Faster Convergence Velocity Research Articles

Full waveform inversion as training a neural network

Full waveform inversion (FWI) is usually solved as a nonlinear least-squares problem to minimize the discrepancy between the recorded signal and the synthetic data by some gradient-based optimization methods. In this paper, we investigate acoustic FWI as training a neural network. We recast the time-domain difference scheme into a forward propagation process of vanilla recurrent neural network (RNN) and then find that the parameters of RNN coincide with the physical parameters in the wave equation. As a result, the FWI problem is resolved as training such a kind of neural network. Some stochastic optimization methods such as the Adam optimizer are applied to update the model parameters. We demonstrate our method numerically with three typical models including the benchmark Marmousi model. The numerical results show that the fast convergence and good velocity inversion result can be achieved.

Reinforcement Learning-Based Lane Change Decision for CAVs in Mixed Traffic Flow under Low Visibility Conditions

As an important stage in the development of autonomous driving, mixed traffic conditions, consisting of connected autonomous vehicles (CAVs) and human-driven vehicles (HDVs), have attracted more and more attention. In fact, the randomness of human-driven vehicles (HDV) is the largest challenge for connected autonomous vehicles (CAV) to make reasonable decisions, especially in lane change scenarios. In this paper, we propose the problem of lane change decisions for CAV in low visibility and mixed traffic conditions for the first time. First, we consider the randomness of HDV in this environment and construct a finite state machine (FSM) model. Then, this study develops a partially observed Markov decision process (POMDP) for describing the problem of lane change. In addition, we use the modified deep deterministic policy gradient (DDPG) to solve the problem and get the optimal lane change decision in this environment. The reward designing takes the comfort, safety and efficiency of the vehicle into account, and the introduction of transfer learning accelerates the adaptation of CAV to the randomness of HDV. Finally, numerical experiments are conducted. The results show that, compared with the original DDPG, the modified DDPG has a faster convergence velocity. The strategy learned by the modified DDPG can complete the lane change in most of the scenarios. The comparison between the modified DDPG and the rule-based decisions indicates that the modified DDPG has a stronger adaptability to this special environment and can grasp more lane change opportunities.

A novel hybrid grey wolf optimized fuzzy logic control based photovoltaic water pumping system

AbstractThis paper describes the novel scientific contribution of a modified hybrid grey wolf optimization‐fuzzy logic controller (GWO‐FLC) based on maximum power point tracking (MPPT) for a photovoltaic (PV) system. The modified GWO eliminates the lower rank (δ) and lowest rank (Ψ) wolves, which provides a rapid search method to track global maximum point (GMPP) in a lesser period. The hybrid GWO‐FLC based MPPT technique provides tracking of GMPP with zero oscillations and high PV power tracking under rapid weather fluctuations. The GWO technique acquires GMPP and provides soft tuning of fuzzy parameters to achieve high PV power tracking with zero oscillations around GMPP. The experimental responses confirm changing environmental situations. The dSPACE real‐time platform has been employed to validate the proposed hybrid MPPT controller design, which provides optimal PV power tracking with accurate efficiency, fast convergence velocity, and less computational period. The switched reluctance motor (SRM) driven PV water pump is proposed in this research work. The proposed PV water pumping has low cost and no voltage, and current sensors are employed to generate a gating pattern of an SRM based mid‐point converter, which reduces the losses.

Effects of Plate Velocity Slowdown on Altering Continental Collision Patterns and Crustal-Lithospheric Deformation During the Collision Process

Continental collision zones are widely distributed across the earth’s surface with diverse types of tectonic processes. Even the same collision zone shows significant lateral tectonic variations along its strike. In this study, we systematically investigated how plate velocity slowdown after the closure of the ocean influences the continental collision evolution, as well as the effects of kinematic characteristics and continental rheology on varying the continental collision modes in a plate velocity slowdown model. From the comparison between the constant plate velocity system (CVS) and the plate velocity-dropping system (VDS), we can conclude the following: Plate velocity dropping promotes the extension inside the slab by decreasing the movement of the surface plate, whereas slab pull increases as subduction continues. The timing of the subducting slab break-off and the polarity alteration was initiated earlier in the plate velocity drop models than in the constant plate velocity models, and fast convergence may have triggered multiple episodes of slab break-off and caused strong deformation adjacent to the collision zone. Parametric tests of the initial subducting angle, plate convergence velocity, and continental crustal rheological strength in VDS indicated the following: (1) Three end members of the continental lithospheric mantle deformation modes were identified from the VDS; (2) models with a low subducting angle, fast continental convergence velocity, and medium-strength overriding crust were more likely to evolve into a polarity reversed mode, whereas steep-subducting-angle, slow-plate-velocity, weak-overriding-crust models tended toward a two-sided mode; (3) a strong overriding continent is more liable to develop a stable mode; and (4) overriding crustal rheological strength plays a significant role in controlling changes in continental collision modes.

