A reinforcement learning-based boundary optimal control algorithm for parabolic distributed parameter systems is developed in this article. First, a spatial Riccati-like equation and an integral optimal controller are...
详细信息
A reinforcement learning-based boundary optimal control algorithm for parabolic distributed parameter systems is developed in this article. First, a spatial Riccati-like equation and an integral optimal controller are derived in infinite-time horizon based on the principle of the variational method, which avoids the complex semigroups and operator theories. Using state data along the system trajectory, a value iteration algorithm via the Bellman optimality principle is proposed to obtain the solution of the spatial Riccati-like equation and the optimal control law. The convergence of the value iteration algorithm is proved. Subsequently, an approximation scheme based on weighted residuals is developed to implement the value iteration algorithm, where radial basis functions are chosen as the basic functions to approximate the solution of the spatial Riccati-like equation. Simulations on the diffusion-reaction process demonstrate the effectiveness of the developed method.
This paper investigates the composition and identification of ancient glass artefacts. First, the collected data were pre-processed to determine the chemical composition content of the glass artefacts before weatherin...
详细信息
ISBN:
(纸本)9798350321050
This paper investigates the composition and identification of ancient glass artefacts. First, the collected data were pre-processed to determine the chemical composition content of the glass artefacts before weathering. Then, based on the variable control method, the effects of decoration and colour of two types of glass, high potassium glass and lead-barium glass, on the weathering of artefacts were determined. The statistical law of the presence or absence of weathering chemical composition on the surface of artefact samples was summarized by establishing a weathering change rate solution model. Meanwhile, based on the weathering change rate, the solution model for the content of chemical substances before weathering is summarized, and the content of different chemical components in ancient glass artefacts before weathering is predicted. Next, the second problem is carried out, which is to classify the glass and select the appropriate chemical composition for subclassification. Ancient glass artefacts were classified into the high-potassium glass and lead-barium glass based on their chemical composition and content, and the classification rules were analyzed. On this basis, the top eight indicators of chemical composition differences between the two types of artefacts were determined by comparison and subclassified by the K-means algorithm. Finally, model experiment results verified the correctness and validity of the models proposed in this paper.
An important task in the identification community for time-varying systems is to track the possible changes in system dynamics as well as possible. For the currently applied identification techniques, this task is usu...
详细信息
ISBN:
(纸本)9798350321050
An important task in the identification community for time-varying systems is to track the possible changes in system dynamics as well as possible. For the currently applied identification techniques, this task is usually implemented at the cost of increasing the computational quantity as the time increases, or of adding a forgetting factor that requires to be determined a priori. In this paper we develop an estimation approach for linear time-varying systems with additive disturbances, which achieves a major computational advantage without determining the additional factor. In particular, we integrate the idea behind Savitaky-Golay filtering into the kernel-based regularization method under the framework of regularized least squares. It is found that the developed approach remains the same computational quantity during the whole time interval in the estimating procedure. A numerical example is simulated to validate the effectiveness of the developed approach.
Heating, Ventilation and Air Conditioning (HVAC) system is a highly nonlinear system with a large amount of complex, coupled inputs. This paper presents a novel prediction method for HVAC energy consumption based on d...
详细信息
ISBN:
(纸本)9798350321050
Heating, Ventilation and Air Conditioning (HVAC) system is a highly nonlinear system with a large amount of complex, coupled inputs. This paper presents a novel prediction method for HVAC energy consumption based on deep neural network (DNN). In order to solve the problem that traditional neural networks tend to fall into local optima, batch normalization and Adam optimization algorithm are significantly incorporated in the DNN. In addition, particle swarm optimization (PSO) is utilized to search for the optimal number of hidden layer nodes and increase the accuracy of prediction model. The cooling tower data of HVAC is used to validate the network. The results show the mean absolute error and the mean square error of the PSO-DNN model, from which it can be seen obviously that our proposed prediction model performs better than traditional ones. Accordingly, DNN optimized by PSO algorithm is of great validity and superiorities for energy consumption prediction of HVAC system.
Due to the existence of strong nonlinearity and external disturbances, the controller design of complex nonlinear systems is a challenging problem. Therefore, it is necessary to design an effective robust predictive c...
详细信息
Due to the existence of strong nonlinearity and external disturbances, the controller design of complex nonlinear systems is a challenging problem. Therefore, it is necessary to design an effective robust predictive controller for this issue. In this article, based on a fuzzy neural network, an iterative learning model predictive control (FNN-ILMPC) is designed for complex nonlinear systems. First, a dynamic linearization technique is used to establish a data-driven model, which only relies on input and output data. Since the established model contains an unknown disturbance term that may have an impact on the control performance, an FNN is used to evaluate the disturbance so that the uncertainty of the system is captured. Subsequently, based on the above data-driven model, an FNN-ILMPC strategy, considering the impact of external disturbances, is developed to eliminate the influence of disturbances. Then, it is proved that the designed controller can make both modeling error and tracking error decrease gradually and ensure the closed-loop system stability. Finally, the experimental results verify the effectiveness and superiority of the designed controller.
