PSI - Issue 50

I. Shardakov et al. / Procedia Structural Integrity 50 (2023) 257–265 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

264

8

a

b

2011 2012 2013 2014 2015 2016 2017 -10 -8 -6 -4 -2 0 Date δUz, mm Real_data Line_interp Line_extrap ARIMA(1,1,1)(0,1,0) ARIMA(1,1,1)(0,1,1)

2011 2012 2013 2014 2015 2016 2017 -10 -8 -6 -4 -2 0 Date δUz, mm Real_data Line_interp Line_extrap ARIMA(1,1,1)(0,1,0) ARIMA(1,1,1)(0,1,1)

Fig. 7. Foundation settlements with seasonal periodicity: ARIMA model forecast for 1-year (a) and 2-year (b).

Table 2 gives an assessment of the quality of forecasting for various types of sedimentary processes. It shows the values of the standard deviation between the predicted and observed values of settlements for 1-year and 2-year periods. As can be seen from the table, the largest forecast errors are observed if the process with an increasing rate and processes with seasonal changes are carried out. Table 2. Root-mean-square deviation in forecasting foundation settlements according to the ARIMA (1,1,1)(0,1,0) model for different types of sedimentary processes Process Type Settlement at a constant rate Settlement at increasing rate Settlement at decreasing rate Seasonal cycle RSME (mm) in 1 year forecast 0.160 0.92 0.16 0.62 RSME (mm) in 2year forecast 0.127 1.7 0.29 0.5 4. Conclusions Completed research demonstrates the possibilities of using statistical forecasting models to predict the settlement of foundations of building structures. The capabilities of several well-known models were tested on a large number of time series. These data are provided by an automated monitoring system for a large group of buildings located in an undermined area. The ETS-Holt-Winters and STL-ETS(A,N) and ARIMA (p, d, q) (P, D, Q) models were tested. The performed numerical experiments showed that the most adequate results were obtained using the ARIMA model. It predicts well the development of soil settlement processes with acceleration and deceleration, as well as with seasonal changes. In this paper, we discuss in detail the results obtained using the this model. Comparison of the prediction results with the data of real observations showed that for most of the considered objects, the proposed approach gives satisfactory results. The maximum difference between the observed and predicted values in terms of the standard deviation for a two-year period was no more than 1.7 mm. Such a value of the forecast error is comparable with the error of the hydraulic leveling measuring system. This allows us to conclude that the considered algorithm for predicting the settlement of individual points of the structure can be used in the analytical unit of the automated system for monitoring the deformation processes of buildings and engineering structures. The predicted values of the settlement of various points of the building foundation can be used to calculate the stress strain state of the building as a whole. Thus, based on the data of long-term observation of the settlement of the characteristic points of the foundation, it is possible to assess the state of the structure in the future, including predicting whether the development of hazardous processes in the structure is expected and predicting their localization.