Sensor system is one of the most important parts of the health monitoring system of bridge structures. Even if much more information of structures can be captured theoretically by using much more sensors on structures, it is impossible and no need to locate sensors on all degrees in structures because of the high expenditure. It is the principal of optimization approaches for sensor placement to get adequate information by using adequate sensors. Several usual sensor placement algorithms are studied deeply firstly in this thesis, and the advantages and disadvantages of these algorithms are analyzed. Secondly, Genetic Algorithm (GA) is introduced, and each algorithm of sensor placement is realized by Genetic Algorithm. On the base of the previous analysis, nonlinear optimization problems are categorized into two kinds, non-stochastic problem and stochastic problem, and non-stochastic problems can be solved well by rank-based methods while the latter kind can be solved well by stochastic methods such as GA method. In the last part of Chapter 3, five criteria's on sensor system are introduced, and the characters of the problem of sensor placement are researched, also. In Chapter 4 and 5, General Genetic Algorithm (GGA) and Niche Genetic Algorithm (NGA) are introduced respectively to design the sensor placement on girders of actual suspension bridge structures. By varying the major three of the variations, orthogonal experimentation with three variations are designed to calculate the optimization placement of sensors on non-symmetry suspension bridges with GGA and NGA. The differences of the two GA are studied in the last part of Chapter 5.
|