The fault diagnosis model based on machine learning can only achieve accurate recognition of the fault types included in the training, but in practical applications, it is limited by the classification mechanism of th...
详细信息
ISBN:
(纸本)9798350321050
The fault diagnosis model based on machine learning can only achieve accurate recognition of the fault types included in the training, but in practical applications, it is limited by the classification mechanism of the diagnosis model and cannot achieve the recognition of new faults, which is unknown faults. To address this problem, this paper proposes a fault identification method for rotating machinery based on the Convolutional Neural Networks (CNN) and the Deep Convolutional Generative Adversarial Network (DCGAN) to identify unknown faults of rotating machinery. The method first trained CNN with each known class data to build the initial diagnosis model, and trained DCGAN to obtain the discriminative network of each known class to build the confidence probability calculation model, and then the results of the initial diagnosis were corrected according to the confidence probability, and finally the intelligent diagnosis of the known and unknown class faults was realized. The analysis results of the centrifugal pump fault simulation showed that the proposed method achieved an average diagnostic accuracy of 96.16% and 91.79% for known and unknown faults, respectively.
Traffic congestion is a major source of discomfort and economic losses in urban environments. Recently, the proliferation of traffic detectors and the advances in algorithms to efficiently process data have enabled ta...
详细信息
Traffic congestion is a major source of discomfort and economic losses in urban environments. Recently, the proliferation of traffic detectors and the advances in algorithms to efficiently process data have enabled taking a data-driven approach to mitigate congestion. In this context, this work proposes a reinforcement learning (RL) based distributed control scheme that exploits cooperation among intersections. Specifically, a RL controller is synthesized, which manipulates traffic signals using information from neighboring intersections in the form of an embedding obtained from a traffic prediction application. Simulation results using SUMO show that the proposed scheme outperforms classical techniques in terms of waiting time and other key performance indices.
Navigation and control near asteroids are challenging due to high levels of uncertainty, necessitating a high degree of autonomy and safety. To address this challenge, we propose a novel method for leveraging data-dri...
详细信息
Navigation and control near asteroids are challenging due to high levels of uncertainty, necessitating a high degree of autonomy and safety. To address this challenge, we propose a novel method for leveraging data-driven approaches, which have garnered significant attention in recent years. Traditional data-driven and machine learning techniques have been applied to space exploration problems;however, they often require large amounts of data in advance and are not robust to unknown environments. Hence, they are typically used in auxiliary roles and are not easily adaptable to real-time modeling and control. We propose the use of the theory of dynamic mode decomposition (DMD) and delayed embedding to construct a data-driven real-time guidance and control system for asteroid landings. Our approach employs low-dimensional information obtained from onboard sensors to construct nonlinear dynamics in the vicinity of asteroids. Based on the constructed dynamical model, we propose a guidance control strategy for accurate and reliable spacecraft landing. Our proposed method has the potential to improve guidance and controlsystems for space exploration by enabling real-time modeling and control with reduced computational cost.
Distinguishability and, by extension, observability are key properties of dynamical systems. Establishing these properties is challenging, especially when no analytical model is available and they are to be inferred d...
详细信息
Distinguishability and, by extension, observability are key properties of dynamical systems. Establishing these properties is challenging, especially when no analytical model is available and they are to be inferred directly from measurement data. The presence of noise further complicates this analysis, as standard notions of distinguishability are tailored to deterministic systems. In this article, we build on distributional distinguishability, which extends the deterministic notion by comparing distributions of outputs of stochastic systems. We first show that both concepts are equivalent for a class of systems that includes linear systems. We then present a method to assess and quantify distributional distinguishability from output data. Specifically, our quantification measures how much data is required to tell apart two initial states, inducing a continuous spectrum of distinguishability. We propose a statistical test to determine a threshold above which two states can be considered distinguishable with high confidence. We illustrate these tools by computing distinguishability maps over the state space in simulation, then leverage the test to compare sensor configurations on hardware.
The complexity and uncertainty in power systems cause great challenges to controlling power *** a popular data-driven technique,deep reinforcement learning(DRL)attracts attention in the control of power ***,DRL has so...
详细信息
The complexity and uncertainty in power systems cause great challenges to controlling power *** a popular data-driven technique,deep reinforcement learning(DRL)attracts attention in the control of power ***,DRL has some inherent drawbacks in terms of data efficiency and *** paper presents a novel hierarchical task planning(HTP)approach,bridging planning and DRL,to the task of power line flow ***,we introduce a threelevel task hierarchy to model the task and model the sequence of task units on each level as a task planning-Markov decision processes(TP-MDPs).Second,we model the task as a sequential decision-making problem and introduce a higher planner and a lower planner in HTP to handle different levels of task *** addition,we introduce a two-layer knowledge graph that can update dynamically during the planning procedure to assist *** results conducted on the ieee 118-bus and ieee 300-bus systems demonstrate our HTP approach outperforms proximal policy optimization,a state-of-the-art deep reinforcement learning(DRL)approach,improving efficiency by 26.16%and 6.86%on both systems.
暂无评论