Abstract: This paper discusses the control algorithm of structural vibration control in civil engineering. It points out that the research on intelligent control algorithms has become an important part of structural vibration control in civil engineering, and its development trend is prospected. Keywords: structural vibration control; civil engineering; control algorithm; intelligent control 1 Introduction The large-scale collapse and damage of buildings during earthquakes poses a huge threat to people's lives and property. Due to the uncertainty of the external load of earthquake excitation, traditional earthquake-resistant structural systems that rely on increasing the strength and deformation capacity of the building structure itself cannot guarantee the safety of the building. Therefore , researching safer, more economical, and reliable structural systems has become an important topic in the field of earthquake-resistant engineering structures. In 1972, Chinese-American scholar Yao Zhiping first proposed the concept of vibration control of civil engineering structures based on classical and modern control theories[1], which is a milestone in the study of structural vibration. Structural vibration control has long been widely used in the fields of machinery, aerospace and shipbuilding. It is an emerging discipline in the field of civil engineering. The study of structural vibration control includes two parts: one is the study of control devices, and the other is the study of control algorithms. Among them, the control algorithm is an important part of structural vibration control. This paper discusses the control algorithm of structural vibration control in civil engineering and looks forward to its development trend. 2 Concept of structural vibration control and vibration reduction mechanism of civil engineering Structural vibration control refers to setting some control devices in a certain part of the structure. When the structure vibrates, a set of control forces or changes the dynamic characteristics of the structure are applied passively or actively to reduce the vibration response of the structure and make it meet higher safety and functional requirements. It is an emerging comprehensive discipline and a high-tech field in civil engineering. Based on this definition, the vibration reduction mechanism of structural vibration control can be simply explained by a structural dynamic equation: (1) Where M, C, K: the mass, damping and stiffness matrix of the structure; I: unit column vector; F(t): external action (including earthquake, wind or other external forces that may be applied) vector; : the acceleration, velocity and displacement response vector of the structure under external action (or load); : the seismic acceleration response vector of the ground. Structural vibration control is to greatly reduce the response of the structure under seismic action by adjusting the natural frequency ω or natural period T of the structure (by changing K, M) or increasing the damping C, or applying external force F(t). Let the allowable structural acceleration, velocity and displacement response values ​​be the values ​​that ensure the safety of the building structure and the people, equipment and decoration facilities in it. Then, solving equation (1) only needs to satisfy: (2) to ensure the safety of the people, equipment and decoration facilities in the structure. Structural vibration control can be divided into passive control, active control and semi-active control according to whether external energy input is required[2]. Among them, semi-active control is the second stage of active control development. The vibration reduction effect of active control and semi-active control is better than that of passive control, and its performance depends to a large extent on the selected control algorithm. Therefore, the control algorithm is the core issue of structural vibration control. 3 Traditional structural control algorithms Structural control algorithms originate from modern control theory, but these theories have generated a series of special problems in the application of civil engineering structures, which need to be further studied and solved. In principle, all control algorithms of modern control theory can be used for active structural control. However, due to the special nature of civil engineering structures, some algorithms can be directly applied, while some algorithms need to be specially processed. Structural control algorithms generally adopt modern control theory based on time domain analysis, especially the commonly used state space method as the main research method. The control of multivariable linear systems and nonlinear systems uses modern mathematical methods as the main analysis means and computers as the main implementation means. So far, the traditional algorithms used for vibration control of civil engineering structures mainly include: (1) Classical linear optimal control algorithm [3] The classical linear optimal control algorithm is based on modern control theory and uses the quadratic performance of the control vector and the state vector as the objective function to determine the relationship between the control force and the state vector or external excitation. The objective function uses a weight matrix to coordinate the relationship between economy and safety. The algorithm requires solving the Riccati equation. Since the algorithm ignores the external load term, strictly speaking, the control obtained by it is not optimal control. However, numerical analysis and limited experiments have shown that although this control algorithm is not optimal, it is feasible and effective. The quality of this algorithm depends to a large extent on the choice of the weight matrix, requiring the designer to weigh the control effect and the required control energy, which is also a disadvantage. (2) Instantaneous optimal control algorithm [3] If