A Research On Prediction Settlement Of Soft Soil Foundation Consolidated By Means Of Preloading

With the developing of Chinese economic, civil aviation industry has entered a new area, and at the same time aerodrome construction throughout the country has been in a rapid development stage. A number of provincial capitals, coastal open cities and special economic zones have carried out the constructions of the aerodrome, The aerodrome construction demands stronger solid foundations and pavements compared to that of the civil architectures and common highways, especially for post-construction settlement and differential settlement. However, the soft soil with poor engineering mechanic parameters widely distributes in coastal areas, the soft soil treatment in the construction of the aerodrome is not only the first problem to be solved, but also an extremely important issue. In the construction of the aerodrome, treatment finished time, effect, and post-construction treatment have to be calculated according to the datum of the field monitoring, and the current calculation method will inevitably be some faulted, so it is significant to apply a rational method on the basis of the monitor datum to forecast the settlement of soft soil.This paper is entitled by the national natural science foundation:"The role and influence of the organic in the process of marine soft soil settlement"(NO.40372122). Take the civil aerodrome construction of soft soil foundation disposal in Jieyang, Guangdong as an example, the settlement forecast of soft soil foundation is researched and analyzed.The aerodrome construction field is located between the southwest of Chaoshan plain and the middle reaches of Rongjiang River. The field area belongs to maritime lake sedimentary plain except the southwest which belongs low hill area. Maritime and terrestrial crossed phased soft soil deposits with thick layer and deep depth is wildly distributed. For this engineering, preloading method to deal with the field area of the soft soil is chosen according to preliminary detailed exploration and the experience in various ground treatments. At the same time, in order to conduct a reasonable and effective foundation treatment, the representative test area is selected for preloading before the construction of large area, and the design and construction of the whole field area are carried out on the basis of feedback, the amendment, verification of represent area result. Test areas are divided into 6 test cell, and each one is tested with different construction parameters according to the monitoring datum to gain the multiple field test and monitored settlement datum.The original monitoring datum of surface subsidence are different from that of verhulst and poisson curve models, for many discontinuous points indwell it, and in the process of soft soil consolidation, the monitored settlement value is influenced by rainfall and construction around which make the soft soil subsidence curve departing from the true value, but the extent of this influence is unknown. So this complexion reduces the accuracy of the prediction of various non-linear theories which make the prediction of actual soft soil settlement difficult. In this paper, wavelet theory is used to disposal the soft soil consolidation settlement value, i.e. extract relatively stable datum from the original datum to better predict the settlement without interference outside. When decomposing the original settlement value into high-frequency ones and low-frequency ones, the wavelet function of the theory is not unique, therefore, in this paper, db10 function of Daubechies wavelets, coif5 function of Coiflets wavelets, bior3.9 function of ReverseBior wavelets are used respectively. In addition, the wavelet functions include different decomposition levels, and different levels can not differentiate the close degree between the datum analyzed with the method of wavelet and the original one, so in this paper, different wavelet foundations and once, twice decomposition levels are used to analyze the original soft soil consolidation curve.As the prediction accuracy with the methods of forecasting for different time periods are not consistent. Usually, with the increase of forecast times, the error between forecast and the actual monitored datum will gradually increase. So the forecast time is divided into short-term forecast, medium-term forecast and long-term forecast (final settlement). The original monitored datum are transformed with the method of the wavelet model which different functions and decomposition levels are used in. Linear methods i. e. Subsequently exponential curve and hyperbolic methods, nolinear method i. e. GM (1,1) model of gray system theory and BP neural network model are used to predict on the basis of wavelet and the prediction accuracy between the calculated value based on above method and the monitored datum is compared respectively which provides a certain reference to research the soft soil settlement.In the paper, GM (1,1) model of gray system theory is applied to predict the settlement of the soft soil. This theory not only has a high degree of generality, but also high accuracy, so the theory is applied to predict the consolidation of the field consolidation. The settlement value and settlement difference value are predicted respectively, the result shows that prediction effect is poor based on the settlement value datum, for the settlement will increase ceaselessly which is counter to the mechanism of soft soil consolidation. Therefore, a suggest is proposed that not only the posterior deviation ratio and the small error probability need to be calculated but also the incremental approach of the consolidation curve i.e. the predicted datum sequence should be checked, for the incremental approach in fact reflects the trend of soft soil consolidation. So predict effect should be checked based on the above conditions. In order to resolve the problem that the predicted settlement value curve is not constringent, the settlement difference value is chosen to predict on the basis of gray system theory. It is concluded that the datum sequence predicted can be convergent, and the posterior deviation ratio, small error probability can all reach the degree of I. Through the calculation result, it is drawn out that short-term forecast error is larger than that by other methods and the medium-term forecast error is less. Usually, the predicted settlement value with the method of hyperbolic method is larger than the actual one, and the exponential method is smaller, and the value predicted with the method of GM (1,1) model of gray system theory is between the value by hyperbolic method and that by exponential method. It is concluded that GM (1,1) model of Gray system theory should be applied in the medium and long term prediction of the soft soil settlement.As the artificial neural network has a strong self-adaptability, associate capacity, self-learning ability, fuzzy reasoning ability, so neural network is very convenient to analyze the non-linear problems. Since the development of artificial neural networks, the theory plays an important role in the datum forecast, image processing and the fitting of the functions, so BP neural network model is applied in the prediction of the soft soil consolidation, and the result is analyzed. Through comparing the datum predicted by BP model and the actual datum, it is drawn out that it is feasible for BP neural network model to predict the settlement of soft soil. And the short-term settlement is high accurate that the error between the predict datum and the actual datum is less than 10%. With the increase of learning samples, the network's forecasting ability and generalization ability will be further improved. But the medium-term predict error is huge which is large than 100%, so it is poor to predict the data that is longer time later.In this paper, linear methods i. e. exponential curve and hyperbolic methods are used to predict the settlement of soft soil, and the approaches are explained. Traditional exponential curve and hyperbolic methods mainly aimed at the settlement value, and in the paper the settlement value is transformed into settlement difference value on the basis of formula transform which offered possibility to apply the wavelet theory into it. Since the extraction of the datum based on wavelet theory more stable, the linear analysis method will become more reliable.Comparing the results calculated based on the four methods above, in terms of short-term predict, the accuracy based on BP neural network model is best, in terms of medium-term predict, the accuracy based on gray system theory GM(1,1) model is best, and in terms of long-term predict i. e. final settlement, the value based on hyperbolic method is maximum which is larger than the actual one according to literature and practical engineering experience, the value based on exponential curve method is minimum which is smaller, and that based on gray system theory GM(1,1) model and BP neural network model is between the value above, so the result is more close to the actual value. The calculating result provide reference for settlement calculating method of different time stages in field engineering. According to the result, the calculated accuracy based on one-layer decomposition and bior3.9 is highest which provide a chosen method in the process of wavelet de-noising in the field.Finally, the consolidation degree is used to validate the effects of soft soil ground treatment, the final settlement value based on the above 4 method is synthetically considered, and the last monitored data is chosen to calculate the consolidation degree in different test cell. According to the result of the final consolidation degree, the degree of SD1 region is highest, so the optimal construction parameters are in follows: the vertical drain is plastic drainage plate, the spacing of piles is 1m, the length of piles is 20m, which provide a reference for foundation treatment design and construction in the next stage of the whole aerodrome.