Gliding hydrodynamic modeling and identification of underwater glider based on differential evolution algorithm

Underwater glider (UG) is a new type of ocean observation platform. An accurate hydrodynamic analysis is necessary for motion performance analysis and motion control purposes. This paper proposes a novel data-driven method based on a differential evolution (DE) algorithm for the UG hydrodynamic. The gliding hydrodynamic identification model is derived first from the dynamic model under stable gliding conditions. Then, based on the trial data, the DE is used to identify gliding hydrodynamic. Finally, the towing tank test (Tank) and computational fluid dynamics (CFD) simulation are carried out to validate the effectiveness of the proposed method. To show the superiority of the DE-based method, the optimization (PSO) algorithm is added for comparison. The result indicates that the DE-based method has higher accuracy and faster convergence velocity in the UG hydrodynamic identification problem. When the gliding angle is set to 14.23∘, the maximum gliding efficiency is 94.00%. This hydrodynamic identification method also provides a reference for other underwater vehicles.

An Effective Method for Minimizing Electric Generation Costs of Thermal Systems with Complex Constraints and Large Scale

In this paper, an improved antlion optimization algorithm (IALO) was proposed to search for promising solutions for optimal economic load dispatch (ELD) problems to minimize electrical generation fuel costs in power systems with thermal units and to ensure all constraints are within operating ranges. IALO can be more effective than the original method, called the antlion optimization algorithm (ALO), because of the high performance of the applied modifications on the new solutions searching process. In order to evaluate the abilities of the IALO method, we completed many tests on thermal generating systems including 10, 15, 20, 30, 60, 80, and 90 units with different constraints and fuel-consuming characteristics. The results suggest that the offered method is superior to the ALO method with more stable search ability, faster convergence velocity, and shorter calculation times. Furthermore, the obtained results of the IALO method are much better than those of almost all the other methods used to solve problems for the same systems. As a result, IALO is suggested to be a highly effective method, and it can be applied to other problems in power systems instead of ALO, which has a lower performance.

An Intrusion Detection System Based on a Simplified Residual Network

Residual networks (ResNets) are prone to over-fitting for low-dimensional and small-scale datasets. And the existing intrusion detection systems (IDSs) fail to provide better performance, especially for remote-to-local (R2L) and user-to-root (U2R) attacks. To overcome these problems, a simplified residual network (S-ResNet) is proposed in this paper, which consists of several cascaded, simplified residual blocks. Compared with the original residual block, the simplified residual block deletes a weight layer and two batch normalization (BN) layers, adds a pooling layer, and replaces the rectified linear unit (ReLU) function with the parametric rectified linear unit (PReLU) function. Based on the S-ResNet, a novel IDS was proposed in this paper, which includes a data preprocessing module, a random oversampling module, a S-Resnet layer, a full connection layer and a Softmax layer. The experimental results on the NSL-KDD dataset show that the IDS based on the S-ResNet has a higher accuracy, recall and F1-score than the equal scale ResNet-based IDS, especially for R2L and U2R attacks. And the former has faster convergence velocity than the latter. It proves that the S-ResNet reduces the complexity of the network and effectively prevents over-fitting; thus, it is more suitable for low-dimensional and small-scale datasets than ResNet. Furthermore, the experimental results on the NSL-KDD datasets also show that the IDS based on the S-ResNet achieves better performance in terms of accuracy and recall compared to the existing IDSs, especially for R2L and U2R attacks.

Hybrid optimisation method of improved genetic algorithm and IFT for linear thinned array

The array thinning technique can greatly reduce the number of the array elements while keeping the performance of the array almost the same. However, the existing algorithms have slow convergence rates and are easy to fall into local optimum. To improve the optimisation performance, a hybrid method based on improved genetic algorithm (GA) and iterative Fourier transform (IFT) technique for linear thinned array is proposed in this study. The population is divided into improved GA group and IFT group according to the convergence of the population and different operations can be done paralleled to generate offspring in each iteration. In the improved GA processing, adaptive crossover rate and mutation rate are used. The mechanism that keeps the fill factor stable is removed for larger search range. The IFT processing is executed paralleled for fast convergence velocity. The proposed hybrid method can obtain the fast convergence velocity and avoid being trapped into the local optimum by the combination of the two approaches. Several examples are simulated to validate the performance of the proposed method.

