Online Self-Sufficient Fuzzy Brain Emotional Learning Based Adaptive Intelligent Controllers For Semi-Active Vibration Control Of Smart Civil Structures

Civil engineering structures,e.g.,tall buildings,long-span bridges,offshore deepwater platforms,nuclear plants,and others,are becoming highly complex and nonlinear and are highly vulnerable to environment loadings such as earthquakes and strong winds.The development of smart structures is the most recent invention to handle this difficulty and decrease the structural response caused by natural forces.The aim is to create a new generation of civil structures with sensors,control devices,and control algorithms that can react in ”real-time” to an earthquake.Control algorithms are the most critical aspects of successfully controlling civil structures subjected to earthquakes.These algorithms are the brain of a smart structure,and their performance highly depends on these algorithms.These algorithms must be highly intelligent,adaptive,and robust to changes to sense the response of the smart civil structures,decide the counteraction,and alter the behavior of the structure in real-time.In recent years,adaptive intelligent control(AIC)algorithms are emerging as an acceptable substitute method to conventional model-based control(MBC)algorithms.These algorithms mainly work on the principles of artificial intelligence(AI)and soft computing(SC)methods that make them highly efficient in controlling highly nonlinear,time-varying,and timedelayed complex civil structures.One of the most recent developments,known as the Brain Emotional Learning Based Intelligent Controller(BELBIC),has caught the attention of scientists as a model-free adaptive control system.It possesses appealing capabilities for dealing with nonlinearities and uncertainties in control frameworks.The modern semi-actively controlled civil structures have a highly uncertain and nonlinear nature following severe earthquake disturbances.As a result,these structures require real-time(online)robust control actions towards changing conditions.This dissertation consists of three parts,One theoretical and two technical parts.In Part I,the detailed state of the art review of these emerging AIC methods has been presented for the case of smart civil structure.This theoretical contribution helps formulate and identify the current trends and the open problem present in the current literature.Based on these investigations,the following two parts of the study are formulated to fill in the research gaps and improve the problem present in the identified literature.The open problem states that the brain emotional learning-based controller for smart civil structure vibration control is highly neglected.Only a few applications based on the offline tuned setting of the BEL model were evidenced.These settings produce rigid control gains of fixed and crisp nature but lack a real-time decision-making system.During the tuning and learning phase,these settings get trapped in local optima also have poor learning capabilities.This dissertation's main objective is to help formulates online self-sufficient BEL-based AIC's that can operate independently using their inbuilt decision-making system and have high intelligence,adaptability,and learnability related to the nonlinear and uncertain dynamics of semi-actively controlled smart building structures operating under earthquake forces.This objective is achieved by combining traditional MBC algorithms and AI and SC-based intelligent systems;hence,two real-time BEL-based AIC methods have been contributed in parts II and III of this dissertation.Part,?,intends to formulate a real-time ”online” BELBIC based controller in two ways: an online self-tuning brain emotional learning-based intelligent controller(ST-BELBIC)is formulated.Then its capabilities in improving the performance of cascaded controller in attenuating seismic vibrations of a scaled threestory building structure are validated.In this case,the central control unit BELBIC is based on sensory inputs(SI)and emotional cues(reward)signals.The main contribution of the proposed controller is a self-attuned version of the standard BELBIC that uses the benefits of a first-order Sugeno fuzzy inference system(FIS)to adapt its parameters online.The proposed control methodology can be a promising model-free controller in terms of online tuning,simplicity of configuration,ease of applicability,less operational time,and neutralizing nonlinearities.The simulation affirms that the proposed controller compared with conventional LQR and intelligent Fuzzy tuned PID(FT-PID)controllers shows a superior performance regarding attenuating seismic responses of the building and can also improve the performance of cascaded FT-PID controller.Part,?,intends to formulate a novel tuning method for the optimum design of a standard Brain Emotional Learning(BEL)controller for smart scaled three story building structures.The principal contribution of the proposed control scheme is the development of an online self-evolving fuzzy BELBIC(OSEFBELBIC)that learns an inference(decision-making)system by itself.The proposed control scheme benefits equally offline and online tuning methods.The online self-attuned routines Takagi-Sugeno-Kang(TSK)Fuzzy Inference System(FIS)and the offline FIS tuning function by the evolutionary genetic algorithm(GA).In this case,the central control unit BELBIC is based on sensory inputs(SI)and emotional cues(reward)signals.Besides,a design methodology is also introduced into the reward signal that combines the classical PID and evolutionary fuzzy logic controller.The proposed control methodology can be a promising model-free AIC in terms of both online and offline BELBIC tuning for the response of each floor in parallel to neutralize nonlinearities.The simulation confirms that the proposed OSEF-BELBIC,compared with offline tuned fixed and crisp-valued Genetic PID-BELBIC(GPID-BELBIC),has shown superior performance regarding attenuating seismic responses of the building and also shows high learning abilities.Future work of these presented self-sufficient BEL-based AIC techniques can equally be applied to active control.Moreover,these techniques can be extended for the more complex and high degree of freedom systems.