Priority-based Task Scheduling in Dynamic Environments for Cognitive MFR via Transfer DRL

In this paper, we address the difficulty of solving large-scale multi-dimensional knapsack instances (MKP), presenting a novel deep reinforcement learning (DRL) framework. In this DRL framework, we train different agents compatible with a discrete action space for sequential decision-making while still satisfying any resource constraint of the MKP. This novel framework incorporates the decision variable values in the 2D DRL where the agent is responsible for assigning a value of 1 or 0 to each of the variables. To the best of our knowledge, this is the first DRL model of its kind in which a 2D environment is formulated, and an element of the DRL solution matrix represents an item of the MKP. Our framework is configured to solve MKP instances of different dimensions and distributions. We propose a K-means approach to obtain an initial feasible solution that is used to train the DRL agent. We train four different agents in our framework and present the results comparing each of them with the CPLEX commercial solver. The results show that our agents can learn and generalize over instances with different sizes and distributions. Our DRL framework shows that it can solve medium-sized instances at least 45 times faster in CPU solution time and at least 10 times faster for large instances, with a maximum solution gap of 0.28% compared to the performance of CPLEX. Furthermore, at least 95% of the items are predicted in line with the CPLEX solution. Computations with DRL also provide a better optimality gap with respect to state-of-the-art approaches.

Non-terrestrial network (NTN) communication is included in the 3GPP standard because of its excellent features such as resistance to ground physical attacks and wide coverage. Since the available power and storage resources are limited on high throughput satellites (HTSs), optimizing the resource allocation can greatly improve the performance of the HTS based communication system. Notably, weather and other factors make the channel status between the satellite and the terrestrial base stations constantly change. In this paper, we first exploit the model-free feature of reinforcement learning to formulate the aforementioned dynamic and unpredictable channel conditions into the power allocation problem in HTS systems. Due to the complexity of environment state and power allocation, a deep neural network is introduced to replace the Q table, and a deep reinforcement learning framework is built. Furthermore, a power allocation algorithm based on a deep reinforcement learning framework is presented. Finally, the simulation results show that the proposed algorithm is superior to existing algorithms in terms of the long-term system throughput performance, and has better adaptability to dynamic channel environment, thereby improving network performance.

Unmanned aerial vehicle–human collaboration route planning for intelligent infrastructure inspection

Effective ways to optimise real-time pump scheduling to maximise energy efficiency are being sought to meet the challenges in the energy market. However, the considerable number of evaluations of popular optimisation methods based on metaheuristics cause significant delays for real-time pump scheduling, and the simplification of traditional deterministic methods may introduce bias towards the optimal solutions. To address these limitations, an exploration-enhanced deep reinforcement learning (DRL) framework is proposed to address real-time pump scheduling problems in water distribution systems. The experimental results indicate that E-PPO can learn suboptimal scheduling policies for various demand distributions and can control the application time to 0.42 s by transferring the online computation-intensive optimisation task offline. Furthermore, a form of penalty of the tank level was found that can reduce energy costs by up to 11.14% without sacrificing the water level in the long term. Following the DRL framework, the proposed method makes it possible to schedule pumps in a more agile way as a timely response to changing water demand while still controlling the energy cost and level of tanks.

Optimal scheduling of island integrated energy systems considering multi-uncertainties and hydrothermal simultaneous transmission: A deep reinforcement learning approach

With the rapid development of smart grid and renewable energy, demand response mechanism has become a key means to balance microgrid supply and demand and improve energy efficiency. As the traditional power demand response algorithm is difficult to deal with the uncertainty of power demand and the adverse effects caused by customers, this paper proposes a power demand response algorithm based on deep reinforcement learning. Firstly, we study and design a price -based layered electricity energy demand response model. Then, the dynamic optimal pricing decision of power trading market is described as Markov decision process, and the learning mechanism under the framework of deep reinforcement learning is expounded. Finally, a solution algorithm based on deep reinforcement learning is designed. The simulation results show that the proposed algorithm can adjust the electricity price adaptively according to the load demand, which can reduce the cost of users by about 17% and reduce the peak load demand.

We present a hierarchical deep reinforcement learning (DRL) framework with prominent sampling efficiency and sim-to-real transfer ability for fast and safe navigation: the low-level DRL policy enables the robot to move towards the target position and keep a safe distance to obstacles simultaneously; the high-level DRL policy is supplemented to further enhance the navigation safety. We select a waypoint located on the path from the robot to the ultimate goal as the sub-goal to reduce the state space and avoid sparse reward. Moreover, the path is generated based on either a local or a global map, which can significantly improve the sampling efficiency, safety, and generalization ability of the proposed DRL framework. Additionally, a target-directed representation for the action space can be derived based on the sub-goal to improve the motion efficiency and reduce the action space. In order to demonstrate the eminent sampling efficiency, motion performance, obstacle avoidance, and generalization ability of the proposed framework, we implement sufficient comparisons with the non-learning navigation methods and DRL-based baselines, with videos, data, code, and other supplemental material shown on our website<xref ref-type="fn" rid="fn1">¹</xref><fn id="fn1"><label>¹</label> https://github.com/zw199502/RL <inline-formula><tex-math notation="LaTeX">$\_$</tex-math></inline-formula> navigation </fn>.

