Contexts in source publication

Context 1

... and threshold parameters of the best fit were determined as 0.69, 179.2 and -0.289, respectively. The maxi- mum p value among 14 theoretical distributions was used to select the best fit. Trend and autocorrolation of the dataset were evaluated using the Mann-Kendall trend test and the correlograms of Figs. 2 and 3, respectively. As depicted in Fig. 4, there is no trend in the TBF dataset. In the Mann-Kendall trend test, the S statistic used for the test and its variance is given ...

Context 2

... temporal diagram of TBFs is expected to show a de- creasing trend due to fatigue, aging and the mechanical tear and wear of different components. However, in this study, only the TBFs for one year of shovel operation were stud- ied. Thus, no noticeable decreasing trend was seen on inspec- tion in the diagram of Fig. ...

Context 3

... to transfer data into a stationary series. One or two orders of differenc- ing are typically enough to prepare data for the method. The combined autoregressive-moving average model in this case, that is, ARMA(p,q), is referred to as ARIMA(p,d,q), in which parameter d is the differencing order. A seasonal pattern occurs in the data illustrated in Fig. 4. The applica- tion of ARIMA in seasonal data needs further differencing in the seasonal portion. In this case, the model is known as seasonal autoregressive integrated moving average (SARI- MA), and it is represented by SARIMA(p,d,q)(P,D,Q) S , with a seasonal differencing order of D and a cycle of S. The P and Q represent the ...

Context 4

... seasonality. The AIC was used to select the SARIMA model's best p, q, P and Q as SARIMA(3,2,2)(1,1,0) 3 . In Table 3, AR(1) is the first coef- ficient of the autoregressive component of the model. Simi- Table 3. The result of the optimal SARIMA in both modeling and predicting the training and testing datasets, respectively, are illustrated in Fig. 4. Of 55 TBF records, 45, or 82 percent, were used for model training (building) and the remainder, or 18 percent, for model testing. The mean absolute percent- age error (MAPE) value for the testing dataset was calcu- lated as 44.6 percent. The scattergrams of the SARIMA and gamma predictions versus the actual observations for TBFs are ...