| 摘要: | 本國中小型營造業自1950年代開始逐漸式微，萬丈高樓從此開始從平地拔起，而勞工安全卻隨著樓層高度，被拋之其後，而其中最大的原因，來自於勞工安全設備被資方認為是隱形成本，若無人員傷亡，就會被視為浪費，導致常常被營造商忽略，從根本來看，若可以有效降低添購及維繫的成本、操作的難易度，以及讓其更容易為人攜帶，則能夠大幅提升營造業對勞工安全設施添購的意願。因此本研究開發一種基於加速度計的跌倒預兆檢測系統，該系統採用基於閾值的檢測方式，由一個小型的自包裝傳感器單元及傳輸控制協定(TCP, Transmission Control Protocol)組成該設備。傳感器單元包括一個三軸加速度計+陀螺儀模組(MPU9250)，並由一套開發版（ESP8266）所組成。本研究設計15組模擬情境實驗，分別為7組跌倒情境及8組日常動作情境，來評估所提出系統的性能，而該系統分別安裝於受試者的頭部、腰部、腳部。結果顯示，雖然加速度計的判別都趨於精準，但頭部及腳部較容易誤將日常生活運動判定為跌倒，而在安裝在腰部的系統則可以得到較高的準確率，因此加速度計閾值設定之門檻為0.25 g。另外本研究採用歐拉角作為第二門檻判釋，減少誤報機率。基於上述成果，建築工人可以方便的攜帶此裝置以獲得三軸的加速度及歐拉角，將獲得的數據計算後傳輸至雲端監控端，依據歷史資料庫之特徵值設定閾值檢測並識別出跌倒或墜落。如果檢測到並識別出跌倒，將傳送訊號至手機，手機將發出警報並自動呼叫緊急聯繫人以進行及時救援。或可透過LORA低功耗長距離無線模式直接傳送訊息至工務所，啟動救援行動。;Since the 1950s, the small and medium-sized construction industry in Taiwan has gradually declined, the tall buildings have been uprooted from the flat ground, but labor safety has been ignored in the construction. The biggest reason is that the labor safety equipment is considered as an invisible cost by the management. If there are no casualties, it will be regarded as a waste; therefore, the construction industry often ignores it. If it can effectively reduce the cost of purchase, maintenance, and operation difficulties, besides, it is easier for people to carry, it can significantly enhance the willingness of the construction industry to purchase labor safety facilities. However, it still exists many limitations in the existing research and corresponding devices for detecting falls. Therefore, this research develops a fall detection system based on an accelerometer. The system adopts the detection method based on a threshold composed of a small self-packaged sensor unit and Transmission Control Protocol (TCP). The sensor unit includes a three-axis accelerometers + gyroscope module (MPU 9250) and a development version (ESP8266). This study designed 15 groups of simulated situation experiments, including seven fall situations and eight daily activity situations, to evaluate the performance of the proposed system, which was installed on the subjects’ heads, waist, and feet, respectively. The results show that although the accelerometer′s discrimination tends to be accurate, it is easier for the head and feet to judge the movement of daily life as falling mistakenly, and higher accuracy can be obtained in the system installed at the waist. Based on the experiment results, the threshold of the accelerometer threshold is revised as 0.25g. In addition, this study uses Euler angles as the second threshold to reduce the false alarm. Conclusively, construction workers can easily carry this device to obtain the triaxial acceleration and Euler angles, calculate and transmit the data to the cloud-monitoring terminal, set the threshold according to the characteristic values of the historical database then identify falls. If a fall is detected and recognized, a signal will be sent to the mobile phone, which will give an alarm and automatically call the emergency contact for a timely rescue. Alternatively, it can send messages directly to the public works station through LORA low-power long-distance wireless mode to start rescue operations. |
