Stress Identification Of Space Structure Based On Multi-type Sensors

Structural health monitoring not only can provide real-time monitoringinformation about the performance and status of the structure, but also canachieve lots of targets, such as maintaining the structures, improving the safetyof the structures, verifying the structural design assumptions, reducing structuraluncertainty and having more awareness of the structures and so on. To monitorthe stress distribution is principle content for structural health monitoring syst em,which has the ability of evaluating the stress level and carrying capacity ofStructural elements directly. Due to a sufficient number of stresses monitoringpoints cannot be arranged as considering the cost and installation technology ofthe structural health monitoring system. So the number of the strain sensorswhich are located on the structure is limited. Meanwhile, there are different typesof sensor installed in a real structural health monitoring system. Therefore, anovel stress identification method using the measurements from multi-typesensors and information fusion in data level is proposed.Main research contents are:(1) Stress identification method based on sole type of sensor consideringuncertainties. Firstly, the input and output of the pattern recognition areconstructed by the measurements from sole type of sensor and the stress valueson the locations free of measurement devices, respectively, while the patternlibrary, the recognition library and the measurements library can be decided.Secondly, considering uncertain factors such as the noise or fault of the sensormeasurements and the error of the finite eleme nt model, the pattern matchingmethod considering uncertainties is proposed. Finally, the dynamic analysis of ashell structure under wind load is simulated, while the effectiveness and therobustness of the proposed method are proofed by comparing the ide ntified errorswith different level of noises in measurements.(2) Stress identification method based on multi-type of sensors andinformation fusion in data level. Firstly, taking the stress identification methodbased on sole type of sensor considering uncertainties, the pattern library and therecognition library are constructed by fusing the measurements of multi-type of sensors. Secondly, the feasibility and effectiveness of the proposed method isproofed by the simulations on a shell structure. Finally, in order to compare thestress identification methods using sole type of sensor and multi-type of sensors,the parameter analysis are carried out, including the number and locations ofsensors, noise levels, and the number of matching patterns and so o n. The stressidentification method using multi-type of sensors is much better, especially onthe noises resistance and engineering applications.(3) Stress identification based on the measurements from structural healthmonitoring system of Shenzhen Bay Stadium. Firstly, the simulation onShenzhen Bay Stadium underground pulsation is given, with the optimalselections on the number and locations of the sensors, the identified stress valuesbased on sole-type of sensor and multi-type of sensors are compared anddiscussed. Secondly, to change the measurements library by using the data fromsimulation on Shenzhen Bay Stadium under seismic actions and themeasurements from structural health monitoring system of Shenzhen BayStadium in three monitoring years, respectively, while the errors of the identifiedstress values are discussed. The method is verified in the actual stressidentification of Shenzhen Bay Stadium and it is known that the proposed stressidentification method is feasible and effective for a real large space steelstructure.