Self-learning Strategy Research Articles

The Research of Air Fuel Ratio Self-Learning Control Strategy Based on UEGO Sensor

This paper begins with physical characteristics and the basic working principle of the traditional switch-type Exhaust Gas Oxygen (EGO) sensor and Universal Exhaust Gas Oxygen (UEGO) senor, and then it analyses the Air-Fuel ratio (A/F) Self-learning Control strategy and designs A/F Self-learning Control algorithm of Electronic Control Turbocharged CNG lean-burn engine based on UEGO sensor. At the end, it uses the written control code into ECU of the test engine, and after bench calibration tests it shows that the designed algorithm can compensate wearing, tiring, aging and other state of the engine and improve A/F control accuracy.

Neural Network Based Self-Learning Control Strategy for Electronic Throttle Valve

Recently, the application of the electronic throttle has been very popular in the automotive industry. However, difficulties in the control of electronic throttle valves exist due to multiple nonlinearities and plant parameter variations. A neural-network-based self-learning control (SLC) strategy that consists of a fuzzy neural network (FNN) controller and a recurrent neural network (RNN) identifier is proposed for electronic throttle valves in this paper. The FNN controller, which combines the semantic transparency of rule-based fuzzy systems with the learning capability of a neural network, is utilized as an SLC scheme and will be robust to plant parameter variations. An RNN identifier is employed to model the plant and provides plant information for the learning of the FNN controller. Both the structure and the learning algorithm of the control system are presented. The proposed controller is verified by computer simulations and experiments.

Robust State of Charge Estimation for Hybrid Electric Vehicles: Framework and Algorithms

State of Charge (SoC) estimation is one of the most significant and difficult techniques to promote the commercialization of electric vehicles (EVs). Suffering from various interference in vehicle driving environment and model uncertainties due to the strong time-variant property and inconsistency of batteries, the existing typical SoC estimators such as coulomb counting and extended Kalman filter cannot perform their theoretically optimal efficacy in practical applications. Aiming at enhancing the robustness of SoC estimation and improving accuracy under the real driving conditions with noises and uncertainties, this paper proposes a framework consisting of (1) an adaptive-κ nonlinear diffusion filter to reduce the noise in current measurement, (2) a self-learning strategy to estimate and remove the zero-drift, (3) a coulomb counting algorithm to realize open-loop SoC estimation, (4) an H∞ filter to implement closed-loop robust estimation, and (5) a data fusion unite to achieve the final estimation by integrating the advantages of the two SoC estimators. The availability and efficacy of each component have been demonstrated based on comparative studiesin simulation with the conventional approaches respectively, under the testing conditions of noises with various signal-noise-ratios, varying zero-drifts, and different model errors. The overall framework has also been verified to rationally and efficiently combine these components and achieve robust estimation results in the presence of kinds of noises and uncertainties.

Self-learning multiscale simulation for achieving high accuracy and high efficiency simultaneously

Biomolecular systems are inherently hierarchic and many simulation methods that try to integrate atomistic and coarse-grained (CG) models have been proposed, which are called multiscale simulations. Here, we propose a new multiscale molecular dynamics simulation method which can achieve high accuracy and high sampling efficiency simultaneously without aforehand knowledge on the CG potential and test it for a biomolecular system. In our method, a self-learning strategy is introduced to progressively improve the CG potential by an iterative way. (1) A CG model, coupled with the atomistic model, is used for obtaining CG structural ensemble, (2) which is mapped to the atomistic models. (3) The resulting atomistic ensemble is used for deriving the next-generation CG model. Two tests show that this method can rapidly improve the CG potential and achieve efficient sampling even starting from an unrealistic CG potential. The resulting free energy agreed well with the exact result and the convergence by the method was much faster than that by the replica exchange method. The method is generic and can be applied to many biological as well as nonbiological problems.

Self-Learning Multiscale Simulation

Molecular dynamics (MD) simulation plays more and more crucial roles in understanding the underlying molecular mechanisms of many biological processes. Unfortunately, due to the large number of degrees of freedom involved and inherently rugged energy surface, the time scale currently reachable by accurate all-atom (AA) simulation is far below typical biologically relevant time scale. Coarse graining the molecular representation can accelerate sampling, but the coarse grained (CG) simulation is unavoidably less accurate in energy estimate. To surmount these problems, a number of strategies have been proposed to integrate the AA and CG simulations, which is often called multiscale simulations. However, traditional multiscale methods heavily rely on the accuracy of the CG model. If the CG potential has its major basins different from those of AA potential, the multiscale simulation is not efficient and sometimes even bias the sampling. Here, we propose a new multiscale simulation method, self-learning multiscale molecular dynamics (SLMS-MD), which can achieve high accuracy and high sampling efficiency simultaneously. Based on the resolution exchange MD between atomistic and CG replicas, a self-learning strategy is introduced to progressively improve the initial CG potential by an iterative way based on the previously sampled CG conformations and their corresponding AA energies. The CG simulation ensures the efficient and broad sampling, and simultaneously the AA energies shape up the accuracy of the CG potential. Testing results show that the SLMS-MD can optimally combine the advantages of the AA and CG simulations, and achieve accurate and efficient multiscale simulations even when the initial CG potential is very poor. The resulting free energy converged to the exact result much faster than that by the replica exchange method. This method is generic and can be applied to many biological as well as non-biological problems.

