Research On Corrosion Detection Of Reinforced Concrete Based On Multi-sensor Information Fusion

Reinforced concrete corrosion is the main reason that affects the durability of building structures.Nowadays,the application of reinforced concrete is very common.Corrosion will bring a series of economic problems,resource problems and social problems.Therefore,it is particularly important to obtain the corrosion status of the steel bars in the building in time.Aiming at the current problems of single sensor for corrosion detection of reinforced concrete and low recognition accuracy of corrosion degree,this thesis uses a multi-sensor detection system to detect corrosion of reinforced concrete.The adaptive Kalman filter algorithm is used to reduce the noise of characteristic parameters,and the mathematical model of multi-sensor detection and recognition is established to realize the information fusion of reinforced concrete corrosion detection.Multi-sensor information fusion is a research method using sensor detection technology,data filtering technology and neural network recognition technology.It can monitor and evaluate the corrosion of reinforced concrete in time.The research work done in this thesis includes the following:(1)Rebar corrosion detection data filtering: The traditional Kalman filter algorithm is improved,measurement noise,system noise and state prediction covariance estimator are designed due to the most likely estimation criterion.This method realizes the dynamic adjustment of the estimation model and reduces the interference of sensor measurement noise and system noise in the detection system.(2)Recognition of steel corrosion degree: Fuzzy neural network mathematical model is established.Aiming at the problems that the traditional particle swarm optimization(PSO)is easy to converge early and the local optimization ability is poor in dealing with complex search problems,an adjustment method of the inertia factor in the PSO algorithm is proposed.The improved PSO algorithm optimizes the fuzzy neural network and gives the analysis of the position convergence and velocity convergence of the particles.The optimized neural network connection weights are obtained through the improved PSO algorithm,which improves the search speed and training efficiency of the algorithm,and reduces the recognition error of steel corrosion.(3)On this basis,this thesis designs a reinforced concrete corrosion detection system based on multi-sensor information fusion,which includes hardware design and software design.The hardware circuit design includes sensor signal acquisition circuit,analog-todigital conversion circuit and wireless communication circuit,etc.,this realizes data acquisition of steel corrosion parameters;The development of the software part of the detection system includes the design of the lower computer and the design of the upper computer.The multi-sensor information fusion algorithm is used to realize the real-time human-computer interaction of steel corrosion parameters.(4)The multi-sensor information fusion algorithm proposed in this thesis is verified by experiments.The experiment shows that the parameter measurement accuracy of the steel corrosion detection system is improved,and the prediction error of the steel corrosion degree identification model is reduced.Thus it verifies the feasibility of the multi-sensor information fusion algorithm.