A cooperative particle swarm optimization with constriction factor based on simulated annealing

As many engineering optimization problems are rather complicated, it is usually necessary to search the optimal solution in a complex and huge search space. When faced with these large-scale problems, conventional optimization algorithms need to traverse the entire search space and it is impossible for them to finish the search within polynomial time. Moreover, it can’t meet requirements in terms of computation velocity, convergence and sensitivity to initial value. So, it is very difficult to apply them to engineering optimization problems. Swarm intelligence methods simulate the collective behaviors of social creatures in the nature and they come from the relationship between the community formed by simple individuals and the environment as well as the interactions between the individuals. A single individual can only perform simple tasks, but the population formed by single individuals can fulfill complex tasks. Such intelligence presented by such population is called swarm intelligence. Due to the limitations of existing optimization algorithms, it is usually impractical to obtain excellent computational performance with only one optimization algorithm. In consideration of the jumping property of simulated annealing, it is not easy to get trapped into local minimum and it has strong local search capability near the optimal value and fast convergence velocity. This paper combines it with particle swarm optimization, proposes a cooperative particle swarm optimization with constriction factor based on simulated annealing (SA-CPSO), offers guidelines on selection of related parameters and dynamically adjusts the particle velocity according to its movement track. In this way, it improves the convergence velocity of the algorithm by improving the spatial search ability of the particle so as to make the particle accept the solution which makes the fitness of the objective function “better” as well as the solution that makes the said fitness “worse” at a certain probability during the flight of the particle. The experiment shows that the SA-CPSO improves the diversity of the particle and enhances its ability to get rid of locally optimal solutions. So, SA-CPSO is not easy to be trapped into local optimum and it has stronger ability of global optimization, a faster convergence velocity and higher convergence accuracy.

Pressure Control for a Hydraulic Cylinder Based on a Self-Tuning PID Controller Optimized by a Hybrid Optimization Algorithm

In order to improve the performance of the hydraulic support electro-hydraulic control system test platform, a self-tuning proportion integration differentiation (PID) controller is proposed to imitate the actual pressure of the hydraulic support. To avoid the premature convergence and to improve the convergence velocity for tuning PID parameters, the PID controller is optimized with a hybrid optimization algorithm integrated with the particle swarm algorithm (PSO) and genetic algorithm (GA). A selection probability and an adaptive cross probability are introduced into the PSO to enhance the diversity of particles. The proportional overflow valve is installed to control the pressure of the pillar cylinder. The data of the control voltage of the proportional relief valve amplifier and pillar pressure are collected to acquire the system transfer function. Several simulations with different methods are performed on the hydraulic cylinder pressure system. The results demonstrate that the hybrid algorithm for a PID controller has comparatively better global search ability and faster convergence velocity on the pressure control of the hydraulic cylinder. Finally, an experiment is conducted to verify the validity of the proposed method.

Hybrid Genetic Algorithm and Modified Iterative Fourier Transform Algorithm for Large Thinned Array Synthesis

In this letter, a hybrid algorithm based on the genetic algorithm (GA) and modified iterative Fourier transform (MIFT) technique called HGAMIFT is proposed for large thinned array synthesis. By employing a perturbation mechanism, the MIFT can achieve a better solution in terms of the peak sidelobe level when compared to the iterative Fourier transform technique. Moreover, a control factor to determine the proportion of individuals from GA and MIFT as well as the crossover and mutation rate is introduced to help the HGAMIFT maintain the diversity of population in the early phase while avoiding stagnation in the late phase. Thus, a resulting enhanced search ability and fast convergence velocity can be obtained simultaneously. Several thinned arrays that comprise different array sizes and aperture shapes have been synthesized, which validate the superior performance of the proposed algorithm.

Modified super twisting controller for servicing to uncontrolled spacecraft

A relative position and attitude coupled sliding mode controller is proposed by combining the standard super twisting (ST) control and basic linear algorithm for autonomous rendezvous and docking. It is schemed for on-orbit servicing to a tumbling noncooperative target spacecraft subjected to external disturbances. A coupled dynamic model is established including both kinematical and dynamic coupled effect of relative rotation on relative translation, which illustrates the relative movement between the docking port located in target spacecraft and another in service spacecraft. The modified super twisting (MST) control algorithm containing linear compensation items is schemed to manipulate the relative position and attitude synchronously. The correction provides more robustness and convergence velocity for dealing with linearly growing perturbations than the ST control algorithm. Moreover, the stability characteristic of closed-loop system is analyzed by Lyapunov method. Numerical simulations are adopted to verify the analysis with the comparison between MST and ST control algorithms. Simulation results demonstrate that the proposed MST controller is characterized by high precision, strong robustness and fast convergence velocity to attenuate the linearly increasing perturbations.