Manually designing an effective and efficient controller for complex mechanics, such as bio-inspired robots or underactuated mechanical system, typically are very difficult. It requires precise motion planning and dynamic control. Reinforcement learning or genetic algorithm based learning methods suffers from representing the high dimensional models. The combination of deep learning and reinforcement learning provide a feasible way to handle such difficulties. However, priori-less searching sometimes tends to be low efficient and usually finds the “mechanic” solution instead of the “natural” one. In this paper, the traditional nonlinear control concept is integrated into the deep reinforcement learning (DRL) framework. The whole process is implemented on the brachiation control problem of a two link bio-primate robot. Deep Deterministic Policy Gradient (DDPG) is used to search for the optimal control policy. The searching process is realized by interacting with the dynamic model instead of real robot. The energy based planning and control concept is adopted, which utilize the fact that when the shoulder joint angle is fixed, energy of the whole system keeps constant. By regulating the angle and energy, the robot can be restricted on a particular trajectory. The energy concept is encoded within the reward function and trained in the Gym environment. For varying targets point-to-point control, the network structure is also modified to accept the target coordinates. Effectiveness of the proposed methods are verified by simulation and experimental results.

This paper proposes a deep reinforcement learning (DRL) framework that learns a reusable policy to find the optimal placement of decoupling capacitors (decaps) on power distribution network (PDN) with an arbitrary probing port. The proposed DRL framework trains a policy parameterized by pointer network, which is a sequence-to-sequence neural network, based on REINFORCE algorithm. The policy finds the positional combination of a pre-defined number of decaps that best suppresses self-impedance of a given probing port on PDN with randomly assigned keep-out regions. Verification was done by allocating 20 decaps on ten randomly generated test sets with an arbitrary probing port and randomly selected keep-out regions. Performance of the policy generated by the proposed DRL framework was evaluated based on the magnitude of probing port self-impedance suppression followed by decap placement over 434 frequencies between 100MHz and 20GHz. The policy generated by the proposed framework achieves greater impedance suppression with fewer samples in comparison to random search heuristic method.

A multi-objective deep reinforcement learning framework

With the booming development of logistics, manufacturing and warehousing fields, the autonomous navigation and intelligent obstacle avoidance technology of automated guided vehicles (AGVs) has become the focus of scientific research. In this paper, an enhanced deep reinforcement learning (DRL) framework is proposed, aiming to empower AGVs with the ability of autonomous navigation and obstacle avoidance in the unknown and variable complex environment. To address the problems of time‐consuming training and limited generalisation ability of traditional DRL, we refine the twin delayed deep deterministic policy gradient algorithm by integrating adaptive noise attenuation and dynamic delayed updating, optimising both training efficiency and model robustness. In order to further strengthen the AGV's ability to perceive and respond to changes of a dynamic environment, we introduce a distance‐based obstacle penalty term in the designed composite reward function, which ensures that the AGV is capable of predicting and avoiding obstacles effectively in dynamic scenarios. Experiments indicate that the AGV model trained by this algorithm presents excellent autonomous navigation capability in both static and dynamic environments, with a high task completion rate, stable and reliable operation, which fully proves the high efficiency and robustness of this method and its practical value.

Many of the strongest game playing programs use a combination of Monte Carlo tree search (MCTS) and deep neural networks (DNN), where the DNNs are used as policy or value evaluators. Given a limited budget, such as online playing or during the self-play phase of AlphaZero (AZ) training, a balance needs to be reached between accurate state estimation and more MCTS simulations, both of which are critical for a strong game playing agent. Typically, larger DNNs are better at generalization and accurate evaluation, while smaller DNNs are less costly, and therefore can lead to more MCTS simulations and bigger search trees with the same budget. This paper introduces a new method called the multiple policy value MCTS (MPV-MCTS), which combines multiple policy value neural networks (PV-NNs) of various sizes to retain advantages of each network, where two PV-NNs f_S and f_L are used in this paper. We show through experiments on the game NoGo that a combined f_S and f_L MPV-MCTS outperforms single PV-NN with policy value MCTS, called PV-MCTS. Additionally, MPV-MCTS also outperforms PV-MCTS for AZ training.

