Study On Multi-objective Sensor Placement Optimization Method Based On Intelligent Algorithm

The damage in civil engineering structure under service may have adverse effects on people’s life safety,wealth safety and social stability,so structural health monitoring system construction is an urgent issue.The sensor placement is the first step of structural health monitoring system construction,and the current sensor placement method can be divided into time domain method and frequency domain method.Many sensor placement criteria,that are based on single objective,have been proposed.Arranging the sensor placement should consider multiple objectives,which can contribute more useful and accurate information for structural health monitoring system.There are millions of locations that are available for sensor placement in a real structure,so the sensor placement optimization,which finds out suitable locations for sensors,is necessary.The development of intelligent algorithm reduces the difficulty of sensor location searching work.Therefore,this thesis apply intelligent algorithm to search the sensor location,so the proposed method can consider multiple objectives of sensor placement criteria.The main contents of this thesis are as follows:(1)The monkey algorithm is improved to find out the optimized sensor placement among many of candidate locations,and the improved algorithm show stable accuracy and high efficiency in sensor location searching considering the time domain sensor index and frequency domain MAC criterion separately.Four different methods are used to search the optimal position of acceleration sensor,and the truss structure is used as an example to verify the effect of monkey swarm algorithm.(2)Based on Pareto theory,the time domain sensor index,the frequency domain MAC criterion and the number of the sensors are considered together as multiple objective,and the multiple type sensor placement optimization method under multiple objectives is proposed,and a beam structure is adopted to verify the proposed method.This proposed method provides the monitored structure a sensor placement,which can improve the identification accuracy of the health monitoring system and avoid the subjective mistakes,so it will be practical.(3)The improved monkey algorithm is combined with BP neural network,which can improving the efficiency of sensor location searching,and the multiple type multiple objective sensor placement optimization method is proposed.The neural network monkey algorithm inputs the sensor position searched by the monkey algorithm and outputs the predicted value of the objective function of the neural network instead of the traditional direct calculation of the objective function value.At the same time,the influence of input layer number on neural network performance is considered,A frame structure is adopted to verify the proposed method.