本文提供以修正的最小平方法來估計三參數的對數常態分配的參數,三參數即最低故障時間參數、尺度參數及形狀參數。一般常用的最大概似估計量(maximumlikelihood estimator)雖是最有效的估計量,然對三參數的估算卻是較麻煩的,本文提供一個較為簡易的方法來估計參數,而其方法經模擬比較與概似估計量在有效性與一致性(consistence)均是相似的。 This investigation provides a modified least square method to estimate three parameters, i.e. threshold parameter, scale parameter, and shape parameter, of the lognormal distribution. MLE (Maximum Likelihood Estimator) is most often used and most efficient for the estimation of these three estimators. However, the estimation is quite complicated in three-parameter lognormal distribution. This article will demonstrate by simulation results the proposed modified least square
method has similar efficiency and consistency to MLE, but is much simpler and easier to understand.