以形狀辨識為基礎的診斷訓練輔助對時間壓力下診斷之影響

Abstract

急迫時間的壓力使複雜系統附加於作業人員的負荷過大。許多研究發現人員在時間壓力之下，資訊處理的能力變得很脆弱。為了輔助作業人員在急迫時間壓力下的作業，一些附註診斷的法則型專家系統因應而生，但是專家系統的運作常緩不濟急。因此另一改善方法可從訓練著手。本研究提出以辨識情境為主的訓練方法，此訓練方法利用整合型顯示將症狀之主要線索呈現出來，教導人員如何辨識主要線索型態，以促進其情境與症狀知配對，而增加其診斷績效。 此實驗中所模擬的系統為汽車冷卻系統的監控作業，監控作業中顯示器的型態分為物體顯示、生態顯示、數位顯示。受試者必須在限定的時間內針對三種不同顯示型態診斷出螢幕中所呈現的系統變數，判斷系統目前的狀態，最後再利用實驗設計與問卷做分析。實驗結果中顯示在組件層次問題、系統層次問題下反應時間和錯誤率的關係為，都以物體顯示最佳。但是在系統層級錯誤率的結果上卻是物體顯示優於生態顯示優於數位顯示，而生態顯示經由訓練後績效有明顯的增加。

Complex systems often require operators to respond in a timely fashion. Previous studies found that human information processing capabilities are quite vulnerable when people are under time stress. One way to enhance operator’s fault diagnosis performance is to provide expert system based diagnosis aids. However, these aids were found to be not as useful as expected. This study proposed a training method which allowed operators learn the pairing between observed data and symptoms of malfunction. By using ecological display, learning of mapping between the patterns of information displays and malfunctions could be facilitated. The results of this study demonstrated that operators could quickly match the observed data with the template of malfunction states and their diagnosis performance can be enhanced. We measure their diagnose time, correct rate by object , ecological ,and digital display in car cooling system.The experimental results show that object display appear better effects in time stress.