Reinforcement Learning-based IoT Sensor Scheduling Strategy For Bridge Structure Health Monitoring

Internet of Things(IoT)based Bridge Structural Health Monitoring(BSHM)is a hot topic in the field of civil engineering and computer science and has received a lot of attention from academia and industry.The lifetime of sensors is much smaller than the lifetime of bridges,which is one of the main technical bottlenecks facing BSHM sensing acquisition.Therefore,how to effectively improve the network lifetime of bridge structural health monitoring systems is the focus of current research.Continuous rigid frame bridges have been extremely widely used in China in recent years.With the passage of time,influenced by external factors such as concrete material characteristics,construction environment,and construction process,some diseases such as internal cracking of concrete box girders,rusting and exposure of reinforcement,deflection of main girders,decrease of prestressing,and reduction of strength and stiffness stability of continuous rigid frame bridges have caused great hazards to the structural safety use of rigid frame bridges,even causing loss of life and property.Therefore,it is necessary to develop a bridge structural health monitoring system and adopt efficient technical means to maintain and overhaul these continuous rigid bridges under construction or in operation,so that the structural integrity of the bridge can be continuously assessed and potential economic or human losses can be avoided for the structural safety of the bridge.This paper fully considers the demand of civil engineering for bridge structural health monitoring and the requirements of computer science for sensor network connectivity and coverage,takes the Guangdong Beijiang Bridge continuous rigid bridge as an example,introduces reinforcement learning and credible information coverage model based on Fisher information matrix,and investigates how to dispatch as few sensor nodes as possible to obtain the required bridge structural information,while ensuring network connectivity and coverage requirements.The study investigates how to maximize the service life of the bridge structural health monitoring system while ensuring the network connectivity and coverage requirements,as well as the accuracy and irrelevance of the observable modalities in the sensor arrangement.The main research work and results accomplished are summarized as follows:(1)In this study,ANSYS was used to model the solids of the Beijiang Bridge in Guangdong,China,to collect the required structural data and construct an algorithmic model.(2)A sensor node dormancy scheduling strategy is proposed.The node dormancy scheduling problem is defined as a multi-objective optimization problem with information validity as the modal confidence criterion,and the network performance is measured by combining node energy consumption,network coverage requirements and network connectivity.The network connectivity and coverage requirements are ensured while minimizing the system parameter identification error,thus maximizing the network lifetime of the bridge health monitoring system.(3)Taking the Beijiang Bridge as the research object,the finite element node location is used as the initial deployable location of sensor nodes.Through a series of simulation experiments,the influence of a series of factors such as coverage,monitoring demand and the number of deployed nodes on the lifetime of the sensor network of the bridge health monitoring system and the absolute value of the determinant of the Fisher information matrix is evaluated,which verifies the feasibility and effectiveness of the method and has a broader engineering application prospect.