the objective function is made to reach the optimal in every small time interval, it is called instantaneous optimal control. The instantaneous optimal control algorithm uses the quadratic form of the instantaneous state and the control force vector as the objective function. Within the time range of the dynamic load, the objective function is minimized at each instant. The algorithm does not require solving the Riccati equation, reducing the amount of computation; the gain matrix is ​​uniquely determined by the characteristics of the controlled structure, and the robustness of the control system is good; it has the property of time stepping and can be extended to nonlinear time-varying structural systems. However, this algorithm is only a local optimal control algorithm. From the perspective of whole process control, it is not optimal control. (3) Pole placement method The system matrix determines the dynamic characteristics of the system. The real and imaginary parts of its eigenvalues ​​give the modal damping and modal frequency of the system, respectively. By selecting an appropriate gain matrix, the dynamic characteristics of the closed-loop system can achieve the design value that meets the designer's requirements. This is the pole placement method. The pole placement method can be easily implemented when only a few modes of vibration that have a significant impact on the structural response are considered. In this method, the selection of the gain matrix is ​​the inverse problem of solving the eigenvalues. The selected gain matrix is ​​usually not unique and must be determined by the designer's experience. Therefore, the control law obtained by the pole placement method is not optimal, but the algorithm is relatively simple and easy to implement. (4) Adaptive control algorithm The research object of adaptive control is mainly a control system with a certain degree of uncertainty. This uncertainty can come from inside the system or from outside the system. Adaptive control is also a method that relies on mathematical models. The difference is that adaptive control has less prior knowledge about the mathematical model of the controlled object. It needs to continuously extract useful information as the control process progresses so that the model can be gradually improved. (5) Variable structure control [4] Variable structure control system refers to a system in which the parameters of the structure have uncertainty or time-varying characteristics. Since the 1990s, American scholars Yang et al. have taken the lead in introducing variable structure control methods into civil structure control. Based on the approach law method of variable structure control law design, the variable structure control method for building structure vibration was studied, showing that the method has a good control effect. The discrete approach law method of variable structure control system design is used to give the method of determining the system switching function. (6) Lyapunov direct method The second method of Lyapunov (direct method) is an effective method for analyzing the stability of nonlinear systems. Its most prominent feature is that it has complete adaptability to the time-varying nature of the system itself and external disturbances. The Lyapunov direct method has a clear physical meaning in structural control. The Lyapunov function represents the vibration energy of the structure. When the derivative of the vibration energy with respect to time is negative, the system is asymptotically stable, and the smaller the value, the faster the system tends to the stable equilibrium point. Therefore, maximizing the vibration energy decay rate of the Lyapunov asymptotically stable system can be used as an objective of structural control. Traditional structural control algorithms usually require the establishment of an accurate structural vibration model. Civil engineering structures are complex systems that are nonlinear, strongly coupled, multivariable, and uncertain, and have complex dynamic characteristics (including stressed structural members and unstressed non-structural members). When designing and calculating structures and modeling control, the effects of non-structural members are usually not considered. However, structural vibration control is mainly aimed at the actual structure after construction. The changes in non-structural members and mass have a great impact on the calculation model. In addition, the actual structure may enter nonlinearity under strong dynamic action such as earthquakes, and the strength and stiffness of structural members may degrade. The model correction of the actual structure will be a prominent problem in structural vibration control. Therefore, the study of intelligent control that does not rely on accurate calculation models and is easy to adjust is a hot topic in the development of structural vibration control. 4 Intelligent control algorithm Civil engineering structures are complex systems that are nonlinear, strongly coupled, multivariable, and uncertain. Intelligent control algorithms that do not rely on accurate calculation models can play an effective role and have become an important aspect of the development of structural control. Intelligent control is an emerging theory and technology [5]. It has the ability to effectively control complex systems globally and has strong fault tolerance. It also has the characteristics of non-mathematical generalized models represented by knowledge and hybrid control processes represented by mathematical models. Intelligent control also has learning, adaptation and organization functions. At present, the application research of intelligent control in the field of structural vibration is mainly focused on neural network control, fuzzy control and genetic algorithm. (1) Fuzzy control Fuzzy control does not rely on the precise calculation model of the structure or system. It mainly realizes