SELECT: Self-Learning Collision Avoidance for Wireless Networks

The limited number of orthogonal channels and autonomous installations of hot spots and home wireless networks often leave neighboring 802.11 basic service sets (BSSs) operating on the same or overlapping channels, therefore interfering with each other. However, the 802.11 medium access control (MAC) does not work well in resolving inter-BSS interference due to the well-known hidden/exposed-receiver problem, which has been haunting the research community for more than a decade. In this paper, we propose SELECT, an effective and efficient self-learning collision avoidance strategy to address the hidden/exposed-receiver problem in 802.11 wireless networks. SELECT is based on the observation that carrier sense with received signal strength (RSS) measurements at the sender and the receiver can be strongly correlated. A SELECT-enabled sender exploits such correlation using an automated online learning algorithm and makes an informed judgment of the channel availability at the intended receiver. SELECT achieves collision avoidance at packet-level time granularity, involves zero communication overhead, and easily integrates with the 802.11 distributed coordination function (DCF). Our evaluation in analysis, simulations, and prototype experiments show that SELECT addresses the hidden/exposed-receiver problem well. In typical hidden/exposed-receiver scenarios, SELECT improves the throughput by up to 140 percent and the channel access success ratio by up to 302 percent while almost completely eliminating contention-induced data packet drops.

Self-Learning and Its Application to Laminar Cooling Model of Hot Rolled Strip

Abstract The mathematical model for online controlling hot rolled steel cooling on run-out table (ROT for abbreviation) was analyzed, and water cooling is found to be the main cooling mode for hot rolled steel. The calculation of the drop in strip temperature by both water cooling and air cooling is summed up to obtain the change of heat transfer coefficient. It is found that the learning coefficient of heat transfer coefficient is the kernel coefficient of coiler temperature control (CTC) model tuning. To decrease the deviation between the calculated steel temperature and the measured one at coiler entrance, a laminar cooling control self-learning strategy is used. Using the data acquired in the field, the results of the self-learning model used in the field were analyzed. The analyzed results show that the self-learning function is effective.

Cooperative Strategy Based on Adaptive<tex>$Q$</tex>-Learning for Robot Soccer Systems

The objective of this paper is to develop a self-learning cooperative strategy for robot soccer systems. The strategy enables robots to cooperate and coordinate with each other to achieve the objectives of offense and defense. Through the mechanism of learning, the robots can learn from experiences in either successes or failures, and utilize these experiences to improve the performance gradually. The cooperative strategy is built using a hierarchical architecture. The first layer of the structure is responsible for assigning each role, that is, how many defenders and sidekicks should be played according to the positional states. The second layer is for the role assignment related to the decision from the previous layer. We develop two algorithms for assignment of the roles, the attacker, the defenders, and the sidekicks. The last layer is the behavior layer in which robots execute their behavior commands and tasks based on their roles. The attacker is responsible for chasing the ball and attacking. The sidekicks are responsible for finding good positions, and the defenders are responsible for defending competitor scoring. The robots' roles are not fixed. They can dynamically exchange their roles with each other. In the aspect of learning, we develop an adaptive Q-learning method which is modified form the traditional Q-learning. A simple ant experiment shows that Q-learning is more effective than the traditional techniques, and it is also successfully applied to the learning of the cooperative strategy.

Intelligent control of air-conditioning systems using machine learning

In this paper, machine learning is implemented in a simulated air-conditioning system based on evolutionary computing methods involving the use of classifier systems and genetic algorithms. The overall objective is to achieve a controller capable of selflearning from its own experience to arrive at the desired performance against a specified evaluation scheme. The simulated results show that the self-learning intelligent control strategy is successful and can be considered for further development for application to on-line computer control of air-conditioning systems. It is envisaged that efficient control of air-conditioning systems employing this methodology will help achieve substantial energy saving.

Coordinating control and self-learning of robot trajectory motion

The problem of the robotic manipulator path motion control is reduced to the task of coordination of outputs of a multivariable plant. On this basis a new approach to the design of robot control systems is developed for the prescribed motion along complex analytical curves. For undetermined paths detected by sensors, an adaptive system and self-learning strategy are proposed. Simulation results confirm the readability and demonstrate an excellent dynamics as well as accuracy of the control systems considered.

Adaptation and Self-Learning of Nonlinear Multivariable Systems

It is proposed a self-learning control strategy for the spatial motion along an arbitrary smooth unknown curve. On the basis of geometrical approach a unified structure of the control system is derived. An algorithm of system adaptation and an approximation mechanism are chosen to support the training motions along the curve and to estimate undetermined functional components of the control law.

A fuzzy controller for construction activities

Abstract Construction activities are highly uncertain systems. They do not lend themselves to the application of traditional control theory because of their highly subjective nature. A fuzzy-based self-learning control strategy is utilized in this paper. The system comprises a rule-base, a self-learning unit and a conflict resolution unit. A new three dimensional implication function was developed and integrated in the system to model the individual rules. A condition assessment procedure is also integrated in the system in order to evaluate the output levels given a set of input states. The condition assessment procedure acts as a transfer function in traditional control theory.

On Self-Learning Control Strategy for Robot Manipulators

This paper is about the self-learning control of a robot manipulator. Using a so-called λ-decision, an improved reinforcement decision function is obtained and the corresponding self-learning control algorithm is derived which exhibits the following properties: 1) It is simple to compute, that makes it easy to implement, 2) It requires only a very rough description of the robot manipulator to be controlled and 3) There is no need for repeated trial in applying it.