Using the Particle Swarm Optimization Model to Evaluate the Wind Power Enterprise Development Ability under Low-Carbon Economy Environment

Electricity is the basic industry in China, which has the important strategic significance to maintain the social stability, ensure the national security and promote the economic development. With the rapid development of power market reform and the establishment of bidding for access mechanism, the competition among the power generation enterprises becomes much drastic. To evaluate the development ability of wind power enterprises in the power new energy, the authors proposed a novel particle swarm optimization (PSO) algorithm, which used the randomness, the rapidity and the global characteristics to obtain the pheromone distribution, and had the faster convergence velocity. The development ability evaluation of 12 wind power enterprises showed that the results given by this model were reliable, and it is feasible to evaluate the development ability using this method.

Application of Quantum Particle Swarm Optimization in Adaptive Notch Filter Design

Abstract: adaptive notch filter is a kind of apparatus which can eliminate single frequency or narrow-band interference, normal adaptive algorithm of notch filter is LMS algorithm, but the faster convergence velocity and the smaller steady error are difficult to gain simultaneously. Aimed at the weakness of LMS, the Particle Swarm Optimization (PSO) is studied deeply in the paper, based on the PSO; the quantum mechanic theory is added to improve it. Quantum Particle Swarm Optimization (QPSO) is researched and applied for adaptive notch filter which is proved more efficient in the noise control by MATLAB simulation. The new QPSO algorithm can balance the maladjustment and the searching ability of adaptive filter with a little calculation, the speed of convergence is faster than LMS and normal PSO algorithm.

A NOVEL IGA-EDSPSO HYBRID ALGORITHM FOR THE SYNTHESIS OF SPARSE ARRAYS

Based on the improvements of both Genetic Algorithm and Particle Swarm Optimization, a novel IGA-edsPSO(Improved Genetic Algorithm-extremum disturbed simple Particle Swarm Optimization) Hybrid algorithm is proposed in this paper. An improved performance of GA is achieved by reducing the array space. By discarding the particle velocity vector in the PSO evolutionary equation, the sPSO (simple PSO) can avoid the problem of slow later convergence velocity and low precision caused by determining the maximal velocity vector factitiously. And the edsPSO can overstep local extremum point more effectively with the help of the extremum disturbed factor. The proposed IGA-edsPSO Hybrid algorithm is used in the design of the sparse arrays with minimum element spacing constraint. Given the array aperture and the number of the array elements, the suppression of the peak sidelobe level (PSLL) with a certain half power beamwidth (HPBW) restriction is implemented with a high efficiency by optimizing the HPBW and PSLL synchronously. The simulation results show that faster convergence velocity (which means less computation time) and lower sidelobe level are obtained using IGA-edsPSO compared to IGA, standard PSO, GA-PSO and GA-sPSO.

Trial mountain climbing algorithm for solving the inverse kinematics of redundant manipulator

Trial mountain climbing algorithm to solve the inverse kinematics problem of redundant manipulator is introduced, and a method of describing a numeral with a special numeration system is given to define the changed step of the trial mountain climbing algorithm. The results show that a likelihood solution can be found quickly in the infinite groups of likelihood solutions within the limited search times, and need not calculate the anti-trigonometric function and the inverse matrix. In addition, this algorithm has many good qualities such as concise algorithm, tiny computation, fast convergence velocity, good stability and extensive adaptability.

The influence of backstop dip and convergence velocity in the growth of viscous doubly-vergent orogenic wedges: insights from thermomechanical laboratory experiments

We investigate the effects of convergence velocity and backstop geometry on the evolution of experimental viscous doubly-vergent orogenic wedges. Experiments are performed in a thermomechanical shortening box, using the Newtonian properties of 52/54 EN type paraffin as analogue to the natural strength distribution in the continental crust. A velocity discontinuity is imposed at the subduction slot, where the base plate is pulled below a rigid backstop. The mobile base plate dips 5°. Two different backstop dips (60° and 30°) are used in combination with constant convergence velocities of 2 and 10 cm h −1. Results point out that backstop geometry exerts a first-order control on the evolution of doubly-vergent experimental wedges. Steeply-dipping backstops cause the stationary uplift of the wedge axial region above the subduction slot, whereas shallow-dipping backstops allow its retroward translation, in concomitance with a more pronounced deformation in the retrowedge. Convergence velocity controls the uplift in the wedge axial region and the amount of the outward migration of the deformation front in both the prowedge and retrowedge regions. For fast convergence velocity and shallow-dipping backstop, deformation in the retrowedge is favoured. An increase in the backstop dip and/or a decrease in the convergence velocity favour deformation in the prowedge. Both prowedge and retrowedge tend to grow in self-similar conditions, with an aspect ratio depending upon the backstop dip and convergence velocity. These results may have important consequences for the evolution and deformation style of ductile-dominated orogenic wedges, suggesting that location of the axial region and deformation progression in natural orogens can be strongly controlled by boundary and kinematic conditions.