Learning ability is the basic characteristic of human intelligence. The July 1, 2005 issue of Science published a list of 125 important questions in science. Among them, the question 94 “What are the limits of learning by machines?”. The annotation “Computers can already beat the world’s best chess players, and they have a wealth of information on the Web to draw on. But abstract reasoning is still beyond any machine”. In recent artificial intelligence has made great progresses. In 1997, the rise of the man-machine war, IBM Supercomputer Deep Blue defeated the chess master Garry Kasparov. On February 14, 2011, IBM’s Watson supercomputer won a practice round against Jeopardy champions Ken Jennings and Brad Rutter. In March 2016, Google DeepMind’s AlphaGo sealed a 4-1 victory over a South Korean Go grandmaster Lee Se-dol. This paper focuses on the machine learning methods of AlphaGo, including reinforcement learning, deep learning, deep reinforcement learning, analysis of the existing problems and the latest research progress. Deep reinforcement learning is the combination of deep learning and reinforcement learning, which can realize the learning algorithm from the perception to action. Simply said, this is the same as human behavior, input sensing information such as vision, and then, direct output action through the deep neural network. Deep reinforcement learning has the potential to learn a variety of skills for the robot to achieve full autonomy. Even though reinforcement learning is practiced successfully, but feature states need to manually set, for complex scene is a difficult thing, especially easy to cause the dimension disaster, and expression is not good. In 2010, Sascha Lange and Martin Riedmiller proposed deep auto-encoder neural networks in reinforcement learning to extract feature, which is used to control the visual correlation. In 2013, DeepMind proposed deep Q-network (DQN) in NIPS 2013, using convolution neural network to extract features, and then applied in reinforcement learning. They continue to improve and published an improved version of DQN on Nature in 2015, which has aroused widespread concern. In order to break through the limits of learning by machines, cognitive machine learning is proposed, which is the combination of machine learning and brain cognition, so that the machine intelligence is constantly evolving, and gradually reaches the human level of artificial intelligence. A cognitive model entitled Consciousness And Memory (CAM) is proposed by author, which consists of memory, consciousness, high-level cognitive functions, perception and motor. High-level cognitive functions of the brain include learning, language, thinking, decision making, emotion, and so on. Learning is a course to accept the stimulus through the nervous system and obtain new behavior, habits and accumulation experience. According to the current research progress of brain science and cognitive science, cognitive machine learning may be interested in learning emergence, procedural memory knowledge learning, learning evolution and so on. For intelligence, so-called evolution is refers to the learning of learning and the structure also follows the change. It is important to record the learning result by structure changing and improve the learning method.

In coupled power distribution and transportation (CPT) system, a joint scheduling framework for mobile energy storage systems (MESSs) and Volt/VAR control (VVC) ensures reliable power distribution grid operations while supporting electric vehicle loads at electric vehicle charging stations (EVCSs). However, conventional model-based optimization methods for MESS scheduling and VVC may yield suboptimal solutions and greater computation times because of MESS operation and VVC in uncertain environment of CPT systems. To resolve this issue, this study proposes a model-free deep reinforcement learning (DRL) framework. In this framework, smart inverters of MESSs and solar photovoltaic (PV) systems cooperate to minimize the real power loss and mitigate the violations of both MESSs’ state of charge (SOC) and voltage in the power distribution network, while MESSs travel via the transportation network to satisfy EV loads at EVCSs. A MESS routing algorithm based on Dijkstra’s algorithm is developed to determine the optimal destinations of the MESSs. In addition, two safety modules are developed to ensure that neither SOC nor voltage violations occur by adjusting real and/or reactive power of MESSs and PV systems during the training process. The developed MESS routing algorithm and safety modules are integrated into the proposed DRL framework, wherein the DRL agent performs the desired MESS scheduling and VVC through safe exploration during the training procedure. The proposed approach is tested in coupled IEEE 33-bus power distribution and 15-node transportation systems and coupled IEEE 57-bus power distribution and 42-node transportation systems. Numerical examples demonstrate the advantages of the proposed approach in terms of training convergence, real power loss, and SOC/voltage violation.

To address the static limitations of current career assessment methods and the lack of a long-term perspective in planning strategies, this study constructs an intelligent analysis framework named Deep learning-driven Value-added Career Analytics Model (DV-CAM). Integrating Deep Learning (DL) with value-added assessment, the framework aims to realize dynamic and accurate evaluation of individual professional competencies and generate personalized career planning paths that maximize the benefits of long-term development. Specifically, DV-CAM first leverages Bidirectional Encoder Representations from Transformers (BERT) and Long Short-Term Memory (LSTM) to extract and dynamically update individual competency profiles from multi-source time-series data. Second, it designs a value-added evaluation index system covering the absolute level of competency, growth slope, growth curvature and stability to quantitatively diagnose development potential. Finally, under the framework of Deep Reinforcement Learning (DRL), this study takes competency status, value-added potential and job requirements as joint state inputs, and solves for the optimal sequence of career development actions by optimizing long-term cumulative rewards. Experiments based on the public Occupational Information Network (O*NET) dataset show that DV-CAM achieves a mean squared error of 0.0312 and a mean absolute error of 0.1289 for competency assessment on the test set. In planning tasks, its long-term cumulative reward and target job matching degree improvement rate reach 185.3% and 42.7% respectively, both outperforming various benchmark models. The results demonstrate that DV-CAM can diagnose competency status more accurately, and effectively guide individuals to achieve long-term career development goals through its forward-looking planning strategies. By combining the concept of value-added assessment with advanced deep learning technologies, this study provides a data-driven, dynamically perceivable and long-term oriented new paradigm for the fields of career planning and competency assessment.