the adjustment or control of the system through the fuzzy logic relationship between the state output and the control input, that is, the fuzzy control rules. It can be used for the control of complex systems such as nonlinear, time-varying and time-delay. In 1994, Goto and Yomada et al. studied the fuzzy control method of structural vibration and its membership function optimization problem. The results showed that the shape and parameters of the membership function have a great influence on the control effect. In 1998, Battani applied fuzzy control to study the ATM control of a three-layer frame Benchmark model. Michael et al. applied fuzzy algorithm to control the semi-active damper in the study of the semi-active isolation scheme of bridge structure and achieved good control effect. Wang Gang and Ou Jinping [6] proposed the fuzzy modeling and fuzzy control rule extraction method of structural vibration in response to the problem that there is a lack of manual operation control experience in structural vibration control. However, due to the low accuracy of fuzzy control and the lack of extensive operational experience or experimental data in civil engineering structural control, the determination of fuzzy control rules is somewhat blind. (2) Neural network control Artificial neural networks have strong nonlinear approximation, self-learning and self-adaptation, data fusion and parallel distributed processing capabilities, and have shown obvious advantages and application prospects in the identification, modeling and control of multivariable, strongly nonlinear and large time delay systems [7]. Applying neural networks to active structural control can effectively solve the time delay problem caused by the measurement and transmission of feedback signals. Neural networks were used in structural identification and control in the late 1980s [8]. Neural networks can model nonlinear structures and identify multi-degree-of-freedom systems with unknown parameters under seismic excitation. Bani-Hani and Ghaboussi et al. first used neural network controllers in system identification and active control experiments. Joghnataie and Ghaboussi proposed neural network identification and prediction of the dynamic response of structures, and used neural networks to learn the optimal control force. By improving the neural network controller with fuzzy rules, they reduced the amount that needed to be predicted and achieved good control results. In China, Zhang Shunbao et al. used BP neural networks to predict the structural state, providing a time difference for active structural control, thereby reducing the time delay effect in active structural control. (3) Genetic Algorithm Genetic Algorithm (Genetic Algorithm) [9] is an optimization algorithm based on natural genetics mechanism. It seeks the global optimal solution of the problem by using a random but directed search mechanism. Arranging active and semi-active control mechanisms on building structures is an effective method to control structural behavior. The problem of arranging control mechanisms on structures belongs to discrete variable optimization design problem. It often has the characteristics of nonlinear, non-convex or discontinuous design space. For this type of optimization design problem, it is difficult to find the global optimal solution using traditional mathematical programming methods, while genetic algorithm provides a new way to solve this type of problem. GA is a better global search optimization method. Using GA to solve the optimal arrangement problem of control mechanisms can obtain satisfactory results and has a fast convergence speed. 5 Conclusion Structural vibration control, as a brand-new and proactive structural countermeasure, is gradually combining with emerging control technology, information technology and new material technology, and developing towards automation and intelligence. Currently, research on intelligent control algorithms has become an important component of intelligent vibration control for civil engineering structures. While the most widely applied methods in civil engineering still rely on fundamental theories of intelligent control (such as fuzzy control, neural networks, and genetic algorithms), new ideas need to be incorporated into the algorithms themselves. Control algorithms applied to vibration control of civil engineering structures are constantly evolving, and each algorithm will be gradually improved based on the characteristics of the controlled structure and its dynamic loads. It is foreseeable that intelligent algorithms applied to the control of civil engineering structures will be a perfect combination of fuzzy control, neural network control, and genetic algorithms. References: [1] Peng Gang, Zhang Guodong. Vibration control of civil engineering structures [M]. Wuhan: Wuhan University of Technology Press, 2002: 1-152 [2] Run Weiming, Zhou Fulin, Tan Ping. Research progress on vibration control of civil engineering structures [J]. World Earthquake Engineering, 1997, 13 (2): 8-18 [3] Song T T. Active Structure Control. Theory and Practice. Longman, London, and Wiley, New York, 1990 [4] Yang JN, Wu JC, Agrawal A K. Sliding Mode Control for Nonlinear and Hysteretic Structures. ASCE, Journal of Engineering Mechanics, 1995, 121 (12): 1330-1339 [5] Wang Yaonan. Intelligent Control System [M]. Hunan: Hunan University Press, 1996 [6] Weng Jiansheng. Semi-active control of vehicle suspension system based on magnetorheological damper [D]. Nanjing University of Aeronautics and Astronautics, 2001 [7] He Yubin, Li Xinzong. Neural Network Control Technology and Its Applications [M]. Beijing: Science Press, 2000. [8] Wu Jianjun. Structural Seismic Response Control Based on Neural Networks [D]. Doctoral Dissertation, Tianjin University, 1998. [9] Sun Zengqi, Zhang Zaixing, Deng Zhidong. Intelligent Control Theory and Technology [M]. Beijing: Tsinghua University Press, 2002.