| Since 80's of the last century, the research on deep foundation pit has been always a hotpot in architecture engineering in china. The design and construction of deep foundation excavation is not only an important key in infrastructure engineering of china's big city, but also a rather complex and difficult problem in rock and soil mechanics subject. As a systems engineering problem and synthesis technology, it has relations to many factors, for example, geological reconnaissance, bracing schemes and structure design, construction excavation, construction organization, climate change and ground environment, etc. The study of deep foundation pit is imperfect in theory, it is necessary to develop this synthesis subject. For instance, choice of bracing schemes rationally, perfecting stability theory, the prediction and control of stability in construction course, the construction organization planning and its optimization, these are worth while to lucubrate.Aiming at some leading problems of deep foundation pit, this paper attempts to present a systematic study, ranging from theory, methodology and its applications, and that is the main results as follows:1. Based on lots of national and international research results, the existent problems of the optimum seeking on schemes, the measure of factual stability, the prediction and control of stability in construction course, the comprehensive optimization of construction network planning and the uncertainty of network planning, among design and construction, are summarized, thus, the direction and content of the paper are advanced.2. This paper systematically analyzes main factors affecting the decision of bracing system on deep foundation pit, establishes an index system of evaluation scheme and a multiobjective fuzzy optimum model of the bracing schemes through fuzzy mathematical theory, and puts forward rational methods of the definiteness of objective weight and quantitating qualitative objects. A case study shows that the decision method is effective and easy to operate.(1) It is emphasized that the optimum design of the bracing schemes of deep foundation excavation should mainly consider technology, cost, environment and duration.(2) There is a fuzzy character in most multiobjective decision problems, multiobjective fuzzy optimum seeking model can solve factual matters at a certain extent, which is a profitable discuss of scientific and systematical decision on schemes design.(3) It is always a difficult problem to quantitate qualitative objects, set occurrence statistics is a kind of development to classic statistics and fuzzy statistics, by using the theory of set occurrence statistics, the distortion of quantitated results and random error in course can be reduced, which can quantitate qualitative objects more impersonally and rationally.(4) Rational definiteness of objective weight directly influence the accuracy of results of multiobjective decision, there is obvious different results by using different approaches, the subjective and objective integrated approach to determine attribute weight can make use of the subjective information provided by decision maker and the objective information to overcome the shortages which occurs in either a subjective approach or an objective approach.3. In order to overcome the shortages of general methods, this paper adopts the BP neural network based on toolboxes of MatLab to solve optimum seeking decision for bracing schemes of deep foundation pit, and presents a new linear equational method to select samples in the multiobjective decision model of artificial neural network, meanwhile, the optimum design of BP neural network is described wholly.4. Fuzziness and randomness are two main uncertainties in the study of stability on deep foundation pit, randomness implies kinds of design parameters being random variables, and fuzziness means that the judgement criterion of stability should be regarded as a vague criterion, which has inexplicit outline, it is in accord with the fact of failure process. Chapter four emphasizes to study the fuzziness of stability analysis of deep foundation pit, by adopting fuzzy sets and probability theory, a generalized model of fuzzy-random has been put forward to evaluate the stability degree, in thismodel, the random character of mechanical parameters and the fuzziness of failure criterion, is considered. And it also discusses the failure probability standard of deep foundation pit. An example is given to calculate the fuzzy failure probability of soil uplift in the bottom of pit, the result indicates that the fuzzy-random method can reflect the actual stability of the foundation pit roundly and rationally.5. Based on the BP neural network method in chapter three, the author attempts to apply the method to the prediction of deformation in construction course, thereby the prediction model of deep foundation pit can be established through using artificial neural network, the model may synthesize the distortion values from different survey points, then it also considers the distortion harmony of different survey points at a certain extent. The theory of the model is advanced and reasonable.6. It is doubtless that a physical available construction network plan should be optimized, particularly for deep foundation fit engineering, because of its investment scale greatness, long duration, great resource demand, high level quantity and high degree of construction risk, these characters bring the higher request to network planning of deep foundation pit construction, a rational, optimized and comprehensive or systemic network schedule can guarantee that the deep foundation pit construction being safe, reliable and saving. On the basis of the time-cost optimization of network planning, chapter five regards completing time as a random variable, through calculating the completing probability of kinds of compressed duration under the condition of specified time, the plans holding higher completing probability are selected as feasible projects, meanwhile, through synthesizing time, cost and quality as three most important judgement objects to optimum decision, by using of improved analytic hierarchy process method, the optimum decision of network planning may become actual, and this chapter also presents a optimum model of resource balance in the condition of limited time.(1) The relations among time, cost and quality are dialectic, but the most important one is quality, the aim of multiobjective optimum research on network planning is to integrate the three objects rationally.(2) These uncertain factors cause the uncertainty of the activity time, like politics,economy, weather, hydrology, construction projects, resource supply and construction environment etc, as a result, the presented problem is how to calculate this kind of time parameters and completing probability under the condition of specified time. PERT method can solve various random problems of uncertain network plan, which is based on probability theory and mathematical statistics.(3) Multiresource balance optimization is always a difficulty in network calculation, the method of maximum modulus with ideal solution is an effective method to solve it.(4) It is easy and simple to estimate the level of engineering quantity by using the linear time-cost model.(5) By employing the improved analytic hierarchy process with self-regulating function in this chapter, the relative good and bad degree of eigenvalue on each scheme to a certain object can be quickly, reasonably determined, then it is easy to determine the ranking of each project when the weight coefficients of objects are clear.7. The PERT model in chapter five considers the uncertainty of the activity time in network plan as randomicity, it is also an usual way currently. But the model of PERT is defective, the author discusses the defections of PERT model on theory and practice in chapter six firstly, the aim is about to explain that there are lots of faults about the model itself, on the other hand, shows why it is incorrect for uncertain problem if only the randomicity being considered. After using L-R fuzzy number to describe activity time, the model of identifying fuzzy critical path is found, and the algorithm of fuzzy time parameters is presented. It is more convenient to adopt L-R triangle fuzzy number than common fuzzy number when calculating, particularly while operating to the correction of fuzzy subtraction, it is easier than common three points fuzzy number. Finally, this chapter presented that it is effective and easy to identify the critical path through ranking method of fuzzy number. Moreover, a algorithm is put forward to calculate completing probability under the condition of specified time in this chapter.(1) The researches show that the average duration of PERT network is almost always less than the average values gained from the factual statistics, the error is about30%. It shows that there are shortages of both theory and method for the random model because that the calculational result is disaccord with the factual. Therefore, the author establishes a fuzzy network planning based on the fuzzy uncertainty, the fuzzy model can synthesize fuzzy, random and unknowable uncertainty.(2) It is not reversible between classic fuzzy subtraction and its addition, thereby, the fuzzy subtraction has to be modified in fuzzy network, it is easy for L-R fuzzy number to make the forward and backward calculation reversible.(3) Whether determining the critical path validly affects the realization of project aim and benefit straightly, through calculating the critical degree of every path to identify the critical path, is s common method, but it is not easy to be realized in great network because of the vast calculation, therefore, the author put forward ranking method of fuzzy number to identify the critical path, a case study shows that it is simple but effective method.
|
PDF Full Text Request