in this study includes: Raspberry Pi PI4. The software is based on the Raspberry Pi's built-in operating system IPython as the core of machine learning, and the identification program is written to the tachometer identification system via IPython. Through the tachometer identification system by the proposed machine learning method, when the engine speed exceeds the safe range, an alarm can be immediately issued to avoid damage to the engine